Systems and methods for imaging and modulating nervous systems using ultrasound-based brain-computer interfaces

Abstract

Devices, methods, and systems related to ultrasound imaging or regulation of the nervous system are described. The apparatus may include, for example, one or more ultrasound transducers, where an ultrasound transducer of the one or more ultrasound transducers may include an implantable ultrasound transducer, where the implantable ultrasound transducer may include an acoustically transparent window. The methods and systems may also include, for example, methods of imaging and / or modulating a nervous system of a subject using the one or more ultrasound transducers. The method of imaging and / or regulating the nervous system of the subject may be based on a closed loop operation, wherein iterations of the imaging and / or regulation of the nervous system are based on previous iterations. The closed loop operation may include ultrasound imaging and ultrasound-based regulation or electrophysiological regulation. The method may also include a method of analyzing ultrasound data, such as through an artificial neural network.

Description

Cross-references to related applications

[0001] This application claims priority to U.S. Provisional Patent Application No. 63 / 511,617, filed June 30, 2023, and U.S. Provisional Patent Application No. 63 / 598,886, filed November 14, 2023, the contents of which are incorporated herein by reference in their entirety. Technical Field

[0002] This disclosure generally relates to systems and methods for imaging and modulating the physiological functions (e.g., the nervous system) of a subject, and more specifically, to systems and methods for imaging and modulating the nervous system of a subject using an implantable ultrasound transducer. Background Technology

[0003] Debilitating brain disorders and diseases resistant to treatment or medication are widespread. Existing neurotechnical solutions are insufficient to address the complexity and individualized nature of human brain dysfunction. Current solutions may be highly invasive, limited in spatial or temporal resolution, spatial or temporal scope, or physically cumbersome, making it difficult to obtain orthogonal homologous measurements from subjects. Advanced monitoring and treatment tools are needed to overcome the limitations of currently available drugs and neurotechnologies.

[0004] For example, neuropsychiatric and cognitive disorders, including depression and neuropathic pain, share common characteristics. These disorders occur in circuits and systems spatially distributed throughout the nervous system. Furthermore, the brain states associated with these disorders evolve slowly over time, ranging from hours to months. Moreover, brain states can vary between individuals, even among those diagnosed with the same brain dysfunction. This dispersed and time-evolving nature of the disorders can benefit from large-scale, long-term approaches to imaging and modulating the nervous system (e.g., for monitoring or treating pathological brain function). Summary of the Invention

[0005] The methods and systems discussed in this paper address the technical problem of the lack of suitable systems and methods for monitoring and manipulating neural activity in human subjects at sufficiently broad temporal and spatial scales and resolution. Existing systems and methods struggle to solve this problem due to the inherent fundamental physical and neurophysiological constraints of the technology. The methods and systems disclosed herein include an ultrasound-based technique that can monitor and manipulate neural activity in humans at meso- or macro-scale coverage (including, but not limited to, whole-brain scale). The methods and systems disclosed herein utilize ultrasound-based physics to achieve macro-scale-level docking with the brain. The macro-scale brain-computer interface described herein can observe and modulate neural circuit dynamics at a broad, whole-brain scale. The methods and systems described herein include an ultrasound-based neurotechnology platform, which may also include a digital diagnostic and therapeutic ecosystem supporting the ultrasound-based platform. The systems and methods disclosed herein can utilize meso- or macro-scale access across brain regions (such as, but not limited to, whole-brain access) to achieve improved treatment of brain dysfunction.

[0006] In addition to improving sensitivity and resolution compared to existing methods, functional ultrasound imaging, as described herein, can also be encapsulated into implantable shape factors, unlike functional magnetic resonance imaging (fMRI), for example. In doing so, the described ultrasound imaging system can facilitate high-resolution neuroimaging while the subject performs natural and clinically relevant behaviors. Beyond clinically relevant neuroimaging applications, the disclosed system can also enable neural stimulation of dysfunctional brain circuits while the subject performs clinically relevant behaviors. The neural stimulation of the subject can be in a closed-loop state (discussed further below), allowing for therapeutic stimulation of relevant brain circuits when relevant neural activity patterns and / or clinically relevant behaviors are observed. The device encapsulation of the system disclosed herein can also be repurposed for more general applications beyond interfacing with neural activity, such as for the monitoring and stimulation of non-brain physiological systems.

[0007] The systems and methods disclosed herein are generally compact and minimally invasive. These systems and methods include at least one (but typically multiple) small ultrasound transducers, referred to herein as “implantable transducers,” “implantable sensors,” or “pucks.” The pucks are designed to fit within a craniotomy opening in the skull (e.g., a drill hole 30 mm in diameter or smaller), which extends the puck’s lifespan and minimizes the risk of infection. The hardware ecosystem supporting the pucks may include implantable technologies such as, but not limited to, rechargeable batteries and wireless data streaming. In this implementation, deploying the pucks for functional ultrasound in the brain can be rapid and relatively inexpensive. In other implementations, the hardware ecosystem controlling the pucks may include custom solutions, such as custom controllers employing specialized management schemes to coordinate a group of pucks. The software ecosystem supporting the disclosed systems for monitoring and manipulating the brain may include tools that provide analysis, visualization, and quantification of relevant metrics and neurobiomarkers. The systems and methods disclosed herein allow for the monitoring and manipulation of neural activity at improved spatial and temporal scales and resolution, even when subjects are performing clinically relevant behaviors.

[0008] Systems and methods for imaging and / or modulating the nervous system using ultrasound, as described herein, can operate in a closed loop, such that ultrasound-based imaging of nervous system activity is used to modulate the nervous system of a subject. Imaging-based modulation of the nervous system can be performed iteratively, such that iterative imaging and modulation direct the subject's neural activity toward a target neural activity state (e.g., brain state). The target neural activity state can be a standard neural state, for example, a neural activity state that does not correspond to the observed neural state or a neural state indicating a subject with psychopathology.

[0009] For example, ultrasound data (such as ultrasound imaging using an implanted transducer) can be used to observe neural activity in a region of a subject's brain. The ultrasound data can then be analyzed, for example, through a trained machine learning model, to determine a set of modulation parameters (e.g., instructions for modulating the subject's neural activity). The model can determine the modulation parameters such that when neural modulation is performed according to the determined parameters, the observed region will reach a target state of neural activity. The modulation parameters can then be communicated to a system used to perform the neural modulation.

[0010] The effects of neuromodulation can then be observed, for example, at selected brain regions using ultrasound imaging. A trained machine learning model can be used to analyze the difference between the neural activity generated by the modulation (observed via ultrasound imaging) and the target neural activity state. This difference can be used to determine updated modulation parameters so that subsequent neuromodulation based on these new parameters causes the subject's neural activity to converge toward the target neural activity state. When the difference between the observed activity and the target neural activity is less than a threshold (e.g., indicating that the observed activity has sufficiently reached the target neural activity), the modulation parameters need not be updated.

[0011] The alternating sequence of observing and subsequently updating the regulation of a subject's neural activity can be performed in real time. In some implementations, the alternating sequence of observation and regulation lasts for a period of time, such as based on clinical and biomedical constraints. In some aspects, each iteration of observing and analyzing a subject's neural activity and regulating that activity based on the observations can bring the observed neural activity closer to the target neural activity, thereby allowing for more efficient and accurate treatment of the subject in a minimally invasive and more personalized manner.

[0012] Although examples of methods and systems for imaging and modulating the nervous system are described relative to the brain, it should be understood that these methods and systems can be performed on other parts of the nervous system. These methods and systems can be used on other parts of the nervous system, such as non-brain parts of the central nervous system (e.g., the spinal cord) or the peripheral nervous system. Although examples of methods and systems for imaging and modulating the nervous system are described relative to ultrasound images, it should be understood that these methods and systems can be performed using other types of ultrasound data, such as radio frequency (RF) data.

[0013] In some embodiments, the method may further include emitting ultrasound waves through one or more implantable transducers, wherein the ultrasound waves are configured to modify the physiological activity of a subject. In any embodiment of the embodiments herein, the system may include one to ten implantable transducers.

[0014] In some respects, this document discloses an implantable transducer comprising: a housing; an acoustic window at least partially disposed at a first end of the housing; and an ultrasonic array disposed within the housing near the first end, the ultrasonic array being configured to emit ultrasonic waves to the external environment through the acoustic window.

[0015] In some implementations, the implantable transducer may also include one or more circuit boards disposed within a housing, the circuit boards including one or more electronic components disposed thereon, the electronic components being configured to transmit one or more signals to an ultrasound array.

[0016] In some implementations, one or more electronic components disposed on a circuit board are configured to process data received from an ultrasound array, the data indicating brain function in a subject.

[0017] In any embodiment of the embodiments described herein, the data includes image data indicating the anatomical features of the subject.

[0018] In any of the embodiments described herein, the implantable transducer is configured to be placed in a hole in the skull of the subject.

[0019] In any of the embodiments described herein, the implantable transducer is positioned to contact the subject's soft tissue.

[0020] In any of the embodiments described herein, the implantable transducer is configured to raise the local body temperature by less than 2°C.

[0021] In any embodiment of the embodiments described herein, the implantable transducer is configured to limit absolute local brain temperature to less than 39°C.

[0022] In any of the embodiments described herein, the implantable transducer is positioned to contact the dura mater of the subject.

[0023] In any of the embodiments described herein, the implantable transducer is located outside the brain parenchyma of the subject.

[0024] In any embodiment of the embodiments described herein, the housing includes a lip disposed at a second end of the housing, the lip being configured to attach to the outer surface of the subject's skull.

[0025] In any of the embodiments described herein, the implantable transducer includes a cable configured to send power or data to or from the implantable transducer.

[0026] In any of the embodiments described herein, the cable extends through the housing of the implanted transducer.

[0027] In any embodiment of the embodiments described herein, the acoustic window comprises a biocompatible polymer.

[0028] In some embodiments, the biocompatible polymer is polymethyl methacrylate (PMMA), or polyether ether ketone (PEEK), polychlorotrifluoroethylene (PCTFE), polytetrafluoroethylene (PTFE), ultra-high molecular weight polyethylene (UHMWPE), polyethylene terephthalate (PET), low-density polyethylene (LDPE), polyether block amide (PEBAX), and / or high-density polyethylene (HDPE).

[0029] In any of the embodiments described herein, the ultrasonic array is fabricated on a complementary metal-oxide-semiconductor (CMOS) application-specific integrated circuit (ASIC).

[0030] In any of the embodiments described herein, the ultrasonic array includes a capacitive micromechanical ultrasonic transducer (CMUT), a piezoelectric micromechanical ultrasonic transducer (PMUT) array, or a lead zirconate titanate (PZT) array.

[0031] In any of the embodiments described herein, the implantable transducer is configured to couple to one or more wires, the implantable transducer is configured to transmit data through one or more wires, and is further configured to receive data through one or more wires.

[0032] In any of the embodiments described herein, the implantable transducer is configured to receive multiple ultrasound waves.

[0033] In any of the embodiments described herein, the ultrasonic array includes a plurality of transducer elements.

[0034] In some implementations, the plurality of transducer elements includes 100 to 199, 200 to 399, 400 to 999, 1,000 to 1,499, 1,500 to 9,999, 10,000 to 11,999, 12,000 to 99,000, or 100,000 to 120,000 transducer elements.

[0035] In some implementations, the ultrasonic array comprises an n×m matrix, where n ranges from 16 to 256 transducer elements and m ranges from 1 to 256 transducer elements.

[0036] In some aspects, this document discloses a system for monitoring or modulating the physiological activities of a subject, comprising: one or more implantable transducers, wherein the implantable transducer in the one or more implantable transducers corresponds to an implantable transducer in any embodiment of the embodiments herein; and a controller coupled to each of the one or more implantable transducers, the controller including a power source and a processor, wherein the power source is configured to power each of the one or more implantable transducers, and wherein the processor is configured to perform a method comprising: transmitting one or more signals to the one or more implantable transducers; and receiving data from the one or more implantable transducers.

[0037] In some implementations, the system may further include: emitting ultrasound waves via one or more implantable transducers, wherein the ultrasound waves are configured to modify the physiological activity of the subject.

[0038] In any embodiment of the embodiments described herein, one or more signals are configured to specify the amplitude or timing of one or more of a plurality of transducer elements.

[0039] In any embodiment of the embodiments described herein, the system may also include one to ten implantable transducers.

[0040] In any of the embodiments described herein, the system may also include a remote center configured to receive data from the controller and further configured to send external data to the controller.

[0041] In some implementations, a remote center may be configured to communicate with a display to provide a user interface for controlling the system.

[0042] In any of the embodiments described herein, the implanted transducer and controller are configured to perform wireless communication.

[0043] In some implementations, the implantable transducer and controller are configured to communicate via Bluetooth, Bluetooth Low Energy, WiFi, or a combination thereof.

[0044] In any of the embodiments described herein, one or more signals are configured to coordinate the transmission of ultrasound waves through the implanted transducer and are further configured to coordinate the reception of ultrasound waves.

[0045] In any of the embodiments described herein, the controller includes a clock, and one or more signals are transmitted based on predetermined intervals associated with the clock.

[0046] In some implementations, the controller includes a central clock, and one or more implanted transducers each include a clock, wherein a signal corresponds to a reset signal associated with the central clock.

[0047] In any of the embodiments described herein, subjects perform clinically relevant behaviors while acquiring data from the implanted transducer.

[0048] In some implementation schemes, clinically relevant behaviors include activities of daily living, motion estimation, motion capture, facial expressions and reaction time, self-reported mood, self-reported cognitive state, heart rate, heart rate variability, respiratory rate, oxygenation, skin conductance response, inertial monitoring, or a combination thereof.

[0049] In any of the embodiments described herein, the system is configured to modify the physiological activities of a subject based on data.

[0050] In some implementation schemes, data-driven modifications to the subject's physiological activities occur in real time.

[0051] In some implementations, real-time occurrence includes a reaction time of 5 seconds or less after the data is received.

[0052] In some implementations, the physiological activities of the subjects are modified to occur at regular, predetermined intervals.

[0053] In any of the embodiments described herein, the physiological activity of the subject is neural activity.

[0054] In some implementations, the subject's neural activity is the neural activity of the central nervous system.

[0055] In some implementations, the subject's neural activity is the neural activity of the brain.

[0056] In some implementations, the neural activity of the brain includes neural activity from the brain's distributed neural networks.

[0057] In any of the embodiments described herein, the subject has or is suspected of having a neurological dysfunction.

[0058] In some implementations, neurological dysfunction is clinical depression, clinical anxiety disorder, neuropathic pain, or a combination thereof.

[0059] In some aspects, this document discloses a method for monitoring the physiological activities of a subject, the method comprising: transmitting one or more signals to one or more implantable transducers via a controller, wherein the controller is positioned remotely from the one or more implantable transducers, and wherein the one or more implantable transducers are mounted on the skull of the subject; and receiving data from the one or more implantable transducers.

[0060] In some implementations, the method may further include: emitting ultrasound waves through one or more implantable transducers based on one or more signals, wherein the ultrasound waves are configured to modify the physiological activity of the subject.

[0061] In any embodiment of the embodiments described herein, the method further includes: modifying the physiological activity of the subject based on ultrasound.

[0062] In some respects, this document discloses a method for determining instructions for modulating neural activity of a subject's nervous system, the method comprising: receiving ultrasound data of the nervous system from an implanted transducer, wherein the ultrasound data indicates the physiological state of the nervous system; processing the ultrasound data of the nervous system; and transmitting the processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data.

[0063] In some implementations, the region of neural activity being modulated is determined based on ultrasound data.

[0064] In any embodiment of the embodiments described herein, the method may further include receiving one or more of data associated with physiological state and data associated with neural activity.

[0065] In any embodiment of the embodiments described herein, physiological state includes neurophysiological state.

[0066] In some implementations, neurophysiological states include hemodynamic activity.

[0067] In some implementations, hemodynamic activity is indicated by power Doppler intensity associated with ultrasound data.

[0068] In any embodiment of the embodiments described herein, hemodynamic activity includes cerebral blood volume (CBV) activity, and changes in CBV activity are proportional to changes in power Doppler intensity.

[0069] In any of the embodiments described herein, modulating neural activity includes stimulating one or more regions of the nervous system.

[0070] In some implementations, one or more regions of the nervous system include one or more regions of the peripheral nervous system, one or more regions of the central nervous system, or a combination thereof.

[0071] In some implementations, one or more regions of the central nervous system include the brain.

[0072] In any embodiment of the embodiments described herein: stimulation of one or more areas of the nervous system includes electrical stimulation via one or more electrodes, and instructions for modulating neural activity include instructions for controlling electrical stimulation via one or more electrodes.

[0073] In some implementations, electrical stimulation is controlled by electromodulation parameters, including amplitude, frequency, pulse width, intensity, waveform, polarity, acoustic pressure, or any combination thereof.

[0074] In any embodiment of the embodiments described herein, electrical stimulation includes deep brain stimulation (DBS), transcranial magnetic stimulation (TMS), repetitive TMS (rTMS), vagus nerve stimulation (VNS), transcranial direct current stimulation (tDCS), electrocorticography (ECoG), or any combination thereof.

[0075] In any embodiment of the embodiments described herein, the modulation of neural activity includes ultrasound neuromodulation, and the method further includes: receiving instructions for modulating neural activity by an implantable transducer; and performing ultrasound neuromodulation by the implantable transducer.

[0076] In any embodiment of the embodiments described herein, the modulation of neural activity includes ultrasound neuromodulation, and the method further includes receiving ultrasound neuromodulation by an implanted transducer.

[0077] In any of the embodiments described herein, the method is performed over a period of time, such as seconds, minutes, hours, days, weeks, months, or years.

[0078] In any of the embodiments described herein, the instructions for modulating neural activity are associated with longitudinal treatment or longitudinal studies.

[0079] In any of the embodiments described herein, the instructions for regulating neural activity are further determined based on pre-experimental physiological information.

[0080] In some implementations, pre-trial physiological information includes pre-trial ultrasound information, functional magnetic resonance imaging (fMRI) information, electrophysiological recordings, structural magnetic resonance imaging scans, diffusion tensor imaging (DTI) information, computed tomography (CT) scan information, or any combination thereof.

[0081] In any of the embodiments described herein, the instructions for modulating neural activity are determined based on the output of a machine learning algorithm.

[0082] In some implementations, the output of the machine learning algorithm is based on the ultrasound data provided to the machine learning algorithm.

[0083] In any of the embodiments described herein, the machine learning algorithm is trained using pre-experimental physiological state information.

[0084] In some implementations, pre-trial physiological information includes ultrasound information, functional magnetic resonance imaging (fMRI) information, electrophysiological recordings, structural magnetic resonance imaging scans, diffusion tensor imaging (DTI) information, computed tomography (CT) scan information, or any combination thereof.

[0085] In any of the embodiments described herein, the machine learning algorithm includes reinforcement learning, Bayesian optimization, generalized linear models, support vector machines, deep neural networks, or any combination thereof.

[0086] In any of the embodiments described herein, the machine learning algorithm is trained offline, tested offline, validated offline, or any combination thereof.

[0087] In any embodiment of the embodiments described herein, ultrasound data includes radio frequency (RF) data or in-phase and quadrature (IQ) data.

[0088] In any embodiment of the embodiments described herein, ultrasound data includes one or more ultrasound images.

[0089] In any embodiment of the embodiments herein, one or more ultrasound images include two-dimensional images, three-dimensional images, or any combination thereof.

[0090] In some implementations, one or more ultrasound images have a resolution of 100 μm to 4 mm.

[0091] In any embodiment of the embodiments herein, the imaging volume of one or more ultrasound images comprises a spherical sector with a conical radius.

[0092] In any of the embodiments described herein, one or more ultrasound images are received at 10 Hz to 257 kHz.

[0093] In any of the embodiments described herein, the instructions for modulating neural activity are determined based on the target neural activity.

[0094] In some implementations, the target neural activity is determined by ultrasound imaging, fMRI imaging, electrophysiological recording, structural magnetic resonance imaging scans, diffusion tensor imaging (DTI), or any combination thereof.

[0095] In some implementations, the target neural activity is determined using ultrasound data.

[0096] In any of the embodiments described herein, the target neural activity is determined based on the output of a transfer learning algorithm.

[0097] In any of the embodiments described herein, the target neural activity is represented as a complex, time-independent state.

[0098] In any of the embodiments described herein, the target neural activity is represented as multidimensional time-series data.

[0099] In some implementations, the temporal or spatial resolution of the multidimensional time series data is equal to or less than the temporal or spatial resolution of the ultrasound data.

[0100] In any embodiment of the embodiments described herein, the method may further include: receiving second ultrasound data of the nervous system from an implanted transducer, wherein the second ultrasound data indicates a second physiological state of the nervous system; Processing the second ultrasound data; and transmitting the processed second ultrasound data, wherein a second instruction for regulating a second neural activity of the nervous system is determined based on the processed second ultrasound data.

[0101] In some implementations, the second instruction for regulating the second neural activity includes a modified first instruction for regulating the first neural activity.

[0102] In some implementations, adjusting the first instruction for regulating the first neural activity includes adjusting electrical regulation parameters, spatial regulation parameters, temporal regulation parameters, or any combination thereof.

[0103] In some implementations, the electrically controlled parameters include amplitude, frequency, pulse width, intensity, waveform, polarity, acoustic pressure, or any combination thereof.

[0104] In any of the embodiments described herein, spatial control parameters include electrode configuration, electrode location, electrode size, electrode placement, orientation, coil orientation, coil location, stimulation focus, stimulation bilaterality, montage, focal size, target location, or any combination thereof.

[0105] In any implementation of the embodiments herein, the timing control parameters include burst, cycle, ramp, frequency, pulse duration, string duration, inter-string interval, total number of pulses, stimulation pattern, duration, inter-stimulus interval, session frequency, pulse repetition frequency, duty cycle, or any combination thereof.

[0106] In some embodiments, the method may further include iteratively performing a receiving step, a processing step, and a transmitting step, wherein: the instructions for modulating corresponding neural activity of the nervous system are determined based on corresponding ultrasound data, and The method stops after reaching a predetermined number of iterations.

[0107] In any embodiment of the embodiments described herein, the method may further include: iteratively performing a receiving step, a processing step, and a transmitting step, wherein: the instructions for modulating corresponding neural activity of the nervous system are determined based on corresponding ultrasound data, and The method stops after reaching a predetermined number of iterations.

[0108] In some implementations, the method is stopped upon determining that the subject exhibits target neural activity for at least a predetermined duration.

[0109] In any embodiment of the embodiments described herein, the method may receive second ultrasound data of the nervous system from an implanted transducer in response to modulation of neural activity.

[0110] In any embodiment of the embodiments described herein, the method may further include: correlating second ultrasound data of the nervous system with modulated neural activity.

[0111] In any of the embodiments described herein, the second ultrasound data of the nervous system is received at a predetermined time period after the modulation of neural activity.

[0112] In any of the embodiments described herein: neural activity is regulated at a first region of the nervous system, and in response to regulation of the first region of the nervous system, a second instruction is determined for regulating a second neural activity of the nervous system at a second region of the nervous system.

[0113] In any of the embodiments described herein, the second instruction is determined at a predetermined time period following the modulation of neural activity.

[0114] In any of the embodiments described herein, the second instruction is executed during a second predetermined time period after the second instruction is determined.

[0115] In any embodiment of the embodiments herein, determining the instructions for modulating neural activity includes: determining the region of the nervous system to be modulated based on ultrasound data.

[0116] In any of the embodiments described herein, the instructions for modulating the region of neural activity are further based on a second physiological state.

[0117] In some implementations, the second physiological state is received together with the first physiological state of the nervous system.

[0118] In any of the embodiments described herein, the second physiological state includes the subject's behavior, the subject's ocular measurements, the subject's hematological measurements, or any combination thereof.

[0119] In some implementation schemes, the behavior of the subjects is determined based on their responses to questionnaires, emotional assessments, or both.

[0120] In any of the embodiments described herein, the subject's eye measurements include eye tracking or pupil dilation measurements.

[0121] In any of the embodiments described herein, the hematological measurements of the subject include blood pressure, blood glucose level, blood cholesterol level, blood hormone level, or any combination thereof.

[0122] In any of the embodiments described herein, the second physiological state is determined by a camera, microphone, wearable device, or any combination thereof.

[0123] In any of the embodiments described herein, wearable devices include electronic watches, electronic rings, or electronic glasses.

[0124] In any of the embodiments described herein, the second physiological state is associated with a positive or negative titer.

[0125] In some implementation schemes, the positive or negative potency is determined based on pre-trial physiological observations, ultrasound data, or both.

[0126] In any of the embodiments described herein, the positive or negative titer is determined experimentally.

[0127] In any of the embodiments described herein, positive or negative valence is used in part to determine the target neural activity.

[0128] In any of the embodiments described herein, the modulation of neural activity is associated with the treatment of chronic pain, depression and anxiety, obsessive-compulsive disorder, Parkinson's disease, essential tremor, epilepsy, post-traumatic stress disorder, memory impairment, or any combination thereof.

[0129] In some implementation schemes, obsessive-compulsive disorder is defined as obsessive-compulsive disorder, substance abuse disorder, or both.

[0130] In any of the embodiments described herein, the subject is a human being.

[0131] In any of the embodiments described herein, instructions for modulating neural activity are sent to the neuromodulation system via a docking device.

[0132] In any of the embodiments described herein, instructions for modulating neural activity are sent to the neural modulation system via a communication protocol.

[0133] In some implementations, the communication protocol may include USB.

[0134] In some aspects, this document discloses a method for determining instructions for modulating neural activity of a subject's nervous system, the method comprising: receiving ultrasound data of the nervous system from an implantable transducer, wherein the ultrasound data indicates the physiological state of the nervous system; processing the ultrasound data of the nervous system; and transmitting the processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data; receiving the instructions for modulating neural activity by the implantable transducer; and performing ultrasound neuromodulation on the subject by the implantable transducer.

[0135] In some respects, this paper discloses an acoustic window for a proximity ultrasound array, comprising a biocompatible polymer; and configured to allow ultrasound waves to be transmitted through the acoustic window.

[0136] In some embodiments, the biocompatible polymers include polymethyl methacrylate (PMMA), polyether ether ketone (PEEK), polychlorotrifluoroethylene (PCTFE), polytetrafluoroethylene (PTFE), ultra-high molecular weight polyethylene (UHMWPE), polyethylene terephthalate (PET), low-density polyethylene (LDPE), polyether block amide (PEBAX), and / or high-density polyethylene (HDPE).

[0137] In any of the embodiments described herein, the density of the biocompatible polymer is greater than or equal to the lower density and less than or equal to the higher density.

[0138] In some implementations, the lower density is approximately 0.31 g / cm3.

[0139] In any of the embodiments described herein, the higher density is approximately 2.75 g / cm3.

[0140] In any of the embodiments described herein, the biocompatible polymer is configured to allow the transmission of ultrasound at a speed greater than or equal to a predetermined lower speed of sound and less than or equal to a predetermined higher speed of sound.

[0141] In some implementations, the predetermined lower speed of sound is approximately 896 meters per second.

[0142] In some implementations, the predetermined higher speed of sound is approximately 3680 meters per second.

[0143] In any of the embodiments described herein, the degradation coefficient of the biocompatible polymer is greater than or equal to a predetermined lower degradation coefficient and less than or equal to a higher degradation coefficient.

[0144] In some implementations, a lower attenuation factor of approximately 0.15 dB / cm / MHz is predetermined.

[0145] In some implementations, a higher attenuation factor is predetermined to be approximately 9.27 dB / cm / MHz.

[0146] In any embodiment of the embodiments described herein, the impedance of the biocompatible polymer is greater than or equal to the lower impedance and less than or equal to the higher impedance.

[0147] In some implementations, the lower impedance is approximately 0.685 MRayl.

[0148] In some implementations, the higher impedance is approximately 2.765 MRayl.

[0149] In any embodiment of the embodiments described herein, the impedance ratio of the biocompatible polymer is greater than or equal to a predetermined lower impedance ratio and less than or equal to a higher impedance ratio.

[0150] In some implementations, a lower impedance ratio of approximately 0.625 is predetermined.

[0151] In some implementations, a higher impedance ratio of approximately 2.765 is predetermined.

[0152] In any embodiment of the embodiments described herein, the biocompatible polymer has a reflectance greater than or equal to the lower reflectance and less than or equal to the higher reflectance.

[0153] In some implementations, the lower reflectance is approximately 0.005.

[0154] In some implementations, the higher reflectance is approximately 0.215.

[0155] In any embodiment of the embodiments described herein, the biocompatible polymer has a transmittance greater than or equal to a lower transmittance and less than or equal to a higher transmittance.

[0156] In some implementations, the lower transmittance is approximately 0.785.

[0157] In some implementations, the higher transmittance is approximately 0.995.

[0158] In any embodiment of the embodiments herein, the total attenuation of the biocompatible polymer at a predetermined frequency is greater than or equal to a predetermined lower total attenuation at the predetermined frequency and less than or equal to a predetermined higher total attenuation at the predetermined frequency.

[0159] In some implementations, a lower total attenuation of approximately 0.01 dB / cm is predetermined.

[0160] In some implementations, a higher total attenuation of approximately 40.95 dB / cm is predetermined.

[0161] In any of the embodiments described herein, the predetermined frequency is approximately 5 MHz.

[0162] In any embodiment of the embodiments described herein, proximity of the encapsulated ultrasonic array to the acoustic window includes the encapsulated ultrasonic array being adjacent to the acoustic window.

[0163] In any of the embodiments described herein, the implantable transducer includes an acoustic window, an ultrasonic array, and a housing.

[0164] In any of the embodiments described herein, the implantable transducer includes a cable configured to send power or data to or from the implantable transducer.

[0165] In some respects, this paper discloses a method for assembling an implantable transducer, comprising: placing an ultrasonic array near an acoustic window; and connecting a housing to the placed ultrasonic array, wherein the placed ultrasonic array is at least partially disposed within the housing.

[0166] In some implementations, the housing includes one or more housing components.

[0167] In any of the embodiments described herein, connecting the housing to the placed ultrasound array includes connecting one or more housing assemblies to the placed ultrasound array.

[0168] In any of the embodiments described herein, assembly or connection includes the use of adhesive methods.

[0169] In some implementations, the bonding methods include laser welding, electron beam welding, TIG welding, thermal welding, epoxy sealing, or combinations thereof.

[0170] In some respects, this paper discloses a method for assembling an implantable transducer, comprising: casting a one-piece acoustically transparent housing, wherein the one-piece acoustically transparent housing includes an acoustically transparent window and an ultrasonic array, and the ultrasonic array is disposed within the cast one-piece acoustically transparent housing.

[0171] In any embodiment of the embodiments described herein, the acoustically transparent shell or acoustically transparent window comprises a biocompatible polymer.

[0172] In some embodiments, the biocompatible polymers include polymethyl methacrylate (PMMA), polyether ether ketone (PEEK), polychlorotrifluoroethylene (PCTFE), polytetrafluoroethylene (PTFE), ultra-high molecular weight polyethylene (UHMWPE), polyethylene terephthalate (PET), low-density polyethylene (LDPE), polyether block amide (PEBAX), high-density polyethylene (HDPE), or any combination thereof.

[0173] In any embodiment of the embodiments described herein, the housing comprises an acoustically transparent material that is not an acoustically transparent window.

[0174] In any of the embodiments described herein, the housing comprises a non-acoustic material.

[0175] In any embodiment of the embodiments described herein, assembly includes assembling the implantable transducer in a dry gas environment.

[0176] In any of the embodiments described herein, assembly includes sterilizing the implantable transducer.

[0177] In some implementations, sterilization includes gamma irradiation, autoclaving, ethylene oxide treatment, and / or combinations thereof.

[0178] In some aspects, this paper discloses a method for training a machine learning model, comprising: receiving one or more ultrasound data from one or more samples of one or more subjects obtained from an implanted transducer, and one or more functional ultrasound image data corresponding to the one or more ultrasound data; converting the one or more ultrasound data into one or more ultrasound arrays; converting the one or more functional ultrasound image data into one or more functional ultrasound arrays; and training a machine learning model using the one or more ultrasound arrays and the one or more functional ultrasound arrays to predict one or more inferred functional ultrasound arrays from the input one or more ultrasound data or the input one or more ultrasound arrays.

[0179] In some implementations, the machine learning model is retrained once or multiple times based on one or more additional ultrasound data, one or more additional ultrasound arrays, one or more additional functional ultrasound image data, or one or more additional functional ultrasound arrays.

[0180] In some implementations, retraining the machine learning model includes fine-tuning the machine learning model based on one or more additional ultrasound data, one or more additional ultrasound arrays, one or more additional functional ultrasound image data, or one or more additional functional ultrasound arrays.

[0181] In any embodiment of the embodiments described herein, training the machine learning model further includes determining one or more metrics that describe the relationship between one or more inferred functional ultrasound arrays and behavioral data of one or more subjects.

[0182] In any implementation of the embodiments described herein, one or more metrics include correlation metrics, regression metrics, classification metrics, model performance metrics, information theory metrics, time metrics, or cross-validation metrics.

[0183] In some implementation schemes, the correlation metrics include Pearson correlation coefficient, Spearman rank correlation coefficient, or canonical correlation coefficient (CCA).

[0184] In some implementations, the regression indicators include the R-squared indicator, the adjusted R-squared indicator, the t-statistic from the generalized linear model (GLM), or the f-statistic from the GLM.

[0185] In any implementation of the embodiments described herein, classification metrics include decoding accuracy, precision, accuracy, recall, F1 score, area under the receiver operating characteristic curve (AUC-ROC), area under the precision-recall curve (AUC-PR), or confusion matrix metrics.

[0186] In some implementations, model performance metrics include mean squared error (MSE), root mean squared error (RMSE), mean absolute error (MAE), explained variance, or log loss.

[0187] In some implementations, information-theoretic metrics include mutual information metrics or transfer entropy metrics.

[0188] In some implementations, time metrics include time signal-to-noise ratio (tSNR) or detection delay.

[0189] In any implementation of the embodiments described herein, the cross-validation metric includes the metric in which the metric has been cross-validated.

[0190] In any embodiment of the embodiments described herein, training the machine learning model includes jointly optimizing the metrics and the error between one or more inferred functional ultrasound arrays and one or more functional ultrasound arrays.

[0191] In any of the embodiments described herein, one or more metrics are used as part of a cost function during the training of the machine learning model.

[0192] In some implementations, the cost function comprises a weighted sum of one or more metrics.

[0193] In some implementations, the weighted sum of one or more metrics is dynamically adjusted during the training of the machine learning model.

[0194] In any implementation of the embodiments herein, behavioral data includes motion data, cognitive task performance data, emotional state data, or any combination thereof.

[0195] In some implementations, motion data is obtained from accelerometers, gyroscopes, or motion capture systems.

[0196] In some implementations, cognitive task performance data is based on reaction time, error rate, or task completion time.

[0197] In some implementations, emotional state data are obtained from physiological signals such as heart rate, skin conductance, questionnaires, or facial expressions.

[0198] In any of the embodiments described herein, training a machine learning model includes using regularization techniques.

[0199] In some implementations, regularization techniques include dropout, L1 regularization, or L2 regularization.

[0200] In any of the embodiments described herein, training the machine learning model includes human-machine loop techniques.

[0201] In some implementations, human-machine loop technology includes an evaluation of the inferred functional ultrasound array by a medical professional.

[0202] In any implementation of the embodiments described herein, the machine learning model includes an attention mechanism.

[0203] In any of the embodiments described herein, ultrasound data or functional ultrasound image data undergoes image enhancement.

[0204] In some implementations, image enhancement includes deconvolution or the application of super-resolution techniques.

[0205] In any of the embodiments described herein, ultrasound data or functional ultrasound image data of one or more subjects are paired with clinical metadata corresponding to one or more subjects.

[0206] In some implementations, clinical metadata includes the subject's age, sex, or medical history.

[0207] In some respects, this paper discloses a method for inferring a functional ultrasound array from one or more ultrasound data, comprising: receiving one or more ultrasound data from one or more samples from one or more subjects; converting the one or more ultrasound data into one or more ultrasound arrays, providing the one or more ultrasound arrays to a trained machine learning model; and outputting one or more inferred functional ultrasound arrays based on the received one or more ultrasound data.

[0208] In some embodiments, the method may further include converting one or more inferred functional ultrasound arrays into one or more inferred functional ultrasound image data.

[0209] In any embodiment of the embodiments described herein, one or more functional ultrasound arrays include power Doppler imaging.

[0210] In any embodiment of the embodiments herein, one or more ultrasound data include ultrasound data sequences.

[0211] In any embodiment of the embodiments herein, ultrasound data includes radio frequency data, in-phase and quadrature (IQ) data, B-mode image data, composite image data, or any combination thereof.

[0212] In any embodiment of the embodiments herein, ultrasound data includes radiofrequency data, in-phase and quadrature (IQ) data, composite image data, or any combination thereof, but excludes B-mode image data.

[0213] In any of the embodiments described herein, the trained machine learning model is trained based on training data including ultrasound data and functional ultrasound image data.

[0214] In any of the embodiments described herein, at least a portion of the training data includes normalized image data or augmented image data.

[0215] In some implementations, normalized image data includes color-normalized image data.

[0216] In some implementations, enhancing image data includes image data that has already been enhanced by removing noise from the image data, increasing the contrast of the image data, adjusting the brightness of the image data, performing convolution with an image kernel, and / or geometric transformations.

[0217] In some implementations, convolution with an image kernel includes convolution with a Gaussian blur kernel, a box blur kernel, an edge detection kernel, a sharpening kernel, an unsharpening masking kernel, or any combination thereof.

[0218] In some implementations, geometric transformations include affine transformations, elastic transformations, flips, clipping, mesh distortion, optical distortion, perspective transformations, transposes, or any combination thereof.

[0219] In some implementations, affine transformations include translation, rotation, scaling, shearing, or any combination thereof.

[0220] In any of the embodiments described herein, the training data is divided into a first training data portion, a first test data portion, and a validation data portion.

[0221] In some implementations, the first training data portion includes 70%, 75%, 80%, 85%, or 90% of the training data, the first test data portion includes 20%, 18%, 15%, 13%, 10%, or 5% of the training data, and the validation data portion includes 20%, 18%, 15%, 13%, 10%, or 5% of the training data.

[0222] In any of the embodiments described herein, the validation data portion includes one or more training image blocks, while the first training data portion includes all training images other than the one or more training images in the validation data portion.

[0223] In any implementation of the embodiments described herein, the training data is divided into a second training data portion and a second test data portion.

[0224] In some implementations, the second training data portion includes 60%, 65%, 70%, 75%, or 80% of the training data, and the second test data portion includes 40%, 35%, 30%, 25%, or 20% of the training data.

[0225] In any of the implementation schemes described herein, the training data undergoes cross-validation.

[0226] In some implementations, cross-validation includes k-fold cross-validation, p-fold leave cross-validation, one-fold leave cross-validation, hierarchical k-fold cross-validation, repeated k-fold cross-validation, nested k-fold cross-validation, or Monte Carlo cross-validation.

[0227] In any implementation of the embodiments described herein, the machine learning model includes an encoder-decoder architecture.

[0228] In any of the embodiments described herein, the machine learning model includes a 3D convolutional filter.

[0229] In some implementations, 3D convolutional filters are configured to extract one or more spatiotemporal features from one or more ultrasound data, one or more ultrasound arrays, one or more functional ultrasound image data, or one or more functional ultrasound arrays.

[0230] In any implementation of the embodiments described herein, the machine learning model includes residual blocks.

[0231] In any implementation of the embodiments described herein, the machine learning model includes a convolutional neural network (CNN).

[0232] In some implementations, CNNs include convolutional functions, activation functions, pooling functions, or any combination thereof.

[0233] In some implementations, the convolution function includes convolving the matrix from the input with a kernel.

[0234] In some implementations, the kernel is randomly initialized and learned by training a neural network.

[0235] In some implementations, learning includes backpropagation and optimization.

[0236] In some implementations, the optimization includes gradient descent, stochastic gradient descent, batch gradient descent, mini-batch gradient descent, Adam optimization, AdaGrad optimization, RMSprop optimization, momentum optimization, or any combination thereof.

[0237] In any of the embodiments described herein, the activation function is a modified linear unit (ReLU) function, a leaky ReLU function, a linear activation function, a nonlinear activation function, a sigmoid activation function, or a hyperbolic tangent activation function.

[0238] In any implementation of the embodiments described herein, the pooling function is a max pooling function, an average pooling function, or an attention-based pooling function.

[0239] In any implementation of the embodiments described herein, the machine learning model also includes a softmax function or an argmax function.

[0240] Incorporate by reference All publications, patents, and patent applications mentioned in this specification are incorporated herein by reference in their entirety, as if each individual publication, patent, or patent application were specifically and individually indicated to be incorporated herein by reference in its entirety. In the event of any conflict between terminology used herein and terminology found in the incorporated references, the terminology used herein shall prevail. Attached Figure Description

[0241] This patent or application document contains at least one color drawing. A published copy of this patent or application with a color drawing will be provided by the Patent Office upon request and payment of the necessary fees.

[0242] The appended claims specifically set forth various aspects of the disclosed methods, apparatus, and systems. A better understanding of the features and advantages of the disclosed methods, apparatus, and systems will be obtained by referring to the following detailed description of illustrative embodiments and accompanying drawings, wherein: FIG. 1 Schematic diagrams are provided representing example systems that can be used for ultrasound-based imaging or modulation of physiological activities according to some implementation schemes.

[0243] FIG. 2A Examples of non-implanted configurations of peripheral controllers and transducer devices (e.g., ultrasonic transducers) capable of delivering ultrasonic pulses according to some implementation schemes are provided.

[0244] FIG. 2B Examples of non-implanted configurations of peripheral controllers and transducer devices (e.g., ultrasonic transducers) capable of delivering ultrasonic pulses according to some implementation schemes are provided.

[0245] FIG. 3A and FIG. 3B Examples of implantable transducers capable of delivering ultrasonic pulses according to some implementation schemes are provided.

[0246] FIG. 4A to FIG. 4CAn external view of an example implantable transducer capable of delivering ultrasound pulses according to some implementation schemes is provided.

[0247] FIG. 5 A schematic diagram of an example circuit illustrating an ultrasonic transducer capable of delivering ultrasonic pulses according to some embodiments is provided.

[0248] FIG. 6A and FIG. 6B Renderings of example ultrasonic transducer devices based on some implementation schemes are provided.

[0249] FIG. 7 Example ultrasonic transducer devices according to some implementation schemes are provided.

[0250] FIG. 8A Cross-sectional renderings of example ultrasonic transducer devices according to some implementation schemes are provided.

[0251] FIG. 8B Renderings of the exterior of an example ultrasonic transducer device according to some implementation schemes are provided.

[0252] FIG. 8C Line drawings of the exterior of an example ultrasonic transducer device according to some implementation schemes are provided.

[0253] FIG. 8D Line drawings of the exterior of an example ultrasonic transducer device according to some implementation schemes are provided.

[0254] FIG. 8E A rendering of the exterior of an example ultrasonic transducer device is provided.

[0255] FIG. 9A and FIG. 9B Exploded view renderings of example ultrasonic transducer devices according to some implementation schemes are provided.

[0256] FIG. 10A and FIG. 10B Renderings of example ultrasonic transducer devices based on some implementation schemes are provided.

[0257] FIG. 11 An exemplary workflow for assembling an ultrasonic transducer device according to some implementation schemes is described.

[0258] FIG. 12 A schematic diagram is provided depicting an ultrasonic transducer device connected to a personal computer according to some embodiments.

[0259] FIG. 13A An example peripheral controller is provided for one or more ultrasonic transducer devices that can control and deliver ultrasonic pulses according to some implementation schemes.

[0260] FIG. 13B An example peripheral controller is provided for one or more non-implantable transducers that can control and deliver ultrasonic pulses according to some implementation schemes.

[0261] FIG. 13C An example peripheral controller is provided for one or more implantable transducers that can control and deliver ultrasonic pulses according to some implementation schemes.

[0262] FIG. 14A and FIG. 14B Example peripheral controllers are provided for connecting to multiple transducers capable of delivering ultrasonic pulses, according to some implementation schemes.

[0263] FIG. 15A Illustrations are provided illustrating example uses of implantable transducers according to some implementation schemes.

[0264] FIG. 15B Illustrations are provided illustrating example implementations of multiple implanted transducers in the skull of a subject according to some implementation schemes.

[0265] FIG. 15C Example ultrasound transmission patterns for imaging and neuromodulation applications are provided according to some implementation schemes.

[0266] FIG. 15D Illustrations are provided illustrating example uses of several implantable transducers according to some implementation schemes.

[0267] FIG. 16A An exemplary workflow for iteratively modulating the nervous system based on observed neural activity is described according to some implementation schemes.

[0268] FIG. 16B An exemplary workflow for iteratively modulating the nervous system via ultrasound neuromodulation based on observed neural activity, according to some implementation schemes, is described.

[0269] FIG. 17A and FIG. 17B A schematic diagram representing example neural activity states according to some implementation schemes is provided.

[0270] FIG. 18 A schematic diagram is provided to represent an example neural activity state after neural modulation based on modulation parameters according to some implementation schemes.

[0271] FIG. 19 An exemplary schematic diagram is depicted, illustrating the adjustment of neural modulation parameters to achieve a desired brain state.

[0272] FIG. 20An exemplary workflow for iteratively modulating the brain based on observed neural activity is described according to some implementation schemes.

[0273] FIG. 21 An exemplary workflow for iteratively modulating the brain based on observed neural activity is described according to some implementation schemes.

[0274] FIG. 22 Example diagnostic and / or treatment platforms based on some implementation schemes are provided.

[0275] FIG. 23A An exemplary method for training a machine learning model for reconstructing one or more functional ultrasound images, according to some implementation schemes, is described.

[0276] FIG. 23B Exemplary methods for reconstructing functional ultrasound images from anatomical ultrasound images, according to some implementation schemes, are described.

[0277] FIG. 24 Example computing devices or systems according to some implementation schemes are described.

[0278] FIG. 25 Example computer systems or computer networks according to some implementation schemes are described.

[0279] FIG. 26 Examples of data depicting simulation results based on some implementation schemes are provided.

[0280] FIG. 27A to FIG. 27J Examples of data associated with simulation-based imaging designs based on some implementation schemes are provided.

[0281] FIG. 28 Examples of data depicting simulation results based on some implementation schemes are provided.

[0282] FIG. 29A An example diagram is provided showing how pressure is applied to a subject's brain when a pulse wave amplitude is delivered using an ultrasonic transducer device.

[0283] FIG. 29B Provided for FIG. 29A The example diagram shown illustrates how an ultrasonic transducer device can deliver pulse wave amplitudes to a material, enabling the delivery of pulse wave amplitudes. FIG. 29B Overlay FIG. 29A superior.

[0284] FIG. 30A Example 3D diagrams are provided showing the pressure applied to a subject from two non-implantable ultrasound transducer devices according to some implementation schemes.

[0285] FIG. 30BExample 3D diagrams are provided showing the pressure applied to a subject from three implanted ultrasound transducer devices according to some implementation schemes.

[0286] FIG. 31A to FIG. 31G Example data on the ultrasonic imaging quality and transparency of an ultrasonic transducer comprising one of several acoustically transparent materials, according to descriptions of some implementation schemes, are provided.

[0287] FIG. 32 Example tabular data on the physical properties of ultrasound waves passing through one of several acoustically transparent materials, according to descriptions of some implementation schemes, are provided.

[0288] FIG. 33 Images of example ultrasonic transducers configured to deliver ultrasound waves, being tested in a laboratory according to some implementation schemes, are provided.

[0289] FIG. 34A to FIG. 34E Data on the characteristics of one or more ultrasonic pulses delivered by an ultrasonic transducer device in a laboratory, according to descriptions of some implementation schemes, are provided.

[0290] FIG. 35A to FIG. 35C Example hydrophone data are provided for one or more ultrasonic pulses that vary spatially when delivered by a non-implantable ultrasonic transducer device, according to some implementation schemes.

[0291] FIG. 36A to FIG. 36C Example simulation data of one or more ultrasonic pulses that vary spatially when delivered by an implantable ultrasonic transducer device, according to some implementation schemes, are provided.

[0292] FIG. 37A Example activity diagrams generated by ultrasound imaging of the brain according to some implementation schemes are provided.

[0293] FIG. 37B Provided based on FIG. 37A Example neural activity traces of the indicated ROI.

[0294] FIG. 38 Images of example ultrasonic transducers for applying one or more ultrasonic pulses to a subject according to some implementation schemes are provided.

[0295] FIG. 39A A schematic diagram of an example coronal section of the human brain according to some implementation schemes is provided.

[0296] FIG. 39B Example imaging data of neural activity from the brain of a subject obtained by ultrasound imaging using an ultrasound transducer, according to some implementation schemes, are provided.

[0297] FIG. 40A and FIG. 40BExample imaging data of neural activity from the brain of a subject obtained by ultrasound imaging using an ultrasound transducer, according to some implementation schemes, are provided.

[0298] FIG. 41A to FIG. 41E Example imaging data of neural activity from the brain of a subject obtained by ultrasound imaging using an ultrasound transducer, according to some implementation schemes, are provided.

[0299] FIG. 42A to FIG. 42B Example imaging data of neural activity from the brain of a subject obtained by ultrasound imaging using an ultrasound transducer, according to some implementation schemes, are provided.

[0300] FIG. 43A to FIG. 43D Example power Doppler imaging data of neural activity from the mouse brain according to some implementation schemes are provided, including imaging data of the mouse brain reconstructed based on a trained artificial neural network model. Detailed Implementation

[0301] This paper discloses devices, methods, and systems for imaging and stimulating the brain using ultrasound. Based on devices that read from or write to the brain, neurotechnology holds the promise of curing neurological and psychiatric disorders, enhancing human experiences by improving cognition, memory, and sleep, and enabling high-bandwidth communication between humans and technology, as well as between humans themselves. However, neurotechnology has not yet delivered on this promise because current approaches are either imperfect or poorly matched to the biological characteristics of the brain. For example, current neurotechnology often sacrifices temporal and / or spatial scale and resolution. The methods and systems disclosed herein address this technological shortcoming in the art. That is, the methods and systems disclosed herein provide the capability to monitor and manipulate the neural activity of human subjects at sufficiently broad yet precise temporal and spatial scales and resolution, even during clinically relevant behaviors, without penetrating the subject's brain.

[0302] Due to inherent physical and neurophysiological constraints, existing systems and methods struggle to provide adequate scale and resolution for monitoring and manipulating neural activity. For example, while electrophysiological systems offer sufficient temporal resolution in terms of sampling frequency, they are generally unsuitable for prolonged recording sessions due to the surgically invasive nature of these techniques (e.g., direct implantation into the subject's brain). The feasibility of monitoring electrophysiological activity for weeks or months typically results in prolonged contact between the probe and the subject's brain, potentially increasing the risk of infection. Electrophysiological methods also tend to offer very specific spatial resolution, but at the cost of coverage. To achieve even mesoscale monitoring of neural activity, multiple recording probes are often required. The invasiveness and poor scalability of electrophysiological methods limit the impact of such techniques.

[0303] Functional magnetic resonance imaging (fMRI) is another existing technique that is generally not ideal for reading and writing neural activity. Magnetic fields (like those generated by fMRI machines) can easily penetrate and image deep brain structures, but fMRI machines and fMRI-based techniques are difficult to miniaturize. Therefore, fMRI is generally not suitable for real-time localized interaction with neuronal circuits. Furthermore, because the bulkiness of fMRI machines typically requires subjects to remain still, fMRI often cannot record neural activity during clinically relevant behaviors. In short, existing modalities capable of reading and writing neural activity struggle to balance invasiveness and performance.

[0304] The methods and systems disclosed herein bridge the gap between invasiveness and performance in current neurotechnologies. The methods and systems discussed herein utilize the physical properties of ultrasound. Ultrasound has a wavelength of approximately 100 μm and travels through soft tissue at a speed similar to sound, approximately 1.5 km / s. These physical properties allow ultrasound energy to propagate throughout the brain with a spatial resolution of approximately 100 μm and a temporal resolution of approximately 1 ms. Ultrasound technology can be focused deep into tissues. For example, focused ultrasound (FUS) is a rapidly developing therapeutic approach for neuromodulation and tissue ablation. Ultrasound's low cost, portability, and safety make it suitable for clinical medicine and facilitate its application in macroscopic brain imaging and modulation. Ultrasound-based physics-based techniques can bridge the gap between invasiveness and performance. The methods and systems described herein relate to a brain-computer interface (BCI) that can use ultrasound to image and / or modulate the neural activity of a subject's nervous system. Imaging and / or modulation of neural activity can be achieved through an implanted ultrasound transducer. The transducer can propagate ultrasound waves and record the patterns of reflected waves to infer and / or modulate the subject's neural activity. The transducers can be connected to a peripheral controller that can provide power and / or organize the activity of multiple transducers (e.g., a group of transducers). The peripheral controller can offload the collected neural activity data to an external server for further processing, which may include analyzing the data based on algorithms, including machine learning algorithms.

[0305] BCI based on ultrasound physics, as described herein, can be implemented in a closed-loop approach that includes strategic neuromodulation of a subject based on ultrasound imaging. For example, imaging and modulation of the nervous system can be complementary, allowing for iterative processes of imaging and modulation to achieve a target state of neural activity (e.g., for treating neurological disorders or diseases). Both the iteration of imaging and modulation can be based on ultrasound physics, such as through implanted ultrasound transducers. In some examples, the iteration of imaging and modulation can be based on both ultrasound-based and electrophysiological methods, such as imaging a subject's neural activity using an implanted ultrasound transducer and modulating that activity using electrodes.

[0306] Beyond providing high-performance and relatively non-invasive techniques for imaging and manipulating neural activity, the systems and methods disclosed herein can also advance other fields. For example, advances in molecular biology and gene therapy are paving the way for further progress in areas where the methods and systems discussed herein will accelerate. For instance, when combined with intravenous microbubbles, ultrasound can temporarily open the blood-brain barrier to allow for precise drug delivery to the brain. Precise delivery can also be achieved by using ultrasound to uncage drugs. The sonogenetic pathway can also leverage the interaction between ultrasound and genetically modified cells to facilitate targeted manipulation of neural activity. The methods and systems disclosed herein can accelerate the development of these approaches and open new avenues in neurology and personalized medicine.

[0307] The systems and methods disclosed herein can also advance silicon manufacturing. When combined with state-of-the-art low-power integrated circuits, on-chip ultrasound technology enables systems and methods for reading and writing neural activity, such as those described herein. The methods and systems described herein complement recent advances in other fields and enable longitudinal recording and modulation of neural activity at macroscopic scales with significantly improved temporal and spatial resolution.

[0308] The section headings used in this article are for organizational purposes only and should not be construed as limiting the topics described.

[0309] Definitions Unless otherwise defined, all technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains.

[0310] As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” include plural references unless the context clearly indicates otherwise. Unless otherwise stated, any reference to “or” herein is intended to cover “and / or.”

[0311] "About" and "approximately" generally refer to the degree of error acceptable for a measured quantity given the nature or precision of the measurement. Exemplary error levels are within 20 percent (%) of a given value or range of values, typically within 10%, and more usually within 5%.

[0312] As used herein, the terms “comprising” (and any form or variation thereof, such as “comprise” and “comprises”), “having” (and any form or variation thereof, such as “have” and “has”), “including” (and any form or variation thereof, such as “includes” and “include”), or “containing” (and any form or variation thereof, such as “contains” and “contain”) are inclusive or open-ended and do not exclude additional, unmentioned additives, components, integers, elements, or method steps.

[0313] As used herein, in claims, the use of sequential terms (such as "first," "second," "third," etc.) to modify claim elements does not imply any priority, precedence, or order of one claim element over another, or the sequence of actions of the method, but is merely a marker to distinguish one claim element having a certain name from another element having the same name (other than the use of sequential terms). Similarly, the use of a), b), etc., or i), ii), etc., does not imply any priority, precedence, or order of steps in the claims. Similarly, the use of these terms in the specification does not imply any desired priority, precedence, or order.

[0314] As used herein, the terms “individual,” “patient,” or “subject” are used interchangeably and refer to any single animal requiring treatment, such as a mammal (including non-human animals such as, for example, dogs, cats, horses, rabbits, zoo animals, cattle, pigs, sheep, and non-human primates). In a particular embodiment, the individual, patient, or subject herein is a human being.

[0315] As used herein, “treatment” (and its grammatical variations, such as “treat” or “treating”) refers to a clinical intervention (e.g., administration of an anticancer drug or anticancer therapy) that attempts to alter the natural course of an individual being treated, and may be performed for prevention or during the clinicopathological process. Desired therapeutic effects include, but are not limited to: preventing the onset or recurrence of disease, relieving symptoms, eliminating any direct or indirect pathological consequences of the disease, preventing metastasis, slowing the rate of disease progression, improving or alleviating the disease state, and mitigating or improving prognosis.

[0316] As used herein, when referring to subjects, “modulation” can refer to the process of altering the nervous system activity of a subject. Modulation of the nervous system can be termed “neuromodulation.” Neuromodulation using ultrasound can involve delivering focused ultrasound waves to specific areas of the nervous system to influence neural activity. Modulation can have a variety of effects on subjects, such as stimulating or inhibiting neural discharges, altering synaptic transmission, modifying neurotransmitter release, and affecting intracellular signaling pathways. Modulation can be used to enhance or inhibit neural signals and / or alter how the regulated area responds to endogenous or exogenous neural signals, but may not include directly stimulating or inhibiting the subject's neural activity.

[0317] As used herein, “imaging via ultrasound” (e.g., ultrasound imaging) can refer to various types of imaging, including non-functional imaging (e.g., anatomical imaging) and / or functional imaging. Ultrasound imaging can include qualitative ultrasound imaging or quantitative ultrasound imaging. Non-functional imaging (e.g., anatomical imaging) can be optimized to capture detailed anatomical structures of the nervous system without capturing time-varying values ​​of physiological processes (e.g., dynamics). Functional imaging can be optimized to capture dynamic physiological processes, such as blood flow, and can be used to infer blood flow in a subject, such as cerebral blood flow, which in turn can be used to infer the subject’s neural activity. Quantitative ultrasound imaging can involve the measurement and analysis of ultrasound properties, thereby providing more information about tissue properties and composition. Furthermore, as used herein, “imaging via ultrasound” does not necessarily strictly refer to obtaining images; for example, it can be interpreted by visual correspondence as visualized data mapped onto physiological landmarks. Imaging via ultrasound refers to obtaining any data generated by sending ultrasound waves to a subject, which can subsequently be visualized by manipulating the data. Such data can include, but is not limited to, radiofrequency data and / or derived data thereof, or in-phase and orthogonal (IQ) data and / or derived data thereof.

[0318] Ultrasound-based brain-computer interface (BCI) system The methods and systems described herein for monitoring and modulating neural activity may include multiple components. These may include, but are not limited to, at least one implantable transducer (e.g., an ultrasound transducer), a peripheral controller for the implantable transducer, an acoustically transparent material for encapsulating the ultrasound-based medical device for the implantable transducer, and methods for analysis related to ultrasound imaging and stimulation (such as, but not limited to, multi-transducer imaging algorithms). Combinations of the above components may constitute systems or methods related to macroscopic BCI.

[0319] FIG. 1 An exemplary macroscale BCI system 100 is illustrated. The exemplary system 100 may include one or more implantable transducers 102, a peripheral controller 120, and one or more tethers 130 configured to couple each implantable transducer 102 to the peripheral controller 120. In some embodiments, the tether 130 may be attached to a rigid body of the implantable transducer 102 and configured to send and receive data from the peripheral controller 120. In some embodiments, the tether 130 may be configured to send power from a battery residing in the peripheral controller 120 to the implantable transducer 102. In some embodiments, one or more implantable transducers may be configured to be mounted on the head of a subject and may be collectively referred to as head unit 110. In some embodiments, the peripheral controller 120 is configured as an implantable unit and is configured to be implanted in the chest of a subject for controlling one or more implantable transducers 102 (e.g., head unit 110). An implantable transducer may include a rigid body configured to be mounted into the skull such that the device has minimal impact on the subject's live activities. The implantable transducer may be configured to detect and / or modulate blood flow or other aspects related to brain tissue.

[0320] In some embodiments, the implantable transducer 102 can be tethered to an external unit for power and data processing, and can function and be compatible with existing technologies, thereby enabling efficient scalability and implementation. In another embodiment, the macroscale BCI system may include a fully integrated, implantable system, making it suitable for long-term free-living clinical research and treatment.

[0321] Implantable transducer 102 In addition to ultrasound data (such as those according to the implementation scheme described herein, including Figures 16 to 17) FIG. 21 Beyond the broader scale provided by the data acquired by the method shown, ultrasound-based imaging allows for greater flexibility when used for multimodal or multidimensional measurements. One or more implantable transducers comprising an ultrasound array can be used to receive ultrasound data.

[0322] The implantable transducer 102 can be minimally invasive because each transducer can be installed in a small hole in the subject's skull without excessive penetration into the brain tissue. The transducer 102 can be located at the top of the brain, flush with the skull, so that the physical size of the transducer 102 does not penetrate across the subarachnoid space and into the subject's brain.

[0323] Furthermore, the implantable ultrasound transducer 102 can be highly miniaturized. For example, the overall volume of the transducer 102 can be comparable to that of a coin. The convenient shape factor of the ultrasound transducer 102 allows for the maintenance of broad and behavioral repertoire. In contrast, traditional methods for observing neural activity, such as fMRI and some electrophysiological techniques, can be cumbersome, invasive, and disrupt the subject's natural behavioral patterns. For example, fMRI may require the subject to lie relatively still in a small imaging chamber. Therefore, the ability to identify neural activity states indicative of brain disorders may be limited. Baseline behavior, as a known indicator of a brain disorder, cannot be correlated with ongoing neural activity patterns because the limitations of the fMRI machine may prevent the subject from exhibiting known behavioral indicators of a brain disorder. Similarly, many electrophysiological systems used to observe neural activity may constrain the subject's natural behavior. For example, they often require invasive surgical implantation, which may require inserting one or more electrodes into the subject's brain. Some traditional methods, such as stereotactic electroencephalography (sEEG), are reserved for subjects with refractory forms of epilepsy. Ultrasound-based imaging is less invasive and less constrained by the subject's natural behavioral patterns. Furthermore, the implantable ultrasound transducer 102 can penetrate deep into the subject's brain tissue, allowing for clearer observation of deeper brain regions. Compared to conventional techniques, the combination of a wider imaging scale and greater flexibility offered by ultrasound-based imaging provides advantages in the observation-modulation paradigm of neural activity. These advantages also allow for the integration of the disclosed system with neurosensing or neuromodulation systems such as MRI, fMRI, PET, EEG, TMS, or optogenetic tools.

[0324] In addition to the miniaturization capabilities of the implantable transducer, the implantable transducer 102 also provides a wealth of data that allows for the determination of neuromodulation commands. For example, multiple aspects of ultrasound data can allow for the determination of commands for neuromodulation: imaging resolution, imaging volume, and capture frequency, as will be described in more detail herein.

[0325] The methods and systems described herein may include an ultrasonic neurosensing and stimulation device enclosed in a rigid housing designed for mounting into a living human skull, such as... FIG. 2A to FIG. 3A As shown. FIG. 2A to FIG. 2BTwo configurations are shown in which the ultrasound transducer and the peripheral controller (e.g., a chest unit) that controls the ultrasound transducer can work together to provide a platform for ultrasound-based imaging and modulation. FIG. 2A In this embodiment, the ultrasound transducer 202 and the peripheral controller 206 are not implanted in a manner that allows for free movement of the subject. For example, given... FIG. 2A The described configuration means that the subject may not be able to move around with the ultrasound transducer 202 and peripheral controller 206 after surgery. FIG. 2A In this configuration, cable 204 extends from the subject's body to a peripheral controller 206, which is also not implanted in the subject's body. The cable can be delivered percutaneously to the subject. Based on the non-implanted peripheral controller, the ultrasound transducer 202 does not require implantation but can be provided during surgery, for example, only for the duration of the surgery.

[0326] In contrast, FIG. 2B In this embodiment, an ultrasound transducer 208, a cable 210, and a peripheral controller 212 are implanted in the subject, allowing the subject to move freely after surgery. Given that the ultrasound transducer 208 is implanted, the ultrasound transducer can be an implantable transducer, and given that the peripheral controller 212 is implanted (e.g., near the subject's chest), the peripheral controller can be a chest unit. The ultrasound transducer in the ultrasound transducer 208 can be configured to fit into a hole (e.g., a drill hole) in the skull and positioned to contact the dura mater, completely outside the brain parenchyma. The drill hole need not be strictly circular in shape and can be any shape, such that the drill hole optimally accommodates the physical position of the ultrasound transducer in the ultrasound transducer 208; for example, the ultrasound transducer can be paired with the drill hole. Therefore, the shape factor of the ultrasound transducer is not intended to limit the scope of this disclosure, and other shapes are contemplated.

[0327] FIG. 3A and FIG. 3B Examples of implantable transducers capable of delivering ultrasound pulses are provided. In some implementations, such as... FIG. 3A As shown, the implantable ultrasound transducer 304 can be positioned to contact the soft tissue of a subject, such as, but not limited to, muscle, fat, blood vessels, nerves, tendons, or any combination thereof. Due to its use in the body, the implantable ultrasound transducer 304 can be hermetically sealed within a housing 302 and sterilized. The implantable transducer 304 can be connected to a cable 306, which can provide power and / or data to and / or from the implantable transducer 302. The cable 306 can connect the implantable transducer 202 to a peripheral controller, such as... FIG. 13A to FIG. 13C Further details are shown, and / or connection to other implantable transducers, such as FIG. 14A and FIG. 14B As shown.

[0328] FIG. 3B An exemplary schematic diagram of an implantable transducer 312 according to an embodiment of the present disclosure is shown. FIG. 3B As indicated, the implantable transducer 308 may include a circuit board (PCB) 310, an ultrasonic ASIC 312, a heat sink 314, and a microelectromechanical system (MEMS) ultrasonic array 316, all of which may be included within a housing (e.g., a shell) that may include an acoustically transparent window. The acoustically transparent window may include one or more layers disposed within the housing of the implantable transducer 308. In some embodiments, the mechanical shell may be made of a biocompatible, chemically stable, and corrosion-resistant material (e.g., a polymer) to withstand the physiological environment imposed by the subject and / or to have optimal heat dissipation characteristics. In some examples, the shell may be made of medical-grade stainless steel 316L or titanium. The mechanical shape of the shell may be approximately cylindrical, but the cylindrical shape factor is not intended to limit the scope of this disclosure, and other shapes are contemplated. The implantable transducer 312 may also have a power source that can be charged via inductive charging.

[0329] FIG. 4A to FIG. 4C An external view of an implantable ultrasound transducer is provided, which may be based on... FIG. 3A to FIG. 3B The implantable ultrasound transducer shown. FIG. 4A to FIG. 4C The implantable ultrasound transducer 400 shown can accommodate a dual-function closed-loop system for both imaging and modulating the nervous system of a subject. FIG. 4A to FIG. 4C The illustrated implantable ultrasound transducer 400 may include a cable 404 that can provide bidirectional data and / or unidirectional power from a peripheral controller to the ultrasound transducer 400. The data communication protocol may operate according to a standardized communication protocol, such as the USB protocol, e.g., USB 3 or a sub-version of USB 3. Alternatively, the data communication protocol may operate according to a non-standardized communication protocol, such as a proprietary communication protocol. FIG. 4A to FIG. 4CAn acoustic window 402 is also depicted through which ultrasound waves are transmitted from an ultrasound array (e.g., a MEMS ultrasound array) contained within a housing of the implantable ultrasound transducer 400. The acoustic window 402 can form a barrier between the ultrasound array contained within the housing of the implantable ultrasound transducer 400 and the physiological environment (e.g., the subject's subarachnoid space or brain), allowing ultrasound waves (e.g., sonic ultrasound) to be transmitted through the acoustic window 402 of the housing without significant attenuation or distortion, while the internal electronics of the ultrasound transducer remain protected. In some embodiments, the sound-transmitting material may include polymethyl methacrylate (PMMA), polyether ether ketone (PEEK), polychlorotrifluoroethylene (PCTFE), polytetrafluoroethylene (PTFE), ultra-high molecular weight polyethylene (UHMWPE), polyethylene terephthalate (PET), low-density polyethylene (LDPE), polyether block amide (PEBAX), and / or high-density polyethylene (HDPE), or other biocompatible polymers with low acoustic impedance and minimal ultrasonic attenuation (i.e., sound-transmitting materials) for use in the sound-transmitting window 402.

[0330] Implantable transducer: ultrasound ASIC 312 In one or more embodiments, the ultrasonic ASIC 312 may include one or more analog front-end (AFE) circuits, such as FIG. 5 As shown. In one or more embodiments, at least one AFE circuit may be located on a single ASIC. In some embodiments, the AFE circuit may span two or more ASICs. In some embodiments, the AFE circuit or a portion thereof may be distributed across one or more ASICs and / or board 310. For example, one or more board assemblies may contribute to the functionality of the AFE. The AFE circuit may be responsible for transmitting ultrasonic energy, as well as amplifying and digitizing the signals received by the ultrasonic ASIC 312.

[0331] The AFE circuit can be designed based on one or more factors. For example, power consumption and heating may be issues, as excessive temperature rise can damage surrounding tissue (there may be a 2°C rise limit allowed by the surrounding tissue). The AFE circuit can be designed to minimize power consumption, reduce overhead or heat dissipation requirements of the head unit 110, and / or accommodate the limited battery capacity available in the implantable transducer 304 and / or head unit 110.

[0332] Alternatively or concurrently, the AFE circuit can be designed based on the amount of offloadable data. In some cases, the disclosed methods and systems retain as much raw data as possible for scientific discovery and for image processing and algorithm development. For example, the raw data rate obtained from a 10-bit Nyquist sampling ADC of 10k CMUT (Capacitive Micromechanical Ultrasonic Transducer) elements can produce 1 Tbps of data. Depending on the intended application, this amount of data may be excessive, both compared to the fundamental information that the disclosed methods and systems attempt to measure and the technical challenges of offloading and processing it.

[0333] The choice of semiconductor technology for AFE circuits directly impacts the overall performance, power consumption, and / or high-voltage requirements of the circuit. Furthermore, the methods and systems discussed herein can utilize MEMS-compatible processes on wafers. Wafers can be of any size, such as, but not limited to, approximately 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 inches. BCD (Bipolar CMOS-DMOS) or silicon-on-insulator (SOI) processes are suitable for fabricating AFE circuits because both BCD and SOI processes allow for high-voltage devices on conventional CMOS wafers while also offering the benefits of smaller process nodes. These processes can be fabricated according to one or more fabs, and for a given fab, the process can be fabricated according to one of several different options. For example, a 180 nm BCD 8-inch process can meet the requirements for high-voltage device wafer size and overall analog performance.

[0334] FIG. 5 The illustrated system-level architecture of the AFE improves the overall performance and efficiency of the ultrasound imaging system. The AFE's system-level architecture can be designed for optimal interfacing with the high-capacitance inputs of a capacitive micromechanical ultrasonic transducer (CMUT) array and for amplifying signals with time-varying and beamforming requirements. The AFE may include at least a transmitter (TX), a low-noise amplifier (LNA), a time gain control (TGC), and a successive approximation register analog-to-digital converter (SAR ADC). For this purpose, these components can be arranged across transducer elements and work together to form a coherent and efficient signal processing chain.

[0335] AFE circuit: data transfer In some examples, a low-noise amplifier (LNA) can be the first stage of the receiver chain in the AFE, responsible for amplifying the signal received from the CMUT array without introducing significant noise or distortion. For CMUT ultrasonic transducers, a transimpedance amplifier (TIA) architecture may be appropriate because it is capable of converting the small capacitive current generated by the CMUT components into a voltage signal while providing sufficient gain and bandwidth for the application. The TIA input impedance can be designed to match the CMUT array to minimize input-dependent noise.

[0336] Following LNA, a Time Gain Control (TGC) stage 504 can be used to compensate for signal attenuation caused by ultrasound waves traveling through different tissue layers. The TGC 504 applies a time-varying gain to the amplified signal, compensating for distance-dependent attenuation by enhancing weaker signals that have traveled deeper into the tissue, while maintaining overall dynamic range. Applying a time-varying gain to the amplified signal helps ensure proper balance of signals from different depths, allowing for more accurate and detailed reconstruction of ultrasound images.

[0337] After the TGC stage, an analog-to-digital converter (ADC) 506 can be used to digitize the data for further processing before offloading to external devices. The successive approximation register (SAR) ADC offers a combination of high resolution, high-speed conversion, and low power consumption, making this technique ideal for ultrasound imaging applications.

[0338] Given the imaging approach disclosed herein, the estimated raw data rate can be values ​​within the following range, such as, but not limited to, approximately 100 Mb / s, 120 Mb / s, 140 Mb / s, 160 Mb / s, 180 Mb / s, 200 Mb / s, 220 Mb / s, 240 Mb / s, 260 Mb / s, 280 Mb / s, 300 Mb / s, 320 Mb / s, 340 Mb / s, 360 Mb / s, 380 Mb / s, 400 Mb / s, 420 Mb / s, 440 Mb / s, 460 Mb / s, 480 Mb / s, and 500 Mb / s per functional image. In some embodiments, the ultrasonic transducer may use a wired transceiver to transmit data to an external processing unit for scientific research and the development of more advanced algorithms. In some embodiments, wireless data transmission technology may be used.

[0339] Implantable transducer: AFE circuit for signal transmission According to embodiments of this disclosure, the AFE circuit may further include transmitter circuitry for the ultrasonic array. The design of the transmitter circuitry for the ultrasonic array may involve a combination of analog components and digital control logic to provide precise beamforming and high-resolution imaging. Basic components of the analog circuitry may include a pulse generator, a high-voltage switching matrix, and a digital-to-analog converter (DAC). The pulse generator generates high-frequency, short-duration electrical pulses required to excite the CMUT elements. These pulses can be selectively delivered to individual elements in the array via the high-voltage switching matrix. The DAC can generate finely tuned voltage waveforms for each transducer element to control the amplitude and phase of the pulses, thereby allowing precise control of the acoustic pressure field.

[0340] The control logic for beamforming directs and focuses the transmitted ultrasound front. Control of the transmitted ultrasound front is achieved by adjusting the time delay and amplitude of the excitation pulse applied to each element in the array. The control logic can use a beamforming algorithm that combines the desired transmit focus and steering angle with the geometry of the CMUT array. By calculating appropriate time delays and amplitudes for each element, the control logic ensures that the sound waves emitted by individual elements undergo constructive interference at the desired focus, resulting in a clear, steerable beam.

[0341] In ultrasound, transmit beamforming is a signal processing technique used to control the propagation of sound waves emitted by multiple transducer elements in an array. The core idea behind transmit beamforming is to adjust the phase and amplitude at each transducer element to produce constructive interference at a specific focal point or desired spatiotemporal acoustic profile. The timing delay of each transducer can be calculated based on the expected depth of focus and the velocity of sound in the tissue. For constructive interference to occur at a specific focal point, waves from one or more (e.g., all) transducers should arrive at that point simultaneously. Transducers farther from the focal point can be activated slightly earlier than those closer to the focal point. These delays can be achieved using delay lines in the beamformer circuitry.

[0342] The process of controlling the timing and amplitude of signals to achieve, for example, focusing a beam is not limited to a single point. By dynamically changing these parameters, the focus can be manipulated over time. Furthermore, by optimizing these parameters (such as plane waves or diverging waves), more complex waveforms can be formed for specific imaging tasks.

[0343] Implantable transducer: MEMS ultrasound array Ultrasound arrays are structured to optimize the performance and accuracy of both ultrasound imaging and neuromodulation. The acoustic design of an ultrasound imaging array is defined by center frequency, total and effective channel counts, array shape and geometry, array quantity, element distribution, bandwidth, and angular sensitivity. Alternatively, neuromodulation array design may be based on the focused volume size and distribution varying with position within the acoustic envelope, achievable acoustic outputs (such as mechanical index, spatial and temporal average intensity), and how these outputs vary as a function of the array. Homogeneity and heterogeneity in the brain Computer simulation The acoustic simulations of propagation in both models are generated.

[0344] like FIG. 3BAs shown, the implantable ultrasound transducer may include a MEMS ultrasound array 316 configured to couple to custom circuitry (e.g., circuit board 310 and / or ultrasound ASIC 312). The MEMS ultrasound array 316 may be configured to image the nervous system (e.g., the brain) and may provide neuromodulation to target areas for clinical treatment. In one or more embodiments, the MEMS ultrasound array 316 may include, for example, a capacitive micromechanical ultrasound transducer (CMUT) array or a lead zirconate titanate (PZT) transducer array, which may be directly fabricated on, bonded to, or otherwise electronically connected to a complementary metal-oxide-semiconductor (CMOS) application-specific integrated circuit (ASIC).

[0345] In one or more embodiments, the choice of ultrasonic array technology can vary depending on the intended application to ensure performance, reliability, and scalability. For example, compared to conventional piezoelectric transducers, CMUTs offer several advantages, including higher receiver sensitivity, wider bandwidth, better integration with electronics, lower acoustic impedance, and reduced crosstalk. These benefits, combined with the ability of CMUTs to be monolithically integrated with CMOS and mass-produced, make CMUTs suitable for the systems and methods disclosed herein. In some cases, large-scale commercial MEMS manufacturing facilities can be used to manufacture and / or source the components used in the systems disclosed herein. By leveraging state-of-the-art CMUT technology and integrating it with CMOS, transducer designs can meet the long-term requirements of head-mounted implantable devices.

[0346] In one or more examples, the system-level specifications of the CMUT array can be customized to meet unique imaging requirements in the brain. For example, for the methods and systems disclosed herein, a CMUT with a center frequency of 5 MHz exhibits greater than 100% bandwidth and a tunable range of 2.5 MHz to 7.5 MHz. These characteristics allow users to selectively configure imaging resolution and depth according to application needs. To obtain a sufficient macroscopic field of view, such as near-whole-brain imaging and modulation, a 17 mm aperture and a 150 μm spacing between the elements of the CMUT array can be used, resulting in an element count of 11,660 (the imaging / modulation solution may not require the simultaneous use of all elements). The spacing between the elements of the CMUT array (e.g.) ， 150 μm) can be based on the maximum coverage angle (e.g. ， 90-degree steering capability The requirements are met, and the wavelength of the nominal operating frequency can be 5 MHz. For example, the appropriate spacing between elements in the CMUT array can be estimated using the following calculations: given a) b) where λ is the wavelength; ,in The speed of sound is approximately 1540 m / s; c) For the frequency, in some implementations it may be approximately 5 MHz; therefore, given the information in a), b), and c), It can have a value of approximately 150 μm, which translates to a spacing of 150 μm between elements of the CMUT array at a center frequency of 5 MHz. According to embodiments of this disclosure, the CMUT array can deliver high-resolution images with broad coverage and optimal performance while maintaining compatibility with CMOS manufacturing processes.

[0347] The implantable transducer design disclosed herein can incorporate various power optimization techniques, such as clock gating and power gating. These techniques can selectively shut down unused or lower-performance portions of the implantable transducer (e.g., ...). ， Circuit boards, ASICs, ultrasonic arrays) or reducing the power supplied to these parts can help minimize power consumption.

[0348] Implantable transducer: mechanical assembly Ultrasonic transducers (e.g., implantable transducers) and methods of manufacturing thereof may include a hermetically sealed housing made of a sound-permeable material, such as, but not limited to, polymethyl methacrylate (PMMA), polyether ether ketone (PEEK), polychlorotrifluoroethylene (PCTFE), polytetrafluoroethylene (PTFE), ultra-high molecular weight polyethylene (UHMWPE), polyethylene terephthalate (PET), low-density polyethylene (LDPE), polyether block amide (PEBAX), and / or high-density polyethylene (HDPE). FIG. 4A to FIG. 4C A physical prototype of an implantable ultrasound transducer is depicted. The ultrasound transducer for ultrasound-based BCI can be constructed according to one or more shape factors, such as... FIG. 6A to FIG. 10B As described, this depends on the application of the ultrasonic transducer. For example, FIG. 6A A CAD rendering of a large shape factor ultrasound transducer is provided. For subjects, a large shape factor ultrasound transducer does not necessarily need to be implantable, but may allow clinicians (e.g., ， Surgeons have a high degree of manual control over the positioning ultrasound transducer, such as... FIG. 38 The description.

[0349] In some implementations, the ultrasonic transducer can be an implantable transducer, such as... FIG. 6B As shown. FIG. 6B A CAD rendering of a small shape factor ultrasound transducer is depicted. Small shape factor ultrasound transducers can have a shape factor small enough to allow implantation into the skull of a subject. FIG. 3AThe positioning of the transducer relative to the skull is depicted. The ultrasound transducer operates according to a data transmission protocol, can be connected to a power source, and functions according to thermal management compatible with cranial implants in humans. A development board for the ultrasound transducer (e.g.) ， Printed circuit boards (PCBs) can include multi-part designs (e.g. ， (Two-part design), this multi-part design can accommodate communication rates of approximately, for example, 1 Gb / s, 2 Gb / s, 3 Gb / s, 4 Gb / s, 5 Gb / s, 6 Gb / s, 7 Gb / s, 8 Gb / s, 9 Gb / s, 10 Gb / s, 11 Gb / s, 12 Gb / s, 13 Gb / s, 14 Gb / s, 15 Gb / s, or 16 Gb / s across cables 602 with lengths of approximately 0.5 m, 0.8 m, 1.0 m, 1.5 m, 2.0 m, 2.5 m, 3.0 m.

[0350] FIG. 7 Images are provided depicting various shape factors of ultrasonic transducers, such as the large shape factor ultrasonic transducer 702 (similar to...). FIG. 6A The large shape factor ultrasound transducer shown), and the small shape factor implantable ultrasound transducer 704 (similar to...) FIG. 6B The small shape factor ultrasound transducer and the miniature shape factor implantable transducer 706 are shown. In some aspects, the shape factor of the miniature shape factor implantable transducer 706 may be smaller than that of the small shape factor implantable ultrasound transducer. FIG. 7 A twenty-five coin 708 is also depicted to provide an understanding of the scale relative to ultrasonic transducers 702, 704 and 706.

[0351] The implantable transducer may include a shape factor small enough to allow implantation into the skull of a subject. Therefore, the shape factor of the implantable transducer can be customized to the subject's skull geometry, such that the transducer's dimensions range from approximately 3 mm to 20 mm in thickness and from approximately 14 mm to 50 mm in width and / or length. Optionally, the housing may include a cable capable of transmitting both power and data to and / or from the device to ensure efficient operation. The cable can allow data relay, such as data communication, with communication rates, for example, approximately 1 Gb / s, 2 Gb / s, 3 Gb / s, 4 Gb / s, 5 Gb / s, 6 Gb / s, 7 Gb / s, 8 Gb / s, 9 Gb / s, 10 Gb / s, 11 Gb / s, 12 Gb / s, 13 Gb / s, 14 Gb / s, 15 Gb / s, or 16 Gb / s, across cables with lengths of approximately 0.5 m, 0.8 m, 1.0 m, 1.5 m, 2.0 m, 2.0 m, 3.0 m, 4 Gb / s, 5 Gb / s, 6 Gb / s, 7 Gb / s, 8 Gb / s, 9 Gb / s, 10 Gb / s, 11 Gb / s, 12 Gb / s, 13 Gb / s, 14 Gb / s, or 16 Gb / s.

[0352] Ultrasonic transducers (including) FIG. 6A to FIG. 10B The electronics included in any of the implantable transducers described may comprise at least two development boards that can form a multi-part (e.g., two-part) assembly. More generally, the implantable transducer can be, for example, in a dry gas environment. In the Multiple segments of the housing are securely connected to protect the internal electronics. Alternatively, implantable transducers can be manufactured not by constructing a multi-part assembly of the housing, but by constructing a single segment (e.g., ， Manufactured using a unibody housing), the individual components can be directly encapsulated (e.g., ...). ， The electronic components included in the hermetic seal form a cohesive interface suitable for sterilization. Sterilization methods may include gamma irradiation, autoclaving, ethylene oxide treatment, and / or combinations thereof. Assembly of the ultrasonic transducer (including the formation of the hermetic seal) may be completed at least partially by bonding methods such as laser welding, electron beam welding, TIG welding, thermal welding, epoxy sealing, or combinations thereof. The housing may be made of any of the aforementioned acoustically permeable materials such as PMMA, PEEK, PCTFE, PTFE, UHMWPE, PET, PEBAX, LDPE, and / or HDPE.

[0353] FIG. 8A and FIG. 8B Additional images depict the assembly of a small shape factor ultrasonic transducer. FIG. 8A A cross-sectional rendering of a non-implantable small shape factor ultrasonic transducer is depicted, revealing the positioning of internal electronics (including MEMS ultrasonic assembly 802) within the transducer's housing. FIG. 8B Depicting with FIG. 7 An external rendering of an ultrasonic transducer with a small shape factor. FIG. 8B Port 804 for data cables, such as USB cables or dedicated cables that can operate according to non-standard data communication protocols, is also described. FIG. 8A and FIG. 8B Each describes a sound-permeable material 806 through which ultrasound can pass. Sound-permeable materials may include, for example, PMMA, PEEK, PCTFE, PTFE, UHMWPE, PET, LDPE, PEBAX, and / or HDPE materials.

[0354] FIG. 8C and FIG. 8D A line drawing depicts the exterior of a small shape factor non-implantable ultrasound transducer. FIG. 8C and FIG. 8DThe image depicts an external view of a small shape factor non-implantable ultrasound transducer, which includes a port 808 for a cable that can receive and / or provide bidirectional data and / or data from external devices (such as peripheral controllers, etc.). FIG. 13A to FIG. 13B As shown in the figure, or the computer) to the unidirectional power of the ultrasonic transducer. FIG. 8A to FIG. 8E A version of a small-shape-factor non-implantable ultrasonic transducer manufactured from multiple housing components is described. FIG. 8D An acoustic window 810 is depicted through which ultrasonic waves (e.g., ultrasonic pulses) can pass. The ultrasonic waves can be emitted from a MEMS array located below the acoustic window 810 within the housing of the ultrasonic transducer.

[0355] FIG. 8E Additional renderings of the exterior of a small shape factor non-implantable ultrasound transducer are provided. FIG. 8E An external view of a small shape factor non-implantable ultrasound transducer is depicted, the transducer including a cable 812, which is compatible with... FIG. 8C and FIG. 7 Port 808, shown as D, is electronically and mechanically paired. Cable 812 can receive and / or supply power from the ultrasonic transducer to external devices (such as peripheral controllers, etc.). FIG. 13A to FIG. 13C As shown, or the computer) bidirectional data and / or power. FIG. 8E A version of a small-shape-factor non-implantable ultrasonic transducer manufactured from multiple housing components is described. FIG. 8E An acoustic window 810 is depicted through which ultrasonic waves (e.g., ultrasonic pulses) can pass. The ultrasonic waves can be emitted from a MEMS array positioned adjacent to the acoustic window 810 within the housing of an ultrasonic transducer. The acoustic window 810 may comprise a biocompatible material that allows for minimal acoustic impedance and minimal ultrasonic attenuation, such as polymethyl methacrylate (PMMA), polyetheretherketone (PEEK), polychlorotrifluoroethylene (PCTFE), polytetrafluoroethylene (PTFE), ultra-high molecular weight polyethylene (UHMWPE), polyethylene terephthalate (PET), low-density polyethylene (LDPE), polyether block amide (PEBAX), and / or high-density polyethylene (HDPE).

[0356] FIG. 9A and FIG. 9B An exploded view rendering of a small shape factor non-implantable ultrasound transducer is provided. FIG. 9A to FIG. 9B The descriptions in sections 902-926 do not necessarily describe small shape factor non-implantable ultrasound transducers exclusively, but may also describe ultrasound transducers of any shape factor (including small shape factor and micro shape factor), and the ultrasound transducers may be implantable or non-implantable. FIG. 9A to FIG. 9BThe depicted ultrasound transducer portions 902-926 are applicable to ultrasound transducers configured for imaging and / or modulating the nervous system of a subject. For example... FIG. 9A and FIG. 9B As shown, the ultrasonic transducer may include one or more of the following: an upper housing 902, which may be made of hard anodized (Type III) 6061 aluminum alloy; an FPGA circuit board 904; a heat sink 906 (e.g., made of 6061 aluminum alloy); a lower housing assembly 908, which includes the ultrasonic transducer; a round head screw 910 (e.g., an M2×5 mm round head screw made of 316 stainless steel); an O-ring 912 (e.g., made of FKM rubber); an M2 washer 914 (e.g., made of PTFE or Nylon 6); a round head screw 916 (e.g., an M2×3 mm round head screw made of 316 stainless steel); a screw assembly 918 (e.g., an M3×3 mm set screw made of 316 stainless steel); and a housing seal 920 (e.g., made of an elastomer). FIG. 8B Further depiction includes: a filled circuit board 922 comprising a MEMS array housed within a lower housing assembly 908; a housing assembly 924 of the lower housing assembly 908; and an acoustically transparent window 926 comprising a biocompatible material, such as PMMA, PEEK, PCTFE, PTFE, UHMWPE, PET, LDPE, PEBAX, and / or HDPE, allowing for minimal acoustic impedance and minimal ultrasonic attenuation. The MEMS array included in the filled circuit board 922 may be a capacitive micromechanical ultrasonic transducer (CMUT) array constructed using MEMS technology, or alternatively, a piezoelectric micromechanical ultrasonic transducer (PMUT) array constructed using MEMS technology. The ultrasonic transducer may be any array constructed using MEMS technology, and is not necessarily limited to CMUT or PMUT arrays.

[0357] FIG. 10A and FIG. 10B Renderings of a non-implantable, small shape factor ultrasound transducer version are provided. FIG. 10A Describing relative to FIG. 10B The rendering of the transducer shown is a rendering of the transducer rotated 180 degrees along the rolling axis. The transducer may include a cap-like structure 1002, which can be used to fix (e.g., adhere) the transducer against the top of the subject's scalp. That is, given... FIG. 10A and FIG. 10B The described transducer is not implanted in the subject's body; the cap-like structure 1002 secures the transducer against the subject's position. The transducer may include a cable 1004, for example... ，The transducer may be a USB 2 cable or any sub-version thereof, or a USB 3 cable or any sub-version thereof, or a cable operating based on a standardized or non-standardized data communication protocol. The transducer may include a housing 1006, which may include a MEMS ultrasonic array (e.g., a CMUT or PMUT array) and corresponding electronics housed within the housing.

[0358] FIG. 11 An exemplary workflow for assembling an implantable transducer according to some embodiments is described. The steps of method 1100 may be performed, for example, using one or more electronic devices that implement a software platform and control a robotic device. In some embodiments, at step 1102, an ultrasonic array may be placed near an acoustic window. In some embodiments, at step 1104, a housing may be attached to the placed ultrasonic array, wherein the placed ultrasonic array may be at least partially disposed within the housing.

[0359] FIG. 12 A schematic diagram depicting an ultrasonic transducer device 1202 connected to a personal computer 1204 is provided. The diagram illustrates an implanted portion of a BCI system connected via a cable (e.g., ， (Implantable ultrasound transducer), the cable can handle bidirectional data transmission and unidirectional power delivery from external devices. External devices can range from custom electronics designed to program and power the implant to standard computing devices (like mobile phones, tablets, and computers), any of which enhances accessibility and ease of use.

[0360] Peripheral controller 120 A head-mounted ultrasound neurosensing and stimulation system (e.g., macroscopic BCI system 100) may include multiple head-implanted transducers that are powered, synchronized, and transmit data using a common peripheral controller. In some embodiments, the peripheral controller may be configured to support three head-implanted transducers. In some embodiments, the peripheral controller may be configured to support one to eight controllers. The number of head-implanted transducers supported by the peripheral controller and used in the macroscopic BCI system is not intended to limit the scope of this disclosure or the peripheral controller.

[0361] FIG. 13A to FIG. 13C An example peripheral controller is shown. For example... FIG. 13A As shown, the peripheral controller 1300 may include at least: a battery 1302, a wireless transmitter / receiver 1304, a DSP ASIC 1306, a power management circuit 1308, an antenna 1310 (e.g., for wireless communication), an antenna or coil 1312 for wireless charging, a connector 1314 coupled to at least one tether, and a processing unit 1316 (e.g., a microcontroller or CPU).FIG. 13B and FIG. 13C A 3D rendering of the peripheral controller 1300 is shown.

[0362] In some implementations, the peripheral controller may not be implanted in the subject's body. FIG. 13B An exploded view of an example peripheral controller that can control one or more ultrasound transducers (e.g., non-implantable ultrasound transducers) is provided. FIG. 13B The illustrated non-embedded peripheral controller may include a programmable push button 1318, a housing 1320 (which may also include entry and exit points for cables providing data and / or power), boards 1322 and 1324 (which may include, for example, a printed circuit board (PCB), a battery, a processor, or an FPGA), and a housing base 1326. Where board 1322 or 1324 includes a PCB, the PCB may facilitate onboard digital signal processing (DSP) and may include one or more processing units (e.g., processing unit 1316). Additionally or alternatively, the PCB may facilitate wireless data transmission, for example, by incorporating a wireless transmitter / receiver 1304, and / or, for example, by incorporating power management circuitry 1308.

[0363] In some implementations, the peripheral controller may be implanted in the subject's body. FIG. 13CAn external view of an example implantable peripheral controller is provided. The implantable peripheral controller can be implanted close to the subject's chest cavity, similar to a pacemaker or deep brain stimulator, and is therefore referred to as a chest unit. The implantable peripheral controller may further include a power source (e.g., a battery), a housing that may be made of titanium, and a PCB. The PCB may facilitate onboard digital signal processing (DSP) and may include one or more processing units (e.g., processing unit 1316). Additionally or alternatively, the PCB may facilitate wireless data transmission, for example by incorporating a wireless transmitter / receiver 1304, and / or facilitate power management, for example by incorporating power management circuitry 1308. Furthermore, a wireless charging coil may be used to charge the power source (e.g., the battery) of the implantable peripheral controller (e.g., chest unit) via inductive charging. The peripheral controller (e.g., chest unit) is adaptable to the physical constraints of the subject's chest cavity and can be within approximately 70 mm long × 50 mm wide × 15 mm high. Alternatively, the peripheral controller may be slightly larger, such as having a length of 75 mm, 80 mm, 85 mm, 90 mm, or 100 mm; a width of 55 mm, 60 mm, 65 mm, 70 mm, or 75 mm; and / or a height of 16 mm, 17 mm, 18 mm, or 19 mm. In some respects, the peripheral controller may be much smaller, such as having a length of 68 mm, 66 mm, 64 mm, 62 mm, 60 mm, 58 mm, 56 mm, 54 mm, 52 mm, or 50 mm; a width of 48 mm, 46 mm, 44 mm, 42 mm, 40 mm, 38 mm, 36 mm, 34 mm, 32 mm, or 30 mm; and / or a height of 14 mm, 13 mm, 12 mm, 11 mm, 10 mm, 9 mm, 8 mm, 7 mm, 6 mm, and 5 mm, wherein the listed lengths, widths, and heights may be arranged in any manner to describe the dimensions of the peripheral controller. The peripheral controller may receive a tether attached to one or more sensing devices. In some implementations, the peripheral controller can provide power management and data transmission to the implantable transducer (e.g., configuration data, synchronization, or other data). The peripheral controller can receive data from the implantable transducer and can perform any subsequent data processing or post-processing techniques before sending image data to an external central device. The external central device can be any device that can operate as a standalone receiver, such as a device that can connect to a wireless network, such as, but not limited to, a WiFi network. In some implementations, the external central device can be a server, computer, or mobile device, such as, but not limited to, a mobile phone or tablet. In some implementations, image data can be wirelessly transmitted to the external central device, can be wirelessly connected to the peripheral controller, and can facilitate the provision of a human-machine interface for controlling the entire system. The human-machine interface can be designed for use by users, such as, but not limited to, clinicians (…).That is The human-machine interface (HMI) can be a user interface, such as a clinician interface. The HMI can be configured to view or retrieve data and / or device status, and / or be programmed for treatment plans and / or other functions related to device use.

[0364] In one or more embodiments, the peripheral controller (e.g., a chest unit) may include a biocompatible material configured for safe implantation into a medical device (such as, but not limited to, medical-grade polymers, titanium, or stainless steel) to prevent adverse reactions or complications in the human body. Additionally, the peripheral controller may be configured to comply with established standards for electronics, wireless communications, and electromagnetic compatibility (EMC)—such as ISO 14708 for implantable medical devices. Compliance with these standards ensures that the implant functions correctly under a wide range of conditions without interfering with other medical devices or systems.

[0365] FIG. 14A to FIG. 14B An example implantable peripheral controller 1404 connected to multiple implantable ultrasound transducers 1402 is depicted. The means of connecting the implantable peripheral controller 1404 to one or more implantable ultrasound transducers 1402 may include an implantable device for ultrasound-based imaging and modulation of a subject's nervous system, and may be installed in conjunction with... FIG. 2B The configuration shown is not similar to the configuration shown.

[0366] Peripheral controller: DSP ASIC In one or more embodiments, the peripheral controller may include a custom DSP ASIC for image processing. A custom DSP ASIC allows for a compact and power-efficient system that facilitates the implantation of an ultrasound transducer. However, a custom DSP ASIC is not necessary for ultrasound transducer implantation. It can perform the main image processing operations on the data received from the AFE ASIC before wireless transmission to external devices. Given that these functions may include highly iterative, cross-functional operations characteristic of the applications of the systems disclosed herein, a custom DSP ASIC can be fabricated using custom techniques without outsourcing components of the design.

[0367] The DSP ASIC design process can include optimizing image processing algorithms on an FPGA (Field-Programmable Gate Array) for rapid prototyping and testing. FPGA-based prototyping and testing allow for modification and optimization of the relevant algorithms before they are submitted to the ASIC design. Algorithms can be programmed in hardware description languages ​​(HDLs) such as Verilog. Once the algorithm has been optimized and tested on the FPGA, it can be digitally synthesized into, for example, a physical ASIC.

[0368] To minimize power consumption and ensure efficient operation, the systems disclosed herein may include small process nodes. Smaller process nodes reduce power consumption and increase transistor density, thereby allowing for greater functionality with a compact form factor and generating less heat for a given computational load. The methods and systems disclosed herein may include process nodes of 22 nm or smaller (e.g., 20 nm, 16 nm, 14 nm, 10 nm, 7 nm, 5 nm, 3 nm, 2 nm, or 1 nm).

[0369] The system design disclosed herein incorporates various power optimization techniques, such as clock gating and power gating. These techniques help minimize power consumption by selectively shutting down unused or lower-performance portions of the ASIC or reducing the power supplied to these portions.

[0370] Peripheral controller: power management The power management circuitry ensures that the peripheral controller and head-mounted transducer receive power from the battery and facilitates wireless charging via an inductive charging system. Using an inductive charging system reduces or even eliminates the use of physical connectors, thereby lowering the risk of infection and ensuring a sealed and sterile environment.

[0371] The methods and systems described herein may include inductive charging systems for peripheral controllers. In some embodiments, the operating frequency may be within the Industrial, Scientific, and Medical (ISM) band, such as 13.56 MHz or 6.78 MHz, consistent with predicated inductive charging links in medical devices. This range provides a balance between power transfer efficiency and tissue heating, ensuring safe and efficient operation. In one or more embodiments, the inductive charging system may include multilayer coils or planar coils with a ferrite backing to maximize coupling efficiency while minimizing magnetic field leakage.

[0372] The methods and systems disclosed herein may also include optimal safety and thermal management. In some embodiments, the chest unit includes a closed-loop temperature control system and thermal sensors to limit temperature rise in surrounding tissues, thereby ensuring safe operation and minimizing potential risks to the patient.

[0373] In some embodiments, the battery may include a lithium-ion polymer battery or a nuclear-powered battery. Lithium-ion polymer (LiPo) batteries are generally a suitable battery technology based on criteria of high energy density, long cycle life, and low self-discharge rate. LiPo batteries offer high energy density, long cycle life, low self-discharge rate, and biocompatible versions are available. In some embodiments, LiPo batteries can be molded into custom shapes to fit the shape factor of the implant. The methods and systems disclosed herein may include custom batteries matched to system specifications, enabling maximum capacity to be achieved within a limited volume. The systems described herein may include capacities up to approximately 100 mAh-20 Ah (e.g., approximately 100 mAh, 200 mAh, 500 mAh, 800 mAh, 1000 mAh, 1200 mAh, 1500 mAh, 1800 mAh, 200 mAh, 2200 mAh, 2500 mAh, 2800 mAh, 3000 mAh, 3100 mAh, 3200 mAh, 3300 mAh, 3400 mAh, 3500 mAh, 3600 mAh, 3700 mAh, 3800 mAh, 3900 mAh, 4000 mAh, 5000 mAh, 800 mAh, 1000 mAh, 2000 mAh, 5000 mAh, 8000 mAh, 10000 mAh, 12000 mAh, 15000 mAh, 18000 mAh). A battery (mAh or 20000 mAh). When the peripheral controller is an implantable peripheral controller (e.g., a thoracic unit), the battery capacity can be adjusted based on the available volume in the subject's thoracic cavity and the energy density of the battery technology. In some embodiments, battery technologies with higher energy densities can produce higher capacities, up to approximately 20000 mAh or higher.

[0374] Peripheral controller: wireless communication Peripheral controllers may include wireless transmitters and / or receivers. The choice of wireless communication technology for peripheral controllers (such as implanted chest units) is critical. In some embodiments, ultrasound data can be used to estimate brain function transmitted from an implant and can be transmitted at a speed of approximately 300 Mb / s. In some examples, a B-mode ensemble of 16 2D slice data from an implant can be estimated at approximately 262 Mb / s. A B-mode ensemble can refer to the collection of B-mode images included in functional ultrasound imaging, where B-mode refers to the ability of an ultrasound system (e.g., an implantable transducer) to transmit sequential ultrasound pulses in different directions to form multiple image lines, such that pulse transmission is performed rapidly and repeatedly to generate ultrasound images.

[0375] Given the described parameters, the chest unit can be configured to support up to three implants, and the wireless link can be configured to support approximately 786 Mb / s over a distance of up to 10 meters outside the body. The number of head-implanted transducers supported by the peripheral controller is not intended to limit the scope of this disclosure.

[0376] The systems disclosed herein may include WiFi 6 implemented using chipsets similar to the nRF7002. WiFi 6 can deliver data rates exceeding 1 Gbps and provides enhanced security features such as WPA3 (Wi-Fi Protected Access 3) encryption, and by extension, the disclosed systems may include wireless chipsets that support WPA3. WiFi 6 chipsets may also include energy-efficient features such as Target Wake Time (TWT), i.e., features that allow devices to negotiate the time and frequency at which they can wake up to transmit or receive data. Using TWT and / or other energy-efficient protocols can significantly reduce power consumption by allowing peripheral controllers to remain in low-power sleep states for longer periods without jeopardizing data transmission. Furthermore, advanced features of WiFi 6, such as OFDMA (Orthogonal Frequency Division Multiple Access) and MU-MIMO (Multi-User Multiple-Input Multiple-Output), allow for the efficient handling of multiple simultaneous data streams, which further improves the energy efficiency of the system.

[0377] Communication across multiple ultrasound transducers for macro-scale brain insonation The systems and methods disclosed herein may include multiple implantable ultrasound transducers, enabling coordinated movement of the ultrasound transducers, such as... FIG. 15A to FIG. 15D As depicted. In FIG. 15A The diagram illustrates the use of elevation focusing to generate 2D plane waves. The timing of the transmission along the elevation axis allows for programmable orientation of the 2D plane wave across a 3D volume. FIG. 15A Three planes are shown to indicate programmability. FIG. 15B The schematic diagram illustrates two implanted transducers 1502 emitting 2D plane waves 1504 from various imaginary points within the patient's skull. FIG. 15C The diagram illustrates schematic waveforms to demonstrate ultrasound transmission patterns when the implanted unit images a subject, compared to when the implanted unit performs neuromodulation on the subject. FIG. 15D The schematic diagram illustrates multiple implantable transducers 1502 used to enhance spatial specificity and acoustic intensity at a desired focal location. Multiple ultrasonic transducers 1502 are used as modulators (e.g.,...). ， Part of the neural modulation system can provide better spatial specificity of the focusing region and increased intensity (in the case of three transducers 1502, constructive interference can allow for a nine-fold increase in intensity), for example, as... FIG. 15D As shown.

[0378] Coordinated activity of the ultrasound transducers 1502 can produce optimal system-level functionality, such as imaging distributed brain regions or neuromodulating target neural circuits. Carefully orchestrating the precise activity of multiple implantable ultrasound transducers 1502 can benefit from custom algorithms for optimizing at least one of several parameters, such as, but not limited to, system-level power consumption, imaging field of view, the magnitude of expected neuromodulation, and / or the temporal resolution of the target brain region used for imaging. While the systems described herein are capable of imaging or neuromodulating a subject's physiological activity using only a single ultrasound transducer (albeit at a reduced sampling rate), implementations of the systems and methods described herein may include multiple ultrasound transducers.

[0379] In some implementations, individual head-implanted transducers have limited achievable performance due to heating limitations, power requirements, physical limitations, or other related reasons. Imaging resolution and depth, as well as power delivery per unit area for neuromodulation, can be improved by using methods that synchronize these sensor devices via a peripheral controller, combined with application-specific algorithms. In some implementations, the synchronization method may include a method in which the sensors are configured such that they have overlapping apertures. Thus, each sensor can individually beamform a target within a shared focal range. The systems described herein may include dedicated algorithms for synchronizing sensor devices with peripheral control devices, such as implanted chest units. The synchronization method may use, for example, a common timing reset signal derived from the peripheral controller or other configuration settings that can be propagated to all sensor devices in the system. Synchronized neuromodulation methods may use algorithms to locate the same treatment point across multiple sensor devices after imaging.

[0380] Another consideration when developing specialized algorithms for ultrasound imaging and neuromodulation within implantable shape factors is whether the available power budget, given thermal constraints, is sufficient to provide adequate system performance for high-impact use cases. Therefore, embodiments according to this disclosure are designed to fall within the desired parameters of thermal management and the physical shape factor.

[0381] Ultrasonic neurosensing and stimulation systems (e.g., macroscopic BCI) may also include multiple implanted sensor devices that can be used in combination to improve the overall performance of the system described herein. The integration of additional implanted sensor devices with the components of the system described herein can benefit from custom integration algorithms, where communication between ultrasonic transducers is further coordinated with auxiliary sensor devices.

[0382] The ultrasound neurosensing and stimulation system according to embodiments of this disclosure can be used for both imaging and neuromodulation applications. Such applications can benefit from custom algorithms that coordinate the functional characteristics of ultrasound transducers, such that, for example, some transducers are configured for imaging applications while others are configured for neuromodulation applications. Based on optimal coordination among the ultrasound transducers for a given system-level task, a given ultrasound transducer can switch from being configured for imaging to being configured for neuromodulation, or vice versa. In one or more examples of custom algorithms capable of coordinating the functional characteristics of implantable transducers, prioritizing real-time monitoring with simultaneous coordination or multi-point neuromodulation may include a set of ultrasound transducers coordinated such that some transducers perform imaging while others perform modulation. The algorithm may select the function performed by each transducer based on the distance between the transducers and / or the aperture of each transducer. A peripheral controller or external center can enable real-time updates between the imaging and transmission functions of the ultrasound transducers.

[0383] Imaging and modulation of the nervous system This paper discloses methods and systems for imaging and modulating the nervous system. The disclosed methods and systems overcome the shortcomings of conventional neurotechniques, such as those described above. Imaging and modulating the nervous system can be performed together in a closed-loop manner. For example, imaging and modulating the nervous system can be complementary, allowing for the achievement of target neural activity states through iterative imaging and modulation (e.g., for the treatment of nervous system disorders or diseases).

[0384] For example, ultrasound data (such as ultrasound imaging using an implanted transducer) can be used to observe neural activity in a region of a subject's brain. The ultrasound data can then be analyzed, for example, through a trained machine learning model, to determine a set of modulation parameters (e.g., instructions for modulating the subject's neural activity). In some implementations, the ultrasound data is processed prior to analysis, as described herein. The model determines the modulation parameters such that neural modulation based on the determined parameters will bring the observed region to a target state of neural activity. The modulation parameters can be communicated to a system used to perform the neural modulation.

[0385] Then, the effects of neuromodulation are observed at selected brain regions, for example, using ultrasound imaging. A trained machine learning model can then analyze the difference between the observed neural activity (observed via ultrasound imaging) generated by the modulation and the target neural activity state. This difference can be used to determine updated modulation parameters, such that subsequent neuromodulation based on these new parameters causes the subject's neural activity to converge toward the target neural activity state. When the difference between the observed activity and the target neural activity is less than a threshold (e.g., indicating that the observed activity has sufficiently reached the target neural activity), the modulation parameters no longer need to be updated.

[0386] The alternating sequence of observing a subject's neural activity and subsequently updating the modulation of that activity can be performed in real time. In some embodiments, the sequence of observation and modulation lasts for a period of time, such as based on clinical and biomedical constraints. In some aspects, each iteration of observing / analyzing a subject's neural activity and modulating that activity based on the observations can bring the observed neural activity closer to the target neural activity, thereby allowing for more efficient and accurate treatment of the subject in a minimally invasive and more tailored manner. In some embodiments, the time between successive iterations of observation / analysis and modulation may depend on the condition being treated. The disclosed systems and methods allow for adjustments to this time to more appropriately treat a specific condition while minimizing power consumption and the patient's commitment to treatment time. For example, for conditions such as epilepsy, the time between successive iterations can be shorter to capture sufficient data points to provide adequate instructions for modulation and treatment. For conditions such as depression, the time between successive iterations can be longer to reduce power consumption while allowing sufficient data points to be acquired for determining modulation instructions and treatment. The disclosed systems and methods can bridge time gaps that can complicate fine-tuning of electrophysiological therapies. The ability to synchronize through the disclosed methods and systems can produce more efficient and effective treatments.

[0387] Although examples of methods and systems for imaging and modulating the nervous system have been described relative to the brain, it should be understood that these methods and systems can be performed on other parts of the nervous system. These methods and systems can be applied to other parts of the nervous system (such as non-brain parts of the central nervous system, e.g., ， (Spinal cord) or peripheral nervous system). Although examples of methods and systems for imaging and modulating the nervous system are described relative to ultrasound images, it should be understood that these methods and systems can be performed using other types of ultrasound data, such as radio frequency (RF) data.

[0388] Maladaptive disorders affecting the nervous system (such as brain disorders) are often resistant to treatment or medication. Furthermore, the underlying causes of nervous system disorders may remain unclear. The resistance to treatment and the unclear etiology of nervous system disorders are partly due to the limitations of conventional neurotechniques. For example, they are limited in resolution, scope, and flexibility when used to observe the nervous system.

[0389] This paper discloses methods and systems for imaging and modulating the nervous system of a subject to achieve a target neural activity state for therapeutic purposes, while overcoming the limitations of conventional neurotechniques. The target neural activity state may differ from the neural activity state associated with maladaptive disorders of the nervous system. The methods and systems disclosed herein use ultrasound data (e.g., received via an implantable transducer) to observe physiological states, such as neurophysiological states or neural activity, within the subject's nervous system. Using ultrasound data to observe neural activity in a subject offers several key advantages compared to conventional methods, which will be described in more detail herein.

[0390] In some embodiments, the disclosed methods and systems treat an individual's nervous system by determining instructions for modulating neural activity. The method includes receiving ultrasound data of the nervous system from an implanted transducer. The ultrasound data may indicate a physiological state of the nervous system, such as neurophysiological state or neural activity. The ultrasound data may be transmitted to a controller. Based on the transmitted ultrasound data, instructions for modulating neural activity in the individual's nervous system can be determined. Neuromodulation can then be applied to the individual based on the determined instruction protocol, for example, by sending instructions to a system for performing neuromodulation.

[0391] In some implementations, the disclosed methods and systems observe neural activity within a closed-loop observation-modulation paradigm. That is, based on the observed neural activity (e.g., ultrasound data received via an implanted transducer, which may indicate modulation-induced neural activity), parameters are determined for modulating the subject, such that applying neuromodulation according to the determined parameters elevates the subject from a pre-modulation neural activity state to a target neural activity state. This closed-loop monitoring ensures the sustained effectiveness of the therapeutic intervention.

[0392] Compared to conventional methods and systems, the disclosed methods and systems allow for the observation of physiological states in a less invasive and more tailored manner with higher resolution, range, and flexibility. This disclosure describes the observation of physiological states using ultrasound data (such as ultrasound imaging). The disclosed implantable transducers, systems, and methods allow for the acquisition of physiological states (e.g., ，Ultrasound data (of neural activity) is available, but such data is otherwise difficult to obtain due to the aforementioned limitations. Using ultrasound data allows for the observation of a subject's neural activity and the determination of modulation of the subject's nervous system (e.g., through algorithms or trained models) to treat disorders or diseases. The disclosed system can be configured to comply with FDA guidance and regulations while utilizing the features and advantages disclosed herein.

[0393] Using ultrasound data to observe a subject's neural activity and perform neuromodulation based on the observations offers additional advantages compared to traditional methods that observe and modulate induced neural responses. Compared to existing technologies, ultrasound data (e.g.) ， Ultrasound images allow for a wider field of view of neural activity. The broad imaging scale provided by ultrasound data is advantageous in the context of neural modulation.

[0394] For example, due to the wider field of view, ultrasound data reduces the need to select specific brain regions for observation. Therefore, the broader imaging scale provided by ultrasound data improves the computational efficiency of the pipeline (e.g., the computational efficiency of devices used to determine neuromodulation commands) because the pipeline can focus on determining the modulation parameters since ultrasound data reduces the need to select specific brain regions for observation.

[0395] For example, the broader imaging scale of ultrasound data improves the accuracy of regulatory commands used to achieve a target neural activity state. Some traditional methods of observing the nervous system measure neural activity from smaller areas of the nervous system, and in doing so, there is a risk of not detecting neural activity in other parts of the nervous system. Because of the inability to detect other neural activity, some traditional methods cannot provide data to more accurately determine new stimulation parameters that would bring the subject closer to the target neural activity state. The broader imaging scale provided by ultrasound data can overcome these shortcomings.

[0396] For example, the broader imaging scale of ultrasound data enables sensitivity to different brain states associated with patient symptoms, such as positive or negative emotional states, tremors, or pain. This data can provide input to systems, including machine learning models, to identify relevant brain regions and correlate them with the patient's emotions (and to train machine learning models for determining treatment instructions).

[0397] As mentioned above, ultrasound data using the physiological state of a subject has many advantages. When combined with appropriate neuromodulation techniques, such as the use of more accurately determined information as mentioned above, ultrasound data can allow subjects to achieve target neural activity states more efficiently and accurately. Examples of these neuromodulation techniques are described in more detail in this article.

[0398] For example, ultrasound can also be used to perform neuromodulation based on ultrasound data of a subject's neural activity. Advantageously, the same device (e.g., an ultrasound transducer) used to receive the subject's ultrasound data can be used to perform ultrasound-based neuromodulation. The ability to use the same device for both data reception and modulation of the subject's nervous system improves efficiency. For example, in cases where both receiving ultrasound data and neuromodulation are performed using the same device, additional surgery or cumbersome clinical setups (such as due to the requirement of combining multiple neurotechnical modalities) may not be necessary. Examples of the disclosed methods and systems for modulating neural activity will be described in more detail below.

[0399] FIG. 16A An exemplary method 1600A for imaging and modulating the nervous system of a subject to treat, for example, the nervous system of a human. The steps of method 1600A may be performed, for example, using one or more electronic devices implementing a software platform. In some examples, a client-server system is used to perform method 1600A, and the steps of method 1600A are separated between the server and client devices in any way. In some examples, the steps of method 1600A are separated between the server and multiple client devices. Therefore, although portions of method 1600A are described herein as being performed by a specific device of a client-server system, it should be understood that method 1600A is not limited thereto. In other examples, one or more client devices are used to perform method 1600A. Examples of server and client devices are described in more detail herein.

[0400] In some embodiments, at step 1602A, ultrasound data of the nervous system is received. In some embodiments, the ultrasound data is received from one or more implantable transducers described herein. In some embodiments, the ultrasound data includes one or more ultrasound images. The ultrasound images may be two-dimensional or three-dimensional images.

[0401] In some implementations, one or more ultrasound images that can be received from the implantable transducer described herein have a resolution of 100 μm to 4 mm. This image resolution allows for a sufficient amount of data to determine neuromodulation commands, as described below. This resolution can depend on imaging parameters such as transmission frequency, device aperture, and imaging depth. For example, full-width axial resolution can be described as follows: 1.206 x l x (z / D) ,in l This refers to the wavelength of the ultrasound wave. z For the imaging depth, and D Where is the aperture diameter, and l = c / f ,in c For the speed of sound, and fFor transmission frequency. These relationships indicate that imaging resolution can be optimized based on operating modes. In some implementations, frequencies in the range of 3 MHz to 15 MHz can be used to cover possible operating modes for imaging deep and superficial brain regions, respectively.

[0402] In some implementations, the imaging volume of one or more ultrasound images can be modeled as a spherical sector whose cone radius depends on the spacing or interval between the ultrasound elements (of the ultrasound array of the implanted transducer) relative to the transmission frequency. For example, assuming the spacing is approximately... l If the value is 2, the turning angle will be 45 degrees. The resulting imaging volume is proportional to the imaging depth as follows:

[0403] Table 2. Imaging Depth (cm) and Imaging Volume (cm²) 3 Example correspondence between ) In some implementations, one or more ultrasound images are received at 10 Hz to 257 kHz. For example, the frequency can be determined based on the operating frequency of a power Doppler filter. Alternatively, it can be determined based on the speed of sound in soft tissue (e.g., ， 1540 m / s) and depth associated with the image (e.g. ， The frequency is determined using a method that measures approximately 3 mm of human cortical thickness. In some embodiments, one or more ultrasound images are received at 100 Hz to 25 kHz. For example, one or more ultrasound images are received at 10 kHz. Because ultrasound acquisition can be limited only by the speed of sound and imaging depth, the disclosed methods and systems provide higher temporal resolution for capturing information about brain state and determining neural modulation instructions compared to other methods with more limiting factors.

[0404] In some embodiments, the ultrasound data includes radio frequency (RF) data. In some embodiments, the RF data includes data with frequencies from 300 kHz to 300 MHz. For example, the frequency may be twice the transmit or receive rate (e.g., the Nyquist frequency of the transmit or receive rate). In some embodiments, the functional state of the brain can be determined from the RF data (e.g., raw RF data) captured by the implanted transducer. For example, the implanted transducer emits ultrasound waves into brain tissue, which penetrate the tissue and subsequently reflect back to the implanted transducer. These reflected waves induce electrical signals that can be captured as RF data, which may include depth-related echo information and corresponding amplitude. This RF data can be processed as described with respect to step 1604A to estimate brain activity. In some embodiments, the RF data includes wide-format RF data (e.g., a matrix), vectors of RF data, long-format RF data, or any combination thereof. These formats of RF data may be received from the implanted transducer or are processed RF data from the implanted transducer.

[0405] In some embodiments, ultrasound data can indicate the physiological state of the nervous system. For example, the ultrasound data is received from one or more implantable transducers described herein. In some embodiments, the ultrasound data includes ultrasound data captured by the implantable transducer at different time points. For example, the ultrasound data is part of a video or ultrasound image stream. Additional examples of ultrasound data are described herein.

[0406] In some embodiments, at step 1604A, the ultrasound data is processed. For example, ultrasound data from one or more implantable transducers may be processed at a controller. In some embodiments, processing of the ultrasound data may include 3D images forming or relating to the subject's nervous system (e.g., image stacks or volumes in Cartesian X, Y, and Z dimensions) or ordered sequences of 3D images (e.g., volumetric video). Processing of the ultrasound data may include filtering noise associated with the received ultrasound data, such as, but not limited to, one or more background subtraction steps, applying 2D filters, such as, but not limited to, comparing one or more images with a predetermined kernel (e.g., ...). ，The ultrasound data can be convolved using a Gaussian smoothing kernel, or one or more images can be input into a Kalman filter, Chebyshev filter, Butterworth filter, Bessel filter, Gaussian filter, Caul filter, Legendre filter, Linkowitz-Rayleigh filter, or any combination thereof. Filters can be applied, for example, to multiple images received sequentially across multiple time points. Processing of the ultrasound data at, for example, a controller may include using one or more trained machine learning models. Processing of the ultrasound data at, for example, a controller may include compressing or reducing the size of the received ultrasound data, for example, using one or more compression algorithms, such as lossy compression algorithms (e.g., transform encoding / decoding algorithms, color quantization algorithms, chroma subsampling, or fractal compression) or lossless compression algorithms (e.g., run-length encoding, area image compression, predictive encoding / decoding, entropy encoding, adaptive dictionary algorithms, chain codes, or diffusion models). In some embodiments, the ultrasound data is not processed, but rather transmitted as described below.

[0407] For example, ultrasound data may be processed (e.g., by the system that determines the instructions or by a controller) before the instructions described herein are determined. This processing may include dimensionality reduction, making it possible to use a low-dimensional representation (e.g., a feature vector) as input to a machine learning algorithm.

[0408] Again Ultrasound data can be processed, for example, by convolving it with an imaging kernel (such as an image sharpening kernel or a Gaussian blur kernel) to improve image quality. Ultrasound data includes time series of ultrasound data (e.g.,...). ， In the case of video, background subtraction from a reference image or preprocessing analysis that treats time as a variable can be used to preprocess images in a time series.

[0409] In some implementations, ultrasound data includes radio frequency (RF) data, and one or more ultrasound images are generated based on RF data. For example, processing steps may occur in the analog domain prior to digitization. First, the RF data (e.g., reflected ultrasound signals) may be amplified using a low-noise amplifier. These signals may also be passed through bandpass filters to eliminate unwanted noise or frequencies outside the desired range. Variable gain amplifiers may be used to dynamically adjust the amplification level to compensate for attenuation effects caused by variations in tissue depth. To prevent aliasing artifacts during digital sampling, anti-aliasing filters may be applied to the analog signals. These pre-digitization procedures prepare the analog signals for efficient and accurate conversion to digital format.

[0410] In some implementations, the RF data is digitized. Once digitized, the RF data undergoes further processing to produce an image. For example, the RF signal is converted to a baseband signal via IQ demodulation. The demodulated signal can be low-pass filtered and subsequently downsampled to reduce the amount of data. Beamforming techniques can then be applied. In Delay-Summarized (DAS) beamforming, plane-wave IQ data from multiple transducer elements are time-aligned and combined to focus the ultrasound signal at different points within the imaging field. This time alignment can be determined based on the time it takes for the ultrasound wave to travel from the transducer to a specific focus and back. The aligned IQ data can then be superimposed to produce a single beamformed signal for each focus. This process can be repeated for multiple focuses to generate 2D or 3D images.

[0411] In some implementations, received ultrasound data (e.g., RF data) undergoes clutter filtering to determine the functional state of the brain. In some implementations, clutter filtering is configured to distinguish dynamic changes (such as blood flow) from stationary or slowly moving tissue signals. In some implementations, clutter filtering includes methods such as high-pass filtering or singular value decomposition (SVD), which are configured to separate physiologically relevant signals, for example, within the brain.

[0412] In some embodiments, at step 1606A, processed ultrasound data is transmitted. For example, the processed ultrasound data is transmitted to a controller. In some embodiments, the controller is a standalone device that communicates with one or more implantable transducers. For example, the controller may be a client device, such as an intermediary device, for transmitting processed ultrasound data to a second device (e.g., a device including algorithms or models) for determining instructions for neuromodulation. Alternatively, the controller may be part of a device or system for determining instructions for neuromodulation. In some embodiments, the controller is integrated with one or more implantable transducers.

[0413] The controller can be configured to receive data associated with physiological states and / or with neural activity. Physiological states may include neurophysiological states, and neurophysiological states may include hemodynamic activity. Hemodynamic activity can be indicated by power Doppler intensity (PDI) values. Power Doppler is a technique that uses the amplitude or intensity of a Doppler signal (e.g., ultrasound data from an implanted transducer) to detect moving substances (such as blood flow, e.g., hemodynamic activity). For example, changes in PDI values ​​may be proportional to changes in a subject's hemodynamic activity (such as changes in a subject's cerebral blood volume (CBV) activity (e.g., CBV signal)). Therefore, changes in PDI values ​​from ultrasound data can be used to determine CBV activity. Physiological states may include non-neural activities that represent neural activity. For example, CBV signals can indicate vascular changes associated with neural activity, such that changes in neural activity can be represented by monitoring changes in CBV signals. That is, CBV signals (such as those acquired via ultrasound data) are a function of neural activity signals. Neurophysiological states may include hemodynamic activity, such as CBV signals obtained from ultrasound data, because CBV signals indicate neural activity patterns, such as the firing patterns of neurons in the subject's nervous system.

[0414] In some implementations, ultrasound data or processed ultrasound data is transmitted for storage or training of machine learning models. For example, ultrasound data and information associated with the neural activity state of the ultrasound data are stored or used to train machine learning models (e.g., for determining neural activity states, for determining instructions for neural modulation, as described in more detail herein).

[0415] In some embodiments, at step 1608A, instructions for modulating neural activity in the nervous system are determined based on processed ultrasound data. For example, a second device or system is configured to receive processed ultrasound data and determine instructions for modulating neural activity. For instance, a device including an algorithm or model receives processed ultrasound data (e.g., from one or more implantable transducers, from a controller) and determines instructions based on the image using the algorithm or model. Alternatively, a processor of one or more implantable transducers determines these instructions. In some embodiments, instructions are further determined based on additional data, such as data associated with physiological states and / or data associated with neural activity. Instructions and additional examples of the determination of instructions are described herein.

[0416] Modulation of neural activity may include stimulating one or more regions of the nervous system. The stimulated regions of the nervous system may include one or more regions of the peripheral nervous system, one or more regions of the central nervous system, or a combination thereof. Regions of the peripheral nervous system may include nerve cells associated with the somatic or autonomic systems, such as, but not limited to, cranial nerves (e.g., […]). ， The central nervous system includes the vagus nerve, spinal nerves, and motor neurons. Regions of the central nervous system may include the spinal cord. One or more regions of the central nervous system may include the brain. Stimulation of one or more regions of the nervous system may include electrical stimulation via one or more electrodes, and instructions for modulating neural activity may include electrical stimulation via one or more electrodes. In some embodiments, electrical stimulation is controlled by electromodulation parameters, including amplitude, frequency, pulse width, intensity, waveform, polarity, acoustic pressure, or any combination thereof. Electrical stimulation may include deep brain stimulation (DBS), transcranial magnetic stimulation (TMS), repetitive TMS (rTMS), vagus nerve stimulation (VNS), transcranial direct current stimulation (tDCS), electrocorticography (ECoG), optogenetics, or any combination thereof.

[0417] Examples of stimulation are described here. Deep brain stimulation (DBS) is used for motor disorders (such as tremor associated with Parkinson's disease, essential tremor) and limited non-motor applications (e.g., ，Surgical techniques for obsessive-compulsive disorder (OCD). DBS involves implanting electrodes into specific brain regions to modulate abnormal impulses via electrical signals. Stereoscopic electroencephalography (sEEG) is primarily used to monitor neural activity associated with refractory forms of epilepsy. Electrodes are implanted in the brain to record electrical activity and locate the source of seizures. This information can then be used to plan surgical ablation or resection. A combination of sEEG and DBS may involve implanting both recording and stimulating electrodes. Transcranial magnetic stimulation (TMS) and repetitive transcranial magnetic stimulation (rTMS) are non-invasive techniques that use magnetic fields to stimulate specific brain regions. rTMS is used to treat depression and other mental disorders. Vagus nerve stimulation (VNS) involves implanting a device under the skin that delivers electrical signals to the brain via the vagus nerve and is commonly used to treat epilepsy and depression. Transcranial direct current stimulation (tDCS) is a non-invasive form of brain stimulation that uses electrodes delivered through electrodes on the head. Electrocorticography (ECoG) may involve placing electrodes directly on exposed surfaces of the brain to record electrical activity. sEEG and DBS can be combined to implant both recording and stimulating electrodes. Stereoscopic EEG-responsive neurostimulation (RNS) is a neurostimulation system developed for refractory focal epilepsy. The RNS monitors brain activity and provides stimulation when an abnormal pattern is detected. Structural connectivity imaging for DBS lead placement uses high-resolution imaging to guide DBS lead placement. Due to the advantages of the ultrasound data described herein, the disclosed methods and systems allow for more efficient and accurate manipulation of these stimulation systems (e.g., in response to instructions determined by the methods described herein) in a less invasive manner for the treatment of neurological disorders or diseases.

[0418] In some implementations, these instructions are used to cause the system to perform neuromodulation (and treat neurological disorders or diseases). Examples of systems used to perform neuromodulation (e.g., systems for stimulation, one or more ultrasound transducers) are described herein. In some implementations, the instructions include adjusting electrical modulation parameters, spatial modulation parameters, temporal modulation parameters, or any combination thereof. The parameters can be advantageously personalized based on the subject.

[0419] Electrical modulation parameters may include amplitude, frequency, pulse width, intensity, waveform, polarity, acoustic pressure, or any combination thereof. Amplitude can be the voltage or current level of the electrical pulse. Amplitude can be the intensity of the current, for example, measured in milliamperes (mA). Frequency can be the rate at which the electrical pulse is delivered, for example, measured in hertz (Hz). Frequency can also be the frequency of ultrasound, for example, in the range of 0.2 MHz to 10.0 MHz for neuromodulation. Pulse width can be the duration of each electrical pulse, for example, measured in microseconds (μs). Intensity can be the strength of a magnetic field, for example, expressed as a percentage of the device's maximum output, or relative to the subject's motion threshold. Intensity can be the power of ultrasound, for example, expressed in watts per square centimeter (W / cm²). 2 Measurements are taken in units of _____. The waveform can be the shape of a magnetic pulse, which can be single-phase or biphase. Polarity refers to the direction of current flow, determined by the positions of the anode and cathode electrodes. Anode stimulation typically excites neuronal activity, while cathode stimulation inhibits it. Acoustic pressure refers to the amount of pressure applied by ultrasound.

[0420] Spatial modulation parameters may include electrode configuration, electrode location, electrode size, electrode placement, directionality, coil orientation, coil location, stimulation focus, stimulation bilaterality, montage, focal size, target location, or any combination thereof. Electrode configuration may refer to the selection of which one or more contacts can be adjusted when using an implanted electrode system (which may have multiple contacts). Electrode configuration may involve the arrangement of multiple electrodes that can be used in more advanced or experimental settings. Electrode location may refer to the placement of the electrode within the target brain region. Electrode location may be adjusted during the initial surgical implantation of one or more neuromodulation techniques. Directionality may refer to the manipulation of the direction of the electric field. Coil orientation may refer to the angle of the coil relative to the scalp. Coil orientation can affect the directionality of the induced current. Coil location may refer to the specific brain region being targeted, typically guided by a neuronavigation system. Stimulation focus may refer to a coil designed to provide more focused stimulation in contrast to a coil providing broader stimulation. Stimulation bilaterality may involve simultaneously stimulating both hemispheres (e.g., such that stimulation of one hemisphere begins before stimulation of the other hemisphere ends) or stimulating both hemispheres alternately. Electrode size can refer to the surface area of ​​the electrode, which can affect current density and thus the stimulation effect. Electrode placement can refer to the location of the anode and cathode on the scalp or other body parts, for example, guided by a 10-20 EEG system or other methods. Electrode configuration refers to a specific combination of electrode size and placement for targeting a particular brain region. Focal size can refer to the size of the area on which ultrasound waves are focused, which can affect the specificity of neural modulation. Target location can refer to one or more locations within the brain where ultrasound waves can be focused.

[0421] Timing parameters may include bursts, cycles, ramps, frequency, pulse duration, series duration, inter-series interval, total number of pulses, stimulation pattern, duration, inter-stimulation interval, session frequency, pulse repetition frequency, duty cycle, or any combination thereof. A burst can refer to a system that allows pulses to be delivered in bursts rather than continuously. A burst can also refer to a burst pattern, which may refer to protocols that deliver pulses in groups or bursts, with frequency within a burst and inter-burst interval as additional parameters. A cycle can refer to a system that can be programmed to turn on and off at set intervals. A cycle can also refer to protocols that involve alternating periods of stimulation and periods of no stimulation. A ramp can refer to a gradual increase or decrease in amplitude over a specific time period. A ramp can also refer to a gradual increase or decrease in current amplitude at the beginning or end of a session to minimize discomfort for the subject. A ramp can also refer to a gradual increase or decrease in intensity or acoustic pressure over a specified time period to minimize potential side effects in the subject. Frequency can refer to the rate at which pulses are delivered, for example, measured in Hertz (Hz). Pulse duration can refer to the length of a magnetic pulse. Pulse duration can also refer to the length of time an ultrasound pulse lasts, for example, measured in milliseconds (ms). Speech duration can refer to the length of time within which a series of pulses (i.e., a train) are delivered. Inter-speech interval can refer to the time between individual pulse trains. Total number of pulses can refer to the total number of magnetic pulses delivered during a session. Patterned stimulation can refer to protocols that use more complex stimulation patterns, such as theta burst stimulation (TBS), which may involve bursts of pulses at a specific frequency. Duration can refer to the length of time an electrical current is applied, for example, measured in minutes. Inter-stimulation interval can refer to a protocol in which multiple sessions are performed, and the time between the end of one session and the start of the next can be called the inter-stimulation interval. Session frequency can refer to the frequency at which sessions are performed, for example, once a day or once a week. Pulse repetition frequency can refer to the rate at which pulses are emitted, which can be measured in hertz (Hz). Duty cycle can refer to the proportion of time ultrasound is active within a given time period, expressed as a percentage.

[0422] Instructions for modulating neural activity can be determined based on target neural activity. The device can receive target neural activity to determine instructions for neural modulation. The target neural activity can be compared to the subject's current neural activity to determine, for example, whether a neural modulation instruction should be initiated or whether neural modulation can be stopped. Target neural activity can be determined from: ultrasound data (e.g., received from one or more implanted transducers), fMRI imaging, electrophysiological recordings, structural magnetic resonance imaging scans, diffusion tensor imaging (DTI), computed tomography (CT) scan information, or any combination thereof. For example, the device receives this information to determine the target neural activity.

[0423] Target neural activity can be determined from processed ultrasound data (e.g., a processed version of the ultrasound data received at step 1602A). Target neural activity can be determined based on the output of a transfer learning algorithm. In some cases, target neural activity can be determined based on a transfer learning algorithm (e.g., as described herein) trained on data types associated with various brain observation modalities, such as electrophysiological time-series traces and / or image sequences, such as volumetric image sequences.

[0424] The target neural activity can be represented by one or more terms that advantageously allow for comparison between the observed neural activity (e.g., a processed version of ultrasound data from one or more implanted transducers) and the representation of the target neural activity to determine neural modulation instructions. The target neural activity can be represented as a composite, time-independent state. That is, a composite, time-independent state can be a collapse (e.g., averaging) of multiple snapshots of neural activity across the subject and / or across time. The target neural activity can be represented as multidimensional time-series data. For example, the target neural activity can be represented as an ordered sequence, such as a video or a matrix or array, rather than a single static composite state. The observed neural activity state is considered to have reached the target neural activity state if two multidimensional time series are correlated with each other by statistical techniques used for comparing the time series (such as, but not limited to, cross-correlation matrices, cross-correlation functions, cross-variance matrices, cross-variance functions, ARIMA models, or multivariate trend regression models). The temporal or spatial resolution of the multidimensional time-series data may be equal to or less than that of the ultrasound data. That is, the target neural activity state may have a resolution lower than that of the ultrasound data of the subject's neural activity.

[0425] Instructions for regulating neural activity can be determined based on the output of a machine learning algorithm. The output of the machine learning algorithm can be based on processed ultrasound data (e.g., sent from step 1606A) provided to the machine learning algorithm.

[0426] Machine learning algorithms may include reinforcement learning, Bayesian optimization, generalized linear models, support vector machines, deep neural networks, representation learning, or any combination thereof. Machine learning algorithms may be trained, tested, or validated offline. These offline operations may not be performed in real time. That is, offline may mean that it does not occur immediately after acquiring new data (such as acquiring new ultrasound data). In some aspects, offline training, testing, or validation of machine learning algorithms may include training, testing, or validating the machine learning algorithm independently of the acquisition of new data and some analysis and / or in parallel with the acquisition of new data and some analysis (e.g., such that the acquisition of new data begins before the analysis of existing data stops), thereby strengthening the model used to determine neural modulation instructions through different sources. Machine learning algorithms may also be trained, tested, or validated online. In some embodiments, training improves the machine learning algorithm's ability to classify neural activity states (e.g., healthy brain states, unhealthy brain states) to determine instructions for neural modulation. In some embodiments, the machine learning model is trained using a processed version of ultrasound data received from an implanted transducer. In some implementations, machine learning models are trained by mapping instructions for neuromodulation to brain states in response to neuromodulation (e.g., so that the machine learning model will better understand the nervous system’s response to specific neuromodulation instructions).

[0427] For example, the machine learning model includes a decision engine, which can be a machine learning model configured to interpret ultrasound data to recommend appropriate neuromodulation parameters. These steps can be performed in real time, and the neuromodulation device can be updated accordingly. Furthermore, patient-reported results or real-time biofeedback can be used to fine-tune the recommendations made by the machine learning model. Additionally, the machine learning model can integrate ultrasound data with data from other methods, such as electrophysiological methods (including but not limited to EEG, ECoG, or deep electrode imaging), and / or imaging methods (such as fMRI, fNIRS, or microscopy), to create a more comprehensive map of neural activity, which facilitates more effective neuromodulation.

[0428] In some implementations, the instructions include the region of neural activity being modulated, and said region can be determined based on processed ultrasound data (e.g., by the aforementioned device). For example, the processed ultrasound data may include an ordered sequence of images sorted relative to time, such as a video or ultrasound image stream. The processed ultrasound data may also include an ordered sequence of images sorted relative to space (such as the X, Y, or Z axis in Cartesian space). That is, the processed ultrasound data can be used to generate a volumetric depiction of the subject's nervous system. For example, a series of 2D images along the X and Y axes can be stacked to generate a volume along the Z axis. The resulting volume can compensate for missing images along the Z axis, so that unexpected jumps along the Z axis do not hinder the construction of the ultrasound imaging volume.

[0429] For example, if the location of the DBS electrode implantation has not yet been determined, one or more implantable transducers capture ultrasound data of the brain to determine these locations. For example, one or more implantable transducers capture macroscopic brain network activity patterns across multiple frequencies and brain regions, thereby establishing a comprehensive functional connectomics. For example, to determine the DBS electrode implantation location, the functional data obtained from this process are combined with structural MRI scans and diffusion tensor imaging (DTI)-based structural connectomics, and the precise stereotactic placement of the DBS electrode is guided (e.g., by feeding ultrasound, MRI, and DTI data to a machine learning model).

[0430] In some implementations, models (such as machine learning models) can be used to infer portions of the ultrasound imaging volume. The ultrasound imaging volume can be a volumetric time series (e.g., recorded at 0.5 Hz, 1 Hz, 2 Hz), for example, a video consisting of images corresponding to data in the X, Y, and Z dimensions, such that for one or more ordered time points, there exists a volume comprising an ordered stack of 2D images. Machine learning models can also be used to infer non-ultrasound image portions of ultrasound data, such as RF data.

[0431] The region of neural activity being regulated may include anatomical portions of the nervous system. For example, a region of the nervous system may at least partially include the hippocampus or amygdala. It should be understood that a region of the nervous system does not need to directly correspond to an anatomical region of the nervous system associated with neuroanatomical landmarks. For example, the identified region of the nervous system may overlap across many marked neuroanatomical portions of the nervous system.

[0432] The instructions for modulating neural activity can be further determined based on pre-experimental physiological information (such as physiological information prior to imaging and modulation). This pre-experimental physiological information can be sent to the device to determine the instructions for modulating neural activity.

[0433] Pre-experimental physiological information may include pre-experimental ultrasound information, functional magnetic resonance imaging (fMRI) information, electrophysiological recordings, structural magnetic resonance imaging scans, diffusion tensor imaging (DTI) information, computed tomography (CT) scan information, or any combination thereof. Pre-experimental physiological information can be used to train machine learning models (e.g., machine learning models for determining neural modulation instructions). Additional examples of pre-experimental physiological information will be described below in relation to secondary physiological information.

[0434] During the early use of ultrasound data in the observation-regulation paradigm, sufficient example data can be obtained for accurate prediction of regulation parameters (e.g. ，Training data (including the same data type output from ultrasound imaging itself, e.g.) ， Ultrasound data is limited. Advantageously, pre-trial information allows the model to be more accurate when there is limited training data of available ultrasound images. In such cases, the model (such as a machine learning model) can be trained based on additional data types (which may also include pre-trial ultrasound data), allowing the algorithm to learn to infer from data types it has not yet trained on extensively (such as ultrasound data from the implanted transducer when limited ultrasound data from the implanted transducer is available) to predict the set of parameters used to modulate the subject's nervous system.

[0435] In some implementations, transfer learning-based algorithms can be used so that the lack of ultrasound data, for example, from an implanted transducer, does not necessarily prevent the use of machine learning models to determine regulatory parameters from observed neural activity. During training, the transfer learning model can be tuned so that the weights of the ultrasound image training data are higher than those of the non-ultrasound data. The machine learning model (such as a transfer learning algorithm) may include one or more machine learning techniques, such as, but not limited to, reinforcement learning, Bayesian optimization, generalized linear models, support vector machines, or deep neural networks.

[0436] In some embodiments, instructions for modulating neural activity are communicated to a system for performing neuromodulation. For example, a device that determines the instructions for modulating neural activity communicates the instructions (e.g., directly, or via a second device) to the system for performing neuromodulation. Examples of such neuromodulation systems are disclosed herein. In some embodiments, docking devices, protocols, or both may be used to allow communication with the neuromodulation system. Communication may include, for example, the communication of the neuromodulation instructions determined in step 1608A.

[0437] In some implementations, instructions for modulating neural activity are transmitted to the neuromodulation system via a docking device. In some implementations, the docking device is part of the controller described herein. For example, the docking device is a Universal Translator Module (UTM). In some implementations, the UTM is a hardware component that can act as a translator between an ultrasound interface (e.g., associated with processed ultrasound data, associated with ultrasound data from an implanted transducer) and the neuromodulation system. It can receive raw or processed ultrasound data, convert the data into a format compatible with the neuromodulation system, and transmit modulation instructions to the neuromodulation system in a compatible format.

[0438] For example, the interface device is a gateway server. In some implementations, a central gateway server is configured to store communication protocols associated with different neuromodulation systems. The disclosed system can transmit data to the gateway server, which can then translate the received data and forward appropriate instructions to the neuromodulation systems.

[0439] For example, components of the disclosed system or ASICs (Application-Specific Integrated Circuits) (e.g.) ， Implantable transducers and controllers can be programmed with communication protocols to directly interface with different neuromodulation systems.

[0440] In some implementations, instructions for modulating neural activity are sent to the neuromodulation system via communication protocols. For example, communication protocols include API-level communication. For instance, a set of APIs has been developed that enables the disclosed system to communicate with the neuromodulation system. This API can be configured by the manufacturer of the neuromodulation system to make it compatible with the disclosed system. As another example, communication protocols include IoT communication protocols, such as MQTT, CoAP, or HTTP / HTTPS for real-time data communication. Yet another example is communication protocols that include wireless communication standards (e.g., ， Bluetooth and WiFi enable seamless, wireless connections between devices.

[0441] In some implementations, the communication protocol includes security protocols for protecting the integrity and confidentiality of neural data. For example, the security protocol includes end-to-end encryption. Ultrasound data and neuromodulation instructions may include sensitive data. End-to-end encryption can be implemented on data transmission to protect sensitive data from being received by an unwanted party. As another example, the security protocol includes an authentication protocol. Authentication protocols may include the use of authentication methods to ensure connectivity between authorized devices.

[0442] A method for closed-loop neuromodulation of a subject’s neural activity based on observed (e.g., imaged) neural activity, wherein at least one of the neuromodulation or observation of neural activity is achieved by an ultrasound-based technique (e.g., focused ultrasound for neuromodulation), and may include iteratively imaging and modulating the subject’s neural activity such that the observed neural activity resembles a target neural activity pattern, e.g., a target neural activity state.

[0443] FIG. 16BAn exemplary method 1600B for imaging and modulating the nervous system of a subject to treat, for example, the nervous system of a human. The steps of method 1600B may be performed, for example, using one or more electronic devices implementing a software platform. In some examples, a client-server system is used to perform method 1600B, and the steps of method 1600B are separated between the server and client devices in any way. In some examples, the steps of method 1600B are separated between the server and multiple client devices. Therefore, although portions of method 1600B are described herein as being performed by a specific device of a client-server system, it should be understood that method 1600B is not limited thereto. In other examples, one or more client devices are used to perform method 1600B. Examples of server and client devices are described in more detail herein.

[0444] In some embodiments, at step 1602B, ultrasound data of the nervous system is received from the implantable transducer, wherein the ultrasound data may indicate the physiological state of the nervous system. In some embodiments, at step 1604B, the ultrasound data of the nervous system may be processed. In some embodiments, at step 1606B, the processed ultrasound data may be transmitted, wherein instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data. In some embodiments, at step 1608B, instructions for modulating neural activity may be received by the implantable transducer. In some embodiments, at step 1610B, ultrasound neuromodulation of the subject may be performed via the implantable transducer.

[0445] FIG. 17A and FIG. 17B Describes the states of neural activity (e.g.) ， Two exemplary diagrams of brain activity states. FIG. 17A and FIG. 17B The depicted brain activity states are illustrated in a schematic heatmap, which exemplifies spatial and / or temporal patterns in brain regions 1702 ( FIG. 17A ) and different spatial and / or temporal pattern regions of the brain 1704. FIG. 17A In this context, brain activity indicates a healthy brain state. A health marker or positive valence associated with a healthy brain state can be determined by associating one or more observed brain states with one or more known indicators of a healthy physiological state (such as one or more behaviors). An observed brain state associated with one or more known indicators of a healthy physiological state (e.g., one or more healthy behaviors) can be considered an example of one or more healthy brain states.

[0446] exist FIG. 17BIn this context, brain activity indicates maladaptive brain states. The maladaptive markers or negative valence associated with a maladaptive brain state can be determined by associating one or more observed brain states with one or more known indicators of a maladaptive physiological state (e.g., brain disorders or diseases). Observed brain states associated with one or more known indicators of a maladaptive physiological state (e.g., one or more disease behaviors) can be considered examples of one or more maladaptive brain states.

[0447] FIG. 18 An exemplary schematic diagram depicts an initial brain activity state 1806 and two example new brain activity states in response to neural modulation according to two different instructions (e.g., neural modulation set A 1802 corresponding to the first instruction and neural modulation set B 1804 corresponding to the second instruction). FIG. 18 Examples of brain responses to different neural modulations are presented in 1808 and 1810, and these responses (and associated neural modulations) can be used to train models for determining neural modulation instructions. This data allows for quantitative assessment of how different neural modulation parameters regulate brain activity and subsequently affect emotional states.

[0448] As illustrated in the figure, when comparing a new brain activity state 1808 with an initial brain activity state 1806, the neuromodulation set A 1802 can induce changes in brain activity. When comparing a new brain activity state 1810 with an initial brain activity state 1806, the neuromodulation set B 1804 can induce different changes in brain activity. The methods and systems disclosed herein aim to guide an initial brain activity state toward a desired or target brain activity state. Determining the modulation parameters is important for guiding a subject's brain activity state toward a target brain activity state. This paper describes algorithms that predict sets of neuromodulation parameters that can efficiently guide a subject's brain activity state toward a target brain activity state. Due to the advantages of the ultrasound data described herein, the disclosed methods and systems allow for more efficient and accurate execution of neuromodulation (e.g., in response to instructions determined by the methods described herein) in a less invasive manner for the treatment of neurological disorders or diseases.

[0449] For example, FIG. 18A schematic diagram depicts brain activity states in response to various neuromodulation techniques, such as DBS electrodes or implantable ultrasound transducers. When DBS electrodes or implantable ultrasound transducers are implanted, the ultrasound system can monitor the brain's functional networks' responses to changes in neuromodulation parameters (e.g., via ultrasound data received from one or more implantable transducers). For DBS-based neuromodulation, parameters used for modulation may include pulse amplitude, pulse width, frequency, duty cycle, and precise stereotactic positioning within the monitored brain region. For ultrasound-based neuromodulation, parameters used to adjust ultrasound-based neuromodulation may include parameters such as intensity, frequency, acoustic pressure, focal size, target location, pulse duration, pulse repetition frequency, duty cycle, burst pattern, cycle, and ramp. Intensity can refer to the power of the ultrasound waves, which can be expressed in watts per square centimeter (W / cm²). 2 The following parameters are measured in units: Frequency (MHz-10.0 MHz) and Acoustic Pressure (APP). Frequency refers to the frequency of ultrasound waves, which, for ultrasound-based neuromodulation of a subject, can range from 0.2 MHz to 10.0 MHz. AAP refers to the amount of pressure applied by ultrasound waves. Focus size refers to the size of the area on which ultrasound waves are focused, which can affect the specificity of neuromodulation. Target location refers to one or more locations within the brain on which ultrasound waves are focused. Pulse duration refers to the length of time each ultrasound pulse lasts, and can be measured in milliseconds (ms). Pulse repetition frequency (PRF) refers to the rate at which pulses are emitted, and can be measured in hertz (Hz). Duty cycle refers to the proportion of time ultrasound is active within a given period, and can be expressed as a percentage. Burst pattern refers to the protocol for delivering ultrasound waves, which can be in the form of groups or "bursts," with intra-burst frequency and inter-burst interval as additional parameters. Cycle refers to the interval between periods after ultrasound is applied. Ramp refers to an increase or decrease in intensity or acoustic pressure over a specified period to minimize potential side effects. The resulting changes in brain network activity can form a dataset. This data allows for quantitative assessment of how different DBS parameters regulate brain activity and subsequently affect emotional states.

[0450] In some implementations, parameters are determined based on a library of previously effective parameters. In other implementations, parameters are determined as part of an algorithmic approach designed to efficiently identify optimal parameters within a large search space (e.g., Bayesian optimization, the machine learning algorithms disclosed herein). FIG. 26 (As described). These parameters are communicated to a neuromodulation system (e.g., a programmable DBS system). Given the possible time-varying and nonlinear relationships between DBS parameters, brain network activity, and individual variability, this process may include iterative fine-tuning to ensure that the identified neuromodulation parameters are uniquely tailored to optimize patient outcomes.

[0451] FIG. 19 An exemplary schematic diagram is depicted illustrating the adjustment of neuromodulation parameters to achieve a desired brain state. For example, system 1902 (which may include the systems disclosed herein) analyzes a brain state 1904 associated with neural activity (e.g., a current brain state determined by ultrasound data received from an implanted transducer) and a target brain state 1906 associated with the target neural activity (e.g., determined as described herein). Based on brain state 1904 and target brain state 1906, system 1902 (e.g., an algorithm of the system) determines neuromodulation parameters (e.g., stimulation parameters, instructions for neuromodulation) and applies a neuromodulation scheme according to the neuromodulation parameters. Then, as described herein, a neuromodulation-induced brain activity state 1908 is measured using ultrasound data. System 1902 then analyzes the neuromodulation-induced brain activity state in feedback to determine the difference between the neuromodulation-induced brain activity state 1908 and the target brain state 1906. Based on the difference, an adjusted set of neuromodulation parameters is determined such that the subject's observed brain state better achieves the subject's target brain state 1906.

[0452] Closed-loop neuromodulation tool Closed-loop neuromodulation tools may include: real-time monitoring, which enhances the ability to monitor brain responses in real time during neuromodulation. The monitored brain responses can be achieved by acquiring functional ultrasound data continuously or intermittently. Closed-loop neuromodulation tools may include: feedback control algorithms that adjust neuromodulation parameters in real time based on the monitored brain responses, with the goal of maintaining or inducing certain brain states; and / or adaptive neuromodulation tools, which may include tools to support adaptive neuromodulation, wherein neuromodulation parameters are dynamically adjusted across sessions based on the subject's response history.

[0453] FIG. 20 and FIG. 21Flowcharts for processes 2000 and 2100 are provided, which describe the connection points enabling closed-loop neuromodulation tools. For example, step 2002 or 2102, where the initial stimulus is optionally identified based on a previous patient pool, can be performed by a software tool. Furthermore, for step 2004 or 2104, where the initial stimulus is optionally identified based on alternative imaging techniques, the initial stimulus can be identified by a software tool. Additionally, for step 2006 or 2106, where the initial stimulus is optionally identified based on an open-loop stimulation protocol, the initial stimulus can be identified by a software tool. For step 2008 or 2108, the software tool can initialize neuromodulation parameters. For step 2010 or 2110, the neuromodulation system can be activated by a software tool. For step 2012 or 2110, the behavior logging system can be activated by a software tool. For step 2014, the imaging system can be activated by a software tool. For step 2016 or 2114, focused ultrasound can be used to monitor neural activity (such as brain activity), and the monitoring can be coordinated and logged by a software tool. For step 2018 or 2112, the patient's state can be recorded using software tools that can stream and / or store the patient's state. For step 2020 or 2116, stimulation can be applied using software tools. For steps 2022 and / or 2118, data can be integrated and / or analyzed using software tools. The software tools used for data integration and / or analysis can be performed offline, for example on an external server, such as a third-party server, such as a server based on Amazon Web Services or Microsoft Azure. Analysis of the neuroimaging data can inform the optimization of parameters used for neuromodulation of the subject (e.g., steps 2024, 2120, or 2122), and this optimization can be performed using software tools. Neuromodulation can be ultrasound-based, but is not limited to, for example, neuromodulation can be electrophysiological. For step 2026 or 2124, updated neuromodulation parameters (e.g., parameters optimized based on the subject's neural recordings) can be used to provide the subject with updated neuromodulation parameters. ， Stimulation parameters).

[0454] It should be understood that, regarding FIG. 20 and / or FIG. 21 The steps described are exemplary. Methods 2000 and / or 2100 may include fewer steps, additional steps, or a different order of steps compared to those described. It should be understood that the steps of methods 2000 and / or 2100 utilize the features and advantages described above with reference to the other accompanying drawings.

[0455] In some implementations, methods 2000 and / or 2100 are part of a neurological treatment. For example, modulation of neural activity caused by instructions determined by methods 2000 and / or 2100 is for neurological treatment. Treatment may include, without limitation, treatment of chronic pain, depression and anxiety disorders, obsessive-compulsive disorder, Parkinson's disease, essential tremor, epilepsy, post-traumatic stress disorder, memory disorders, or any combination thereof. Obsessive-compulsive disorder may include obsessive-compulsive disorder, substance abuse disorder, or both.

[0456] Closed-loop ultrasound-based neuroimaging and ultrasound-based neuromodulation In some examples, a closed-loop approach to modulating and imaging the nervous system, including the subject's brain, to achieve a target state of neural activity may include: imaging neural activity from the nervous system via ultrasound-based neuroimaging (e.g., via an ultrasound transducer); and modulating neural activity from the nervous system via ultrasound-based neuromodulation (e.g., via an ultrasound transducer). That is, iterative ultrasound imaging and ultrasound neuromodulation protocols (e.g., an "ultrasound-ultrasound" protocol) may be used on the subject. For example, an ultrasound-ultrasound protocol may include an ultrasound transducer that can be acutely placed extradurally within the skull, just outside the dura mater (e.g., as at least...). FIG. 2A , FIG. 2B and FIG. 24 The described implantable ultrasound transducer, or through a similar method FIG. 10A and FIG. 10B The depicted rendering shows an ultrasound transducer placed non-invasively on the scalp. The flexibility of pure ultrasound systems (e.g., imaging and modulation systems without electrophysiological devices) can include the temporary deployment of ultrasound transducers across a wide timescale ranging from minutes to 30 days, thus meeting a diverse range of diagnostic and therapeutic needs. The implantable portion of the system can be connected via a cable that handles bidirectional data transmission and unidirectional power delivery from external devices. External devices can vary from custom electronics specifically designed to program and power the implant to standard computing devices such as mobile phones, tablets, and computers, enhancing accessibility and ease of use. Systems that exclusively include ultrasound-based technologies for both imaging and modulation of the nervous system can serve a wide range of neurological functions, from monitoring and diagnostic purposes to therapeutic interventions across various neurological disorders and injuries.

[0457] Algorithm for imaging with multiple ultrasound transducers The methods and systems described herein address the development of advanced imaging algorithms for ultrasound-based medical devices with multiple transducers. These algorithms optimize the performance and accuracy of ultrasound imaging and neuromodulation arrays through careful design of array characteristics, including center frequency, channel count, array geometry, bandwidth, and angular sensitivity. In one or more embodiments, these algorithms can be implemented in a custom digital signal processor (DSP) chip, a field-programmable gate array (FPGA), or a peripheral central device. The methods and systems described herein utilize the synthesis of array parameters with novel imaging sequences and a comprehensive range of simulation results, while also integrating effects such as skull geometry, phase correction, transducer placement optimization, acoustic intensity, and estimation of effective treatment volume. Furthermore, the approaches and systems described herein employ different imaging strategies, including plane wave, focused wave, and divergent wave approaches, while considering hardware design constraints and power requirements. The described methods and systems may include 'transmit-receive' algorithms that allow transmission on multiple ultrasound transducers while receiving from other transducers, which may be overlapping or separate. In some implementations, the described methods and systems may include standard imaging algorithms, such as delay-stacking algorithms, where a delay can be superimposed on the transducer of the implantable unit, such that signals from a particular direction or the implantable unit are aligned before superposition. Results from non-limiting exemplary simulations of ultrasound imaging sequences further support the feasibility of such methods as described herein, which can cover the entire brain volume using a small implantable ultrasound device suitable for burr hole surgery. The methods and systems described herein employ the steering of the array and ultrasound beams to large angles in both azimuth and elevation. In doing so, the ultrasound-based methods and systems disclosed herein achieve broad coverage of the target biological regions of the subjects used in the study.

[0458] Analysis of ultrasound imaging data from ultrasound transducers This document discloses methods and systems for analyzing ultrasound imaging data from ultrasound transducers, including implantable ultrasound transducers. The imaging data being analyzed may include, for example, raw radio frequency data, raw in-phase and quadrature (IQ) data, beamformed IQ data, and intensity values ​​based on beamformed IQ data. The imaging data may include anatomical data (e.g.,...). ， Non-functional ultrasound imaging data (e.g.) ， B-mode imaging data) and / or functional ultrasound imaging data (e.g. ， Power Doppler imaging data).

[0459] Processing raw radiofrequency data from functional ultrasound imaging can involve any of several operations that transform the data, and any of these operations can be performed in software and / or as part of an algorithm. For example, raw radiofrequency data generated by an ultrasound transducer can be converted into in-phase and quadrature (IQ) data, which can then be converted into beamformed IQ data, which can be converted into a brightness mode (B-mode) image. This brightness mode image can be used to determine the amplitude of power Doppler data, which can be equivalent to an intensity value representing changes in cerebral blood volume, and which can be correlated with and visualized by neural activity data. These data transformations are also related to the formation of composite images. That is, in plane wave imaging (such as ultrasound imaging), multiple images are formed by sending acoustic energy into a medium at different angles. To form a composite image, backscattered echoes from the transmitted acoustic energy are received, and then the backscattered echoes are beamformed separately and superimposed. The amplitude of the images can then be calculated and logarithmically compressed to form a B-mode image. The composite images can then be filtered to extract power Doppler images, thus distinguishing motion data from noise, such as motion data originating from non-blood motion (e.g., tissue motion), from blood motion, which can represent changes in cerebral blood volume and, by extension, neural activity.

[0460] Algorithms (including algorithms implemented on a computer, such as software) can be used to perform finer-grained processing steps, such as any of the processing steps involved in transforming a beamformed IQ image into a B-mode image. Given that the B-mode image is derived from the complex values ​​of the beamformed IQ image, software for analyzing ultrasound data, as discussed herein, can participate in any of the three general operations involved in converting an IQ image to a B-mode image: a) beamforming; b) amplitude determination; and / or c) brightness mapping. During beamforming of the IQ image, the received ultrasound echoes can be aligned and superimposed to form a complex IQ image, which may include amplitude and phase information from the ultrasound waves. During the amplitude determination process, the IQ image may include complex values ​​representing the IQ components of the signal. The amplitude of the complex values ​​can be determined using the following formula: Here, I represents the in-phase component of the signal, and Q represents the quadrature component. During brightness mapping, the calculated amplitude values ​​are then mapped to grayscale intensities to form a B-mode image. When visualizing the B-mode image, higher amplitude values ​​correspond to brighter pixels, representing stronger ultrasonic echoes. In contrast, lower amplitude values ​​correspond to darker pixels, representing weaker ultrasonic echoes.

[0461] Artificial Neural Network Reconstruction of Functional Ultrasound Images Existing methods for functional ultrasound imaging rely on singular value decomposition (SVD) to transform a series of B-mode or IQ images into estimates of blood flow. While effective, this technique demands significant computational resources due to the need to acquire and process large amounts of data (typically around 200 to 400 images). Such demands inherently limit the temporal resolution of the imaging system and increase its power consumption.

[0462] In contrast, by training a model to simulate the output generated by the SVD algorithm, the modern deep learning approach indicated in this paper can potentially reduce the number of required B-mode images by up to 95%. Furthermore, the actual amount of interest can extend beyond static maps representing the output of SVD and can instead include dynamic changes in vascular maps over time, which are crucial for understanding brain function.

[0463] The methods described herein may include artificial neural networks (ANNs) that significantly improve the use of deep learning in functional ultrasound by modifying the ANN's cost function to prioritize sensitivity to dynamic changes in brain function. Modifying the cost function can be achieved by incorporating behavioral relevance metrics that quantify the relationship between neural stimulation and corresponding vascular responses. These metrics guide the model to detect subtle, functionally relevant fluctuations within the cerebrovascular system that indicate neural activity. Furthermore, the cost function may include terms that optimize the predictability of functional images to simultaneously acquired functional data.

[0464] Functional ultrasound data for training the ANN can be captured during a controlled behavioral task known to elicit a specific neural response, ensuring that vascular changes are both predictable and relevant. The ANN model can then be trained using a modified cost function that may include novel terms reflecting the statistical correlation between the measured vascular response and the behavioral stimulus. Training the ANN model can be optimized to prioritize small but functionally significant changes in pixel intensity over larger but less informative changes, ensuring the model is fine-tuned to detect functional variations.

[0465] Using ANNs to reconstruct functional ultrasound images (such as power Doppler images) offers several advantages. For example, ANNs provide improved functional sensitivity compared to conventional methods (such as SVD-based power Doppler image reconstruction). By focusing on detecting small-scale changes within the vascular map corresponding to neural activity, ANN models offer unparalleled sensitivity in functional imaging. Furthermore, ANNs provide improved data utilization efficiency compared to conventional SVD-based methods. ANNs can reduce the amount of data required to reconstruct functional ultrasound images (e.g., power Doppler images). ， The image required (e.g., power Doppler image) ，The number of B-mode images (or composite images) is reduced, thereby decreasing computational load and enabling higher frame rates for real-time imaging applications. Furthermore, compared to conventional methods such as SVD, the power consumption for reconstructing power Doppler images is correspondingly reduced by decreasing the computational load used for reconstructing power Doppler images based on B-mode images. The use of ANNs also allows for the direct integration of subject behavioral data during ANN training or deployment. Direct integration of behavioral data into the imaging process allows for more precise mapping of regions of relevant neural activity in the subject. The improved efficiency in data requirements and power consumption translates into direct clinical benefits. Clinically, the use of ANNs promises to provide real-time monitoring of treatment, offering immediate feedback on treatment efficacy and improving diagnostic accuracy for psychosis.

[0466] Figure 23A An exemplary method 2300A for training a machine learning model for reconstructing one or more functional ultrasound images is described. The steps of method 2300A may be performed, for example, using one or more electronic devices implementing a software platform. In some examples, a client-server system is used to perform method 2300A, and the steps of method 2300A are separated between the server and client devices in any way. In some examples, the steps of method 2300A are separated between the server and multiple client devices. Therefore, although portions of method 2300A are described herein as being performed by a specific device of a client-server system, it should be understood that method 2300A is not limited thereto. In other examples, one or more client devices are used to perform method 2300A. Examples of server and client devices are described in more detail herein.

[0467] In some embodiments, at step 2302A, one or more ultrasound data from one or more samples of one or more subjects obtained from an implantable transducer, and one or more functional ultrasound images corresponding to the one or more ultrasound data are received. In some embodiments, at step 2304A, the one or more ultrasound data are converted into one or more ultrasound arrays. In some embodiments, at step 2306A, one or more functional ultrasound image data are converted into one or more functional ultrasound arrays. In some embodiments, at step 2308A, a machine learning model is trained using one or more ultrasound arrays and one or more functional ultrasound arrays to predict one or more inferred functional ultrasound arrays from the input one or more ultrasound data or the input one or more ultrasound arrays. The ultrasound array or functional ultrasound array may be an ND ( nThe ultrasound data can be an array (or a matrix or tensor of arbitrary dimensions). Converting ultrasound data to an ultrasound array may include multiplying each element of the ultrasound data by 1, which may (but does not necessarily) change the computational object type of the ultrasound data if the ultrasound data is computationally instantiated. Similarly, converting functional ultrasound image data to a functional ultrasound array may include multiplying each element of the functional ultrasound image data by 1, which may (but does not necessarily) change the computational object type of the ultrasound data if the functional ultrasound image is computationally instantiated. The ultrasound data or functional ultrasound data may have already been formatted as an array prior to the conversion, in which case the conversion may include any operations that maintain the array format of the ultrasound data or functional ultrasound data.

[0468] Figure 23B An exemplary method 2300B for reconstructing functional ultrasound images from anatomical ultrasound images is described. The steps of method 2300B may be performed, for example, using one or more electronic devices implementing a software platform. In some examples, a client-server system is used to perform method 2300B, and the steps of method 2300B are separated between the server and client devices in any way. In some examples, the steps of method 2300B are separated between the server and multiple client devices. Therefore, although portions of method 2300B are described herein as being performed by a specific device of a client-server system, it should be understood that method 2300B is not limited thereto. In other examples, one or more client devices are used to perform method 2300B. Examples of server and client devices are described in more detail herein.

[0469] In some embodiments, at step 2302B, one or more ultrasound data from one or more samples are received from one or more subjects. In some embodiments, at step 2304B, the one or more ultrasound data are converted into an ultrasound array. In some embodiments, at step 2306B, the one or more ultrasound arrays can be provided to a trained machine learning model. In some embodiments, at step 2308B, one or more inferred functional ultrasound arrays can be output based on the received one or more ultrasound data. The ultrasound array or functional ultrasound array can be an ND (ND) nThe ultrasound data can be an array (or a matrix or tensor of arbitrary dimensions). Converting ultrasound data to an ultrasound array may include multiplying each element of the ultrasound data by 1, which may (but does not necessarily) change the computational object type of the ultrasound data if the ultrasound data is computationally instantiated. Similarly, converting functional ultrasound image data to a functional ultrasound array may include multiplying each element of the functional ultrasound image data by 1, which may (but does not necessarily) change the computational object type of the ultrasound data if the functional ultrasound image is computationally instantiated. The ultrasound data or functional ultrasound data may have already been formatted as an array prior to the conversion, in which case the conversion may include any operations that maintain the array format of the ultrasound data or functional ultrasound data.

[0470] In implementing the disclosed methods, any of a variety of machine learning approaches and algorithms (wherein, as mentioned herein, machine learning models include trained machine learning algorithms) can be used, such as an ANN configured to reconstruct one or more functional ultrasound images from anatomical ultrasound images. For example, the machine learning model may include a supervised learning model ( Right now Models trained using labeled training datasets, and unsupervised learning models. Right now Models trained using unlabeled training datasets, and semi-supervised learning models. Right now A machine learning model can be a model trained using a combination of labeled and unlabeled training data, a self-supervised learning model, or any combination thereof. In some examples, the machine learning model may include a deep learning model (…). Right now This includes models with many coupled "node" layers, which can be trained in a supervised, unsupervised, or semi-supervised manner.

[0471] In some cases, the disclosed methods may be implemented using one or more machine learning models (e.g., 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, or more than 10 machine learning models) or combinations thereof. In some cases, one or more machine learning models may include statistical methods for analyzing data. Machine learning models may be used for data classification and / or regression. Machine learning models may include, for example, neural networks, support vector machines, decision trees, collective learning (e.g., bag-based learning (such as random forests) and / or boosting-based learning), k Nearest neighbor algorithms, linear regression-based models, and / or logistic regression-based models. Machine learning models may include regularization, such as L1 and / or L2 regularization. Machine learning models may include the use of dimensionality reduction techniques (e.g., principal component analysis, matrix factorization, and / or autoencoders) and / or clustering techniques (e.g., hierarchical clustering, ...). kMean clustering, distribution-based clustering (such as Gaussian mixture models), or density-based clustering (such as DBSCAN or OPTICS) are all possible clustering methods. One or more machine learning models can solve (e.g., optimize) an objective function through multiple iterations based on a training dataset. Iterative solutions can be used even when the machine learning models include those with closed-form solutions (e.g., linear regression).

[0472] In some cases, the machine learning model may include an artificial neural network (ANN), such as a deep learning model. For example, one or more machine learning models / algorithms for implementing the disclosed methods may include an ANN, which may include any of a variety of computational motifs / architectures known to those skilled in the art, including but not limited to feedforward connections (e.g., skip connections), recurrent connections, fully connected layers, convolutional layers, and / or pooling functions (e.g., attention, including self-attention). The artificial neural network may include a differentiable nonlinear function trained via backpropagation.

[0473] Artificial neural networks (e.g., deep learning models) typically consist of a set of interconnected nodes organized into multiple node layers. For example, an ANN architecture may include at least an input layer, one or more hidden layers, and so on. Right now An ANN or deep learning model may include any total number of layers (e.g., 3, 4, 5, 6, 7, 8, 9, 10, 15, 20 or more) and any number of hidden layers (e.g., 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20 or more), where the hidden layers act as trainable feature extractors that allow mapping the input dataset to a preferred output value or a set of output values. Each layer of a neural network includes multiple nodes (e.g., at least 10, 25, 50, 75, 100, 200, 300, 400, 500, 600, 700, 800, 900, 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, or more than 10000 nodes). Nodes receive input data (e.g., genomic feature data such as variant sequence data, methylation status data). wait The system can process non-genomic feature data (e.g., digital pathology image feature data) or other types of input data (e.g., patient-specific clinical data) and perform specific operations (e.g., summation). The input data can come directly from one or more input data nodes, or from the output of one or more nodes in a previous layer. In some cases, the connection from the input to the node is associated with weights (or weighting factors). In some cases, a node can, for example, be linked to all pairs of inputs.X i and its associated weights W i The sum of the products. In some cases, the weighted sum is calculated using the deviation. b Offset. In some cases, a threshold or activation function can be used. f To gate the output of the node, where f It can be a linear or nonlinear function. The activation function can be, for example, the rectified linear unit (ReLU) activation function or other functions, such as saturated hyperbolic tangent, identity, binary step, logic, arctangent, softsign, parameterized modified linear unit, exponential linear unit, softPlus, bending identity, softExponential, sine curve, sine, Gaussian or sigmoid function or any combination thereof.

[0474] Weighting factors, bias values, thresholds, or other computed parameters of a neural network (or other machine learning architecture) can be "taught" or "instructed" during the training phase using one or more training datasets (e.g., 1, 2, 3, 4, 5, or more than 5 training datasets) and a specified training path configured to solve (e.g., minimize) the loss function. For example, an iterative solver (such as gradient-based methods, e.g., backpropagation) can be used to determine adjustable parameters of an ANN (e.g., a deep learning model) based on input data from the training dataset, such that the output values ​​computed by the ANN (e.g., sample classification or disease outcome prediction) are consistent with the examples included in the training dataset. Training the model ( Right now The determination of adjustable parameters of the model using an iterative solver can be performed using or without the same hardware used to deploy the trained model.

[0475] In some cases, the disclosed methods may include retraining any machine learning model (e.g., iteratively retraining the model using one or more training datasets different from those initially used to train the previously trained model). In some cases, retraining a machine learning model may include using a continuous (e.g., online) machine learning model. Right now The model is periodically or continuously updated or retrained based on new training data. This new training data can be provided by, for example, a single deployed local operating system, multiple deployed local operating systems, or multiple deployed, geographically distributed operating systems. In some cases, the disclosed method may employ, for example, a pre-trained ANN, and the pre-trained ANN can be fine-tuned based on additional datasets input into it.

[0476] BCI software The embodiments disclosed herein include a comprehensive software framework designed to support macroscale BCI devices. This software framework, combined with the device, provides an integrated framework for data acquisition, device control, stimulus presentation, and standardized format data storage. The standardized software framework ensures the deployment of a variety of clinical and research protocols for patients, thereby accelerating scientific discovery and the development of novel clinical applications.

[0477] In one or more embodiments, the components of this system may include: Equipment control and configuration The software framework enables users to control and configure BCI devices to meet specific clinical or research objectives. This can include customizing device settings such as sampling rate, resolution, and anatomical targets, as well as defining and implementing specific stimulation paradigms or experimental protocols (if applicable).

[0478] Stimulus presentation and interaction The system can be integrated with clients to present visual, auditory, or other stimuli in response to BCI data or as part of a predefined experimental protocol. This functionality may involve: developing APIs or plugins to interface with popular stimulus presentation software or custom-built display clients; or designing real-time data processing pipelines to extract relevant features or patterns from BCI data that can be used to trigger or modulate stimulus presentation.

[0479] Data collection and management The software framework can acquire brain data and associated metadata from BCI devices in real time during operation. The data acquisition process may involve: implementing communication protocols for interfacing with BCI devices and metadata sources, such as Bluetooth, USB, Wi-Fi, or DAQ; developing efficient buffering and streaming mechanisms to process large amounts of continuous, high-resolution brain data and metadata with minimal latency; and / or implementing synchronization mechanisms to ensure accurate timing between brain data and metadata.

[0480] Data storage and standardization The software framework can standardize the format for storing collected brain data, metadata, and related information to facilitate retrieval, sharing, and analysis. Data storage processes may include: defining standardized file formats, data structures, and naming conventions to ensure data consistency and compatibility across different users, experiments, or analysis tools; developing data export and transformation tools to facilitate data sharing with external platforms, databases, or data analysis software; and / or implementing data upload and archiving mechanisms to ensure data integrity and availability.

[0481] User Interface and Workflow The system can be equipped with a user-friendly graphical user interface (GUI) to enable researchers and clinicians to easily interact with the BCI device, configure settings, monitor data acquisition, and control stimulus presentation. The user interface provides intuitive controls, menus, and visualizations for device configuration, data monitoring, and experimental setup.

[0482] Safety and Compliance The software framework complies with relevant data privacy, security, and regulatory requirements to ensure the security of sensitive patient information and adherence to ethical guidelines. This compliance may include: ensuring the software framework meets relevant industry standards and regulations (such as HIPAA or GDPR); incorporating mechanisms for data anonymization or de-identification to protect personal health information; and / or implementing data encryption, access control, and user authentication mechanisms to safeguard data storage, transmission, and processing.

[0483] Standardized software frameworks may also include other analytical tools, such as, but not limited to: Preprocessing tools These tools may include: motion correction algorithms for adjusting images to compensate for patient movement during data acquisition; temporal filtering for applying frequency-based filters to remove non-physiological noise and preserve functional signal fluctuations; spatial normalization for enabling the alignment of functional ultrasound images with standardized anatomical space for group comparisons; image segmentation algorithms, including both automatic and manual methods, for delineating regions of interest (ROIs) in ultrasound images; and / or algorithms for registering ultrasound data to anatomical MRI, which may include using sophisticated registration algorithms to align functional ultrasound data with corresponding anatomical MRI scans to aid visualization and alignment with standardized atlases.

[0484] Analysis tools These tools may include: a general linear model (GLM), which may include a statistical framework for modeling observed data as a linear combination of multiple predictors, including experimental tasks or conditions and confounding covariates; multi-voxel pattern analysis (MBPS), which may include advanced methods for identifying patterns across multiple voxels (rather than isolated individual voxels) associated with different states or conditions, and may support importance maps and exploration light analysis for functional localization; and / or time-course extraction algorithms, which may include tools for obtaining signal time courses from specific ROIs or voxel clusters for further examination or secondary analysis.

[0485] Visualization tools These tools may include: orthogonal viewing tools for examining ultrasound data from different angles (axial, coronal, and / or sagittal) to aid spatial understanding; 3D viewing tools for examining ultrasound data superimposed on a reconstructed grid of brain anatomy; time-series plotting tools for visualizing the temporal evolution of signals from selected ROIs or voxels; statistical plotting tools for visualizing 2D or 3D plots indicating the statistical significance of results from various analytical methods; and / or tools for visualizing registration results, such as tools for visualizing 2D or 3D plots indicating the statistical significance of results from various analytical methods.

[0486] Focus Target Specification Tool These tools may include: visualization-guided target localization for developing interactive 3D visualization tools that allow users to manually specify a focal target within the context of an anatomical scan or atlas template; and / or parametric map-based target localization that provides results based on statistical parametric maps to define the functionality of the focal target. This can be useful when the target is defined based on the results of functional analysis (e.g., brain regions showing significant activation or connectivity changes).

[0487] Neural modulation parameter adjustment tool These tools may include: a parameter tuning interface that allows users to manually adjust neuromodulation parameters, such as ultrasound frequency, intensity, duty cycle, and phase, and provides visual feedback to help users understand the spatial range and intensity of the resulting neuromodulation; and optimization algorithms that automatically adjust neuromodulation parameters based on a predefined objective function to maximize the target response. Neuromodulation parameter adjustment software tools may include those tailored to specific neuromodulation modalities. For example, in cases where the neuromodulation modality is based on ultrasound physics, the software tool may adjust parameters specific to ultrasound technology, such as ultrasound frequency, intensity, duty cycle, and phase. Graphical and / or graphical user interfaces may display to the subject a hypothetical or simulated delivery of neuromodulation (e.g., direction and intensity of delivery), optionally taking into account the subject's biomechanical constraints, such as cranial features. The neuromodulation parameter adjustment tool can also be used to adjust parameters of alternative neuromodulation modalities (such as electrophysiological methods), including deep brain stimulation (DBS), transcranial magnetic stimulation (TMS), repetitive TMS (rTMS), vagus nerve stimulation (VNS), transcranial direct current stimulation (tDCS), electrocorticography (ECOG), or any combination thereof. For example, settings in the neuromodulation parameter adjustment software can be used to select a specific neuromodulation method, which can generate a submenu showing the user the parameter settings corresponding to said specific neuromodulation method.

[0488] Safety monitoring tools These tools may include: safety limits that can enforce safety limits on ultrasound neuromodulation parameters (e.g., maximum intensity, total ultrasound energy) to prevent potential tissue damage; and / or warnings and emergency stops that may include warning signals when parameters approach safety limits, and easily accessible emergency stop functionality.

[0489] Integration and Usability Tools These tools may include: scripting interfaces, such as command-line interfaces for automating data processing and analysis tasks; GUIs, which are intuitive graphical user interfaces for interactive data exploration and pipeline configuration; interoperability tools that ensure compatibility with standard data formats and integration with existing neuroimaging software; documentation and tutorials, which may include comprehensive user guides and tutorials for different user levels, from beginners to advanced users; and / or open-source and community-driven tools that foster an open development culture where users can contribute, share experiences, and collaboratively improve the software.

[0490] Digital Diagnostic and Therapeutic Applications The methods and systems disclosed herein can be used in downstream applications such as digital diagnostics and therapy. Digital diagnostics and therapy can each include separate software, i.e., a medical device (SAMD), whose regulatory pathway is independent of the macro-scale BCI supporting said software. The development of the SAMD or neurological application (nApp) can be used to obtain further analysis and information that may be useful for clinical applications.

[0491] Figure 22 An exemplary illustration of a system is provided, comprising a patient 2020 using an application communicating with a macroscopic BCI system. The system can transmit data to one or more external devices 2024 (e.g., external centers, servers, mobile devices), which can be used by one or more developers or researchers 2022 to update existing applications and / or provide new applications (e.g., nApps) for patient use. Data can be collected from the system by an external center and input into the patient application (e.g., nApp). Data can also be transmitted to the cloud for remote access to patient data by clinicians or researchers who can provide patient monitoring or new or updated patient treatments.

[0492] nApps can be used in digital pharmacy applications. For example, when a clinician treats a patient and wishes to prescribe an nApp, they can place the order through their Electronic Health Record (EHR) system. The EHR system can then obtain payer approval for the nApp and deploy it to the patient's implanted macroscale brain imaging system (BCI). Clinicians may be able to set and assign patient-specific parameters, and data collected by the macroscale BCI (both regarding the provision of the nApp and the creation of any relevant brain recordings) can be logged to a cloud-based data lake and also placed into the patient's EHR. If the patient consents, the data can be de-identified and made available to support further research and development. Payments for nApp prescriptions (along with applicable co-payments) can also flow from the payer to the nApp's developers. Digital pharmacies can be designed to provide nApps with the same functionality that cash pharmacies offer relative to existing medications.

[0493] While the implantation of a macroscale BCI and the prescribing of nApps can utilize a clinical basis, once implanted, it is feasible to provide patients with both health-related and non-health-related applications. This functionality primarily stems from the macroscale BCI's ability to interact with the large volume of the brain, thus transcending the circuitry directly related to the clinical indications upon which the implantation is based. In fact, as long as such applications focus on monitoring rather than stimulation, it is feasible to include a set of "automated safety" functions in the methods and systems disclosed herein, and the macroscale BCI can then serve as a non-clinical interface between the patient and any number of applications. Following digital pharmacy applications, a separate application marketplace can be provided to create access to data related to the methods and systems disclosed herein.

[0494] Data obtained from the methods and systems disclosed herein can also be associated with shared registries. Registries allow developers to quickly locate existing patients and invite them to participate in new studies based on inclusion and exclusion criteria. In some cases, creating such a registry may involve clinical sites inviting subjects using the systems and methods discussed herein to join the registry. Subjects can choose to join or leave the registry.

[0495] In one or more embodiments, the nApp according to the embodiments of this disclosure can target one or more use cases summarized in Table 1 below:

[0496] Computer Systems and Networks Figure 24 An example of a computing device or system according to one implementation is illustrated. Device 2400 may be a host computer connected to a network. Device 2400 may be a client computer or a server. Figure 24As shown, device 2400 can be any suitable type of microprocessor-based device, such as a personal computer, workstation, server, or handheld computing device (portable electronic device), such as a telephone or tablet. The device may include, for example, one or more processors 2410, input devices 2420, output devices 2430, memory or storage devices 2440, and communication devices 2460. Software 2450 residing in memory or storage devices 2440 may include, for example, an operating system and software for performing the methods described herein. Input devices 2420 and output devices 2430 may generally correspond to those described herein and may be capable of being connected to or integrated with a computer.

[0497] Input device 2420 can be any suitable device that provides input, such as a touchscreen, keyboard or keypad, mouse, or voice recognition device. Output device 2430 can be any suitable device that provides output, such as a touchscreen, haptic device, or speaker.

[0498] Storage device 2440 can be any suitable device providing storage (e.g., electrical memory, magnetic memory, or optical memory, including RAM (volatile and non-volatile), cache, hard disk drive, or removable storage disk). Communication device 2460 can include any suitable device capable of sending and receiving signals over a network, such as a network interface chip or device. Computer components can be connected in any suitable manner, such as via wired media. ( For example, a physical system bus 2480, an Ethernet connection, or any other wired transmission technology) or wirelessly (e.g., Bluetooth®, Wi-Fi®, or any other wireless technology).

[0499] Software module 2450, which can be stored as executable instructions in storage device 2440 and executed by processor 2410, may include, for example, an operating system and / or processes embodying the functionality of the methods of this disclosure (e.g., as embodied in a device as described herein).

[0500] Software module 2450 may also be stored and / or transferred in any non-transitory computer-readable storage medium for use or in connection with an instruction execution system, apparatus, or device such as those described herein, from which instructions associated with the software can be retrieved and executed. In the context of this disclosure, a computer-readable storage medium can be any medium, such as storage device 2470, which may contain or store processes for use or in connection with an instruction execution system, apparatus, or device. Examples of computer-readable storage media may include memory cells (such as hard disk drives, flash drives) and distributed modules operating as a single functional unit. Furthermore, the various processes described herein may be embodied as modules configured to operate according to the embodiments and techniques described above. Moreover, while processes may be shown and / or described individually, those skilled in the art will understand that the aforementioned processes may be routines or modules within other processes.

[0501] Software module 2450 can also be propagated within any transmission medium for use or in conjunction with an instruction execution system, apparatus, or device such as those described above, from which instructions associated with the software can be retrieved and executed. In the context of this disclosure, the transmission medium can be any medium capable of transmitting, propagating, or transferring a program for use by or in conjunction with an instruction execution system, apparatus, or device. Transmission readable media can include, but are not limited to, electrical, magnetic, optical, electromagnetic, or infrared wired or wireless propagation media.

[0502] Device 2400 can be connected to a network (e.g., network 2504, such as...). Figure 25 As shown and / or described below, the network can be any suitable type of interconnected communication system. The network can implement any suitable communication protocol and can be protected by any suitable security protocol. The network may include network links of any suitable arrangement that enable the transmission and reception of network signals, such as wireless network connections, T1 or T3 lines, cable networks, DSL, or telephone lines.

[0503] Device 2400 can be implemented using any operating system (e.g., an operating system suitable for operation on a network). Software module 2450 can be written in any suitable programming language, such as C, C++, Java, or Python. In various embodiments, application software embodying the functionality of this disclosure can be deployed in different configurations, such as in a client / server setup or via a web browser as, for example, a web-based application or web service. In some embodiments, the operating system is comprised of one or more processors (e.g., ， Executed by processor 2410.

[0504] Figure 25 An example of a computing system according to one embodiment is illustrated. In system 2500, device 2500 (e.g., as described above and as...) Figure 25 (As illustrated) is connected to network 2504, which is also connected to device 2506. In some embodiments, device 2506 is a peripheral control unit capable of controlling at least one ultrasonic transducer.

[0505] Devices 2400 and 2506 can communicate via network 2504, for example, using a suitable communication interface. various like Local area network (LAN), virtual private network (VPN), or the Internet. In some implementations, network 2504 can be, for example, the Internet, intranet, VPN, cloud network, wired network, or wireless network. Devices 2400 and 2506 can communicate partially or entirely via wireless or hardwired communication (such as Ethernet, IEEE 802.11b wireless, etc.). Alternatively, devices 2400 and 2506 can communicate via a second network ( Such as Communication may occur via mobile / cellular networks, for example, using suitable communication interfaces. Communication between devices 2400 and 2506 may also include communication with or with various servers (such as mail servers, mobile servers, media servers, telephone servers, etc.). In some embodiments, devices 2400 and 2506 may communicate directly (instead of through network 2504, or in other ways than through a network), for example, via wireless or hardwired communication, such as Ethernet, IEEE 802.11b wireless, etc. In some embodiments, devices 2400 and 2506 communicate via communication 2508, which may be a direct connection or may occur through a network (e.g., network 2504).

[0506] One or both of devices 2400 and 2506 typically include logic (e.g., HTTP web server logic) or are programmed to perform the following operations: formatting data accessed from a local or remote database or other data and content source for use in providing and / or receiving information over network 2504 according to the various examples described herein.

[0507] Example Examples 1 through 3 described herein utilize a simulation framework for rapid prototyping of imaging and neuromodulation solutions. The simulation approach is based on maximizing the physical, anatomical, and physiological realism of the system. It generates ultrasound images based on fundamental principles of wave propagation, reflection, aberrations, reverberation, and scattering within soft tissue and blood vessels. These methods are based on numerical solution tools for solving the full-wave equations. The physics-based approach presented in the examples allows for rapid design iterations of 1) transducer arrays, 2) imaging sequences, and 3) optimized imaging parameters. Acoustic propagation within the brain volume is simulated to determine the total illumination capacity of the array. The array configuration (including array characteristics such as placement geometry and frequency) and tissue characteristics (e.g., scatterer contrast, clutter, target depth, neurovascular flow characteristics) are then iteratively simulated to optimize the imaging process. The originally large parameter space is limited by implementing use-case-specific constraints. For example, the simulation is limited to physically feasible transducer placement locations using physiological data extracted from human CT / MRI scans. Furthermore, the imaging field can be constrained by using brain regions of interest associated with specific neuropathologies.

[0508] Example 1 - Indicative Thermal Budget Thermal load can depend on specific imaging or neuromodulation sequence parameters, including the number of transmissions, duration, and duty cycle. These parameters can be optimized in conjunction with hardware development. The thermal budget for power consumption can be estimated by solving the Penne bioheat transfer equation for a given hardware form factor using numerical simulation tools. Power consumption is distributed among the implanted components in the head and chest. The estimation of electrical power consumption uses conversion efficiency estimates of the CMUT array, per-element power consumption of the analog front end and digitization, the digital computational costs of filtering, demodulation, beamforming, and power Doppler estimation, and other chip processes (e.g., power management, timing circuitry, etc.). By keeping the maximum temperature rise within accepted standards, various imaging and neuromodulation sequence parameters affecting heat generation can be optimized.

[0509] During operation of a macroscopic BCI system, one or more components (e.g., peripheral controllers and implantable transducers) may experience temperature rise. For example, the implantable transducer is expected to experience a temperature rise when emitting and / or receiving ultrasound waves. A major constraint affecting the imaging solution is the power budget available to the disclosed system while meeting the thermal safety requirements of the implanted system. That is, the system described herein should not heat the subject's tissue temperature by more than 2 degrees Celsius, and for the brain, the absolute limit is 39 degrees Celsius (ISO 14708-3). Computational studies can provide accurate preliminary estimates for modeling the thermal characteristics of cranial implants (e.g., implantable transducers) according to embodiments of this disclosure.

[0510] Figure 26Non-limiting exemplary results from numerical simulations of an example cranial implant 2612 are depicted, with a thermal power budget of approximately 350 mW while complying with ISO tissue heating safety standards. Within the cranial implant 2612, forming a single 2D functional brain image per second consumes approximately 20 mW of power. Therefore, the envisioned system could support the acquisition of approximately 16 2D functional brain data slices per second, which could provide high spatial coverage for our indicative imaging solution. Figure 15B An exemplary imaging solution can be visualized, where a 2D plane is shown as a cross-section superimposed on an anatomical MRI, thereby exhibiting significant and programmable coverage. A more in-depth feasibility analysis of the components of the ultrasound imaging system is performed according to the embodiments disclosed herein. Specifically, the cranial implant 2612 and the remote processing unit of the ultrasound imaging system are analyzed.

[0511] This example illustrates the results of a numerical model of the thermal effects of an implanted transducer on adjacent tissue. FDA safety constraints were applied to the estimated power budget using simulation software (e.g., COMSOL). The results are presented in... Figure 26 (It depicts visualizations of the scalp 2614, brain 2616, and subarachnoid space 2618.) In some respects, the numerical model assumes that the cranial implant 2612 is a three-dimensional cylindrical model with a diameter of 25 mm, extending through the skull 2610 to a depth such as... Figure 2B and Figure 26 The implant extends 35 mm onto the surface of skull 2610 to provide a physical base for securing the implant to the skull and dissipating additional heat. The components of the physical model are defined using validated model parameters, as depicted in Tables 3 and 4. Numerical simulation results indicate a steady-state thermal budget of 350 mW to 400 mW. Figure 26 As instructed.

[0512] Table 3: Model Parameters

[0513] Table 4: Model Parameters These results provide an upper limit to the expected power budget for each implantable transducer. Using additional power would heat the implantable transducer beyond what is considered safe under current guidelines.

[0514] Example 2 - Parameter Optimization for Imaging Using Ultrasonic Transducers According to embodiments of this disclosure, a BCI system may include multiple implantable transducers. In such embodiments, placing the implantable transducers on the skull of a subject affects the system's imaging and neuromodulation functionality. Optimizing the placement of the implantable transducers can produce constructive or destructive interference. To efficiently deliver ultrasound energy to desired regions of the brain, determining the optimal placement of the implantable transducers on the skull is useful. Simulations used to determine potential multi-implantable transducer designs are detailed below.

[0515] Sequence design and validation pathways can be explored through simulations of imaging and neuromodulation pulses using multi-array designs. In the anatomically and acoustically aligned simulations, the imaging and neuromodulation pulses are modeled as either (i) independent plane waves or (ii) a “one-to-one” composite plane wave. Results are based on two virtual transducers implanted at a 45-degree angle relative to each other, with an aperture of 20 mm and a spacing of [missing information]. Other configurations are possible regarding placement, array size, spacing, slits, frequency, and number of arrays. This can be achieved using magnetic resonance (MR) scans of one or more human subjects. Figure 27A To determine the general target area ( Figure 27B ) and the location of a specific loop ( Figure 27C (This refers to the nucleus accumbens, anterior cingulate cortex, and orbitofrontal cortex, areas associated with circuit-based mechanisms of depression.) The scan data was then converted into a map of acoustic properties. Figure 27D The acoustic properties include sound velocity, density, attenuation, nonlinearity, and scatterer brightness. A plane wave (5 MHz, 3 cycles, δθ=4) propagates into the sound field. Reflection, based on the first principles of propagation, uses a conventional delay-superposition composite beamformer to generate an ultrasonic image. Figure 27F , Figure 27H ), and intensity fields generated based on heterogeneous propagation paths in brain tissue ( Figure 27E , Figure 27G ).

[0516] Two imaging approaches are presented, illustrating the differences in image quality based on the following imaging parameters: (i) a single plane wave emitted by a single array ( Figure 27F , Figure 27G (ii) 23 composite plane waves emitted by two arrays oriented at 45 degrees relative to each other. Figure 27H , Figure 27I By matching and simulating ATS widget phantoms in two scenarios, further quantification of resolution and contrast is provided. Figure 27J , Figure 25 Because this homogeneous material is unlikely to be affected by the same image degradation sources that may be encountered in the brain, the phantom can provide an ideal imaging scenario.

[0517] By designing arrays and ultrasound beams that can be easily turned to large angles in both azimuth and elevation, it is possible to cover brain volume using small, implantable ultrasound devices that can be used in burr hole surgery. This capability is inversely proportional to element size or spacing, with improved performance at half the wavelength of the transmitted frequency. The plateau phase is reached at this point, which is common in sector scanning imaging. Despite the presence of large rectangular elements in PZT... Linear arrays have been widely used in clinical applications for over two decades, but their use in matrix imaging has been limited to a few research implementations due to their fabrication complexity. In some respects, reliable CMUT fabrication techniques can be used, which simultaneously address array cutting, sensitivity, and electronic interconnection issues, thereby enabling fully addressable arrays. Spacing matrix array. Furthermore, in some respects, large angular sensitivity can be even more pronounced in volumetric imaging: for example, a single ultrasound array with a base of 16.2 mm, an imaging depth of 80 mm, and an angular sensitivity of 45 degrees can have an imaging volume of 315 cc, which is about 25% of the average volume of the human brain.

[0518] Determining the spatial resolution limitations of a single array system can be relatively straightforward. However, achieving resolution and sensitivity to fUS signals using a multi-array imaging system is complex and depends on a wide range of parameters, including hardware capabilities (e.g., noise level in amplifiers, digitization bit depth), the choice of transmission sequence (e.g., plane wave, focused imaging, synthetic aperture approach), frame rate or collective length, and post-processing algorithms (e.g., beamforming, single-valued decomposition (SVD) filters), the brain volume being imaged for the task (e.g., whole-brain imaging vs. targeted imaging), and the temporal resolution of obtaining the fUS signal (continuous monitoring vs. on-demand imaging). Furthermore, these choices and the trade-offs they represent can have complex dependencies on power consumption and heating, which can be strictly limited to avoid thermobiological effects.

[0519] Example 3 - Feasibility Analysis of Imaging the Nervous System The systems and methods described herein can generate high-quality, functional images with large coverage and high frame rates within the constraints of implantable shape factors. The disclosed systems and methods may include transmission and reception techniques optimized for low-power applications, as well as methods for subsequent processing. The disclosed systems and methods can support multiple imaging modes optimized for different use cases, such as those determined by requirements such as imaging depth, coverage, and frame rate.

[0520] Different sequences of functional imaging include wide-beam / explosive scanning, row-column addressing, sub-aperture plane-wave recombination, and sparse-aperture imaging approaches. These methods involve trade-offs in imaging depth, frame rate, recombination efficiency, and sensitivity to heterogeneity, and their relevance varies depending on the specific requirements of the use case. The simulations described in this paper can generate a first-order characterization of image sequence design, including point spread function analysis, contrast-to-noise ratio, electronic and thermal noise, grating lobes, and their frequency dependence. If the model proves to improve generalization accuracy from modeling to in vivo operation, it can be extended to include (modeling within the aforementioned validation framework and...) simulation The effects of brain heterogeneity, off-axis and reverberant clutter, blood pool contrast, and neurovascular flow differences can be considered. Analysis may include the effects of recombination, frame rate, noise sensitivity, and array design. The effects of receive beamforming pathways, apodization, beamforming operations, recombination, partial recombination, coherence factor, and multiplexing can be tested, while understanding the hardware implementation of the solution is crucial, and therefore includes constraints such as the number of digitizable channels, communication bandwidth between the cranial implant and the remote processing module, and storage and computational constraints in digital processing. Therefore, imaging design can be performed as part of hardware co-optimization. Optimization parameters may include the estimated total number of frames required to fill the brain volume, beamforming mesh size / position, RF and IQ processing, the required number of channels, and the complexity of operations within the signal processing chain, as well as how these different processing methods are partitioned among the different hardware modules constituting the system (e.g., within the cranial implant, thoracic implant, or remote externalization device).

[0521] In some implementations, power consumption feasibility calculations based on 2D plane wave imaging serve as a baseline for characterizing power requirements and the resulting functionalities of indicative transmission, reception, and processing pathways. In this context, functional brain data can be acquired layer by layer, programmed by a subset of transducer elements that can be activated on a 2D transducer matrix and their timing, such as... Figure 15A As shown. This layer-by-layer approach can be viewed as analogous to the way volumetric imaging can be performed in functional MRI. For these indicative calculations, the methods and systems described herein employ a split-encapsulated shape factor, consisting of a cranial implant and a remote processing module, which may be externalized in some implementations of the described system and implanted in the chest in others. The cranial implant may include a sensor array, integrated analog processing and digitization, digital preprocessing, and wired data communication. The remote processing module is responsible for digital processing.

[0522] Cranial implants: Cranial implants may include implantable transducer arrays, integrated analog processing and digitization, digital preprocessing, and in some implementations, wired data communication. Specific variables affecting power consumption include low-noise amplifier (LNA) noise, bit depth of digitization, imaging sequences that determine the number of elements used for transmission, reception, and digitization, the number of transmissions used for power Doppler estimation of cerebral blood volume, and the reception window (among other factors). This example provides indicative power consumption calculations for 2D plane-wave imaging sequences, separately for each major processing stage. This example assumes that each functional 2D image is generated from 250 images created from different transmission and reception events. This value is the minimum that allows for accurate detection of functional brain activity using advanced processing methods validated with non-human primate data.

[0523] Assume that the sub-apertures of 1000 transducer elements on an ultrasonic array are energized to generate 2D plane waves, the capacitive load is 2 pF, and the operating frequency is... And the inter-peak voltage is Then the instantaneous transmission power can be estimated as follows: This is consistent with current state-of-the-art systems. For imaging, this peak power is only visible during the transmitted pulse, which is short (per image). Assuming 250 images are acquired, the estimated time-averaged power is: To estimate transmit power consumption, use P = 0.5 × C × V 2 × f This is used as an approximation to estimate the power consumed in an AC circuit with a capacitive load. Here, C Indicates capacitance. V This indicates the peak voltage swing, and f Indicates the signal frequency. P It is an expression for the energy stored in the capacitor. E = 0.5 × C × V 2 The result is obtained, and then multiplied by the frequency. f This takes into account the energy changes in the AC signal over time.

[0524] For an imaging depth of 8 cm, a sound velocity of 1540 m / s, an AFE of approximately 0.66 mW / element, and digitization power consumption (in continuous operation), with approximately 1000 receiving elements and microwave beamforming reduced to approximately 128 channels before digitization, the average power consumption of the receiving link is estimated to be... The power consumption of the analog front-end (AFE) and digital receiver chain used for ultrasound imaging can be estimated based on element-level power consumption. For integrated designs, an element-level power consumption of approximately 0.5 mW to 1 mW per element is sufficient to obtain 10-bit digital data. Assuming 1000 elements are active to receive backscattered echoes, and these elements are superimposed along the elevation axis in the analog domain, the channel count is reduced to 128. An indicative power consumption of 0.66 mW per element is chosen. Energy consumption can be calculated based on the time of flight of the received ultrasound signals and the total number of images. Receive_power=(receive_window_per_image)×AFE_power_consumption× number of images Receive_power=(2×imaging_distance / speed_of_sound)×(AFE_per_element_ power × number_of_active_elements)×number_of_images Receive_power=(2×0.08 / 1540)×(0.66e -3 (× 1000)×250=17.14 mW Once the raw data is digitized, it can be transmitted directly or processed channel-by-channel (such as IQ demodulation, smoothing, and decimation) to reduce data load. For this application, assume the following transmission of raw digital data: 128 digital channels, a sampling frequency twice the transmission frequency (e.g., 10 MHz), and a bit depth of 10. These parameters will result in a data transmission rate of 332 Mb / s. This rate can be calculated based on the data time per image and the total number of images. The data per image is a function of the number of channels, sampling rate, sampling window, and bit depth. Therefore, the total data is given by the following formula: total_data=data_per_image×number_of_images total_data=(receive_window_per_image×sampling_rate×number_of_ channels×bit_depth)×number_of_images total_data=((2×imaging_distance / speed_of_sound)×sampling_rate× number_of_channels×bit_depth)×number_of_images total_data=(2×0.08 / 1540×10e 6 (×128×10)×250=332Mb / s.

[0525] The channel loss of a wired transmitter depends on design, materials, frequency, and distance. Given our data rate and the relatively short distance between the cranial implant and the chest unit, we can assume an energy efficiency of approximately 1.5 pJ / bit. Therefore, the power consumption at a data rate of 332 Mb / s would be roughly equivalent to... Therefore, if the feasibility analysis is limited to the active processes of transmitting, receiving, and preprocessing 2D plane wave image data, we expect the total skull energy consumption to be: These power consumption values ​​will allow for the acquisition of approximately 16 2D functional brain data slices per second, while remaining within the calculated thermal budget of approximately 325 mW.

[0526] Remote Processing Unit: The remote processing module is responsible for digitizing the raw ultrasound data into functional brain images. To understand the power requirements, the number of operations required for each processing stage (e.g., multiply-accumulate or MAC operations) can be estimated, and then the power is estimated based on an indicative process (e.g., 22 nm). For feasibility calculations, the following three processing stages are estimated: 1) IQ demodulation, filtering, and decimation of RF data to baseband; 2) image formation (“beamforming”); and 3) power Doppler estimation. Alternative processing schemes are possible, but this ordering is understood to minimize power consumption. Given our system parameters, demodulation, filtering, and decimation require approximately 400 million MACs. Regarding IQ demodulation, when analyzing ultrasound, the raw RF signal is demodulated into in-phase and quadrature (IQ) components. This typically involves multiplying the RF signal by a cosine wave (for the in-phase component) and a sine wave (for the quadrature component), each at the carrier frequency. Regarding filtering and decimation, the data is passed through a low-pass filter after demodulation. The number of MACs used for this operation depends on the number of filter taps and the number of samples. If the filter has M taps (e.g., 5 taps for a dual second-order IIR filter) and there are N samples, then each I or Q branch uses M×N MACs. Therefore, the total number of MACs for this stage will be 2×M×N. Decimation may involve downsampling, which does not require MAC operations (although anti-aliasing filters may be included).

[0527] MACs= 2 × M × N × number_of_images MACs= 2×5×1038 × 128 ×256 = 340e 6 This results in two multiplication operations per channel per sample: one for the in-phase component and one for the quadrature component. Therefore, the number of MACs is... 2×N×number_of_images ,in N This represents the total number of samples per image. In our case: Image formation via beamforming and coherent composite can utilize approximately 8.2 billion MACs. The anticipated number of required MACs justifies the development of a custom DSP chip. The final processing stage estimates the power Doppler from the collective of the collected images. This processing can be performed on-chip within an embedded shape factor, or alternatively, the reconstructed B-mode image can be off-device for further processing. For example, a collective B-mode of 16 2D data slices will yield approximately 262 Mb / s after composite.

[0528] Regarding the incorporation of beamforming and coherent recombination in the feasibility analysis, the analysis posits that in RF space, information from a single scatterer is dispersed across channels, determined by the time of flight from the scatterer's position in the medium to the transducer's position, thus forming a hyperbolic shape in the mixed spatiotemporal space of the RF signal. Beamforming is the process of superimposing along the hyperbola to integrate all energy originating from the scatterer at every point in physical space. In practice, the value of the hyperbola is determined by pre-calculated interpolation. Therefore, the beamforming operation for each beamformed pixel is proportional to the number of channels and the interpolation factor. In practice, this number is reduced based on the effective f-number of the imaging setup, thus taking into account the inhomogeneity of the effective signal-to-noise ratio of the backscattered echo across the hyperbolic characteristics. This is primarily determined by the directivity of the transducer elements. Here, we conservatively estimate and assume the number of MACs per beamformed pixel is approximately 500. This means the total number of MACs is dominated by the number of beamformed pixels per image (here we assume a 256×256 image) and the number of images: MACs= number_beamformed_pixels×500×number_of_images MACs= 256×256×500×250 = 8.2e 9 Example 4 - Feasibility Analysis of Neural Modulation The design of the neural modulation sequence imposes fewer constraints on the acoustic design of the transducer (e.g., focusing capability and sensitivity) but more constraints on the acoustic load. The reduced constraints are due to the fact that neural modulation uses pulse trains lasting several milliseconds, while imaging pulse trains are typically <1 µs. The focus gain estimate is mapped across the intervention volume to estimate the required pressure level at the transducer surface. An estimate of the upper limit is then obtained via derating tank measurements, and a more accurate representation of the effects of attenuation, aberrations, nonlinearity, registration error, and focus characteristics is provided through simulation of the heterogeneous tissue characteristics described in the imaging section.

[0529] Although ultrasound imaging and neuromodulation can occur using the same hardware, their technical requirements are fundamentally different. Ultrasound imaging works by delivering brief (<1 μs) pulses of mechanical energy into the brain, measuring the backscattered echo, and estimating the neurovascular state of the tissue. Given the extremely short pulse duration, most of the power consumed by ultrasound imaging is in the receiving processing. In contrast, ultrasound neuromodulation typically relies on the ability to deliver extended pulse trains of approximately 10 ms without any receiving processing, such as... Figure 15CAs depicted. Therefore, the main technical challenge of ultrasound stimulation relates to the ability to deliver sufficient acoustic intensity to the neuromodulation target over an extended duration characterizing the effective neuromodulation parameters in published literature. A non-limiting exemplary result employs a system consisting of three transducers, each a square grid composed of 112 × 112 elements spaced 150 μm apart (e.g., the same array used for imaging). Using multiple transducers as part of a neuromodulation enhancement system can provide better spatial specificity of the focusing region and increased intensity through the constructive superposition of beam profiles generated by each array (in the case of three arrays, constructive interference can allow for a nine-fold increase in intensity). Figure 15D This exemplary design concept is depicted. Results show that the system described herein can deliver approximately 69, 27, and 14 W / cm at focal depths of 4, 6, and 8 cm, respectively. 2 The spatial peak temporal mean intensity (I_SPTA) is approximately the largest reported in the neuromodulation literature. I_spta Twice that. Furthermore, the system described herein can deliver 25 W / cm² at a depth of 8 cm over a prolonged time period within a 30 ms pulse duration. 2 Spatial peak pulse average ( I_sppa This matches the average peak pulse intensity recorded in the literature.

[0530] The feasibility analysis described above can be examined in further detail. More precisely, this feasibility analysis focuses on the disclosed system's ability to insonify the focal target location using delivery specifications that have been validated to elicit neuromodulation effects. Parameters related to technical feasibility include delivering sufficient acoustic intensity (e.g., I_spta Up to 7 W / cm 2 And the ability to maintain extended pulse trains (e.g., up to about 30 ms).

[0531] The spatial peak time-averaged intensity that can be delivered to the focal target at several representative depths ( I_spta The non-limiting example system described herein consists of three transducers, each a square grid composed of 112 × 112 elements spaced 150 μm apart. Using multiple transducers as part of a neuromodulation enhancement system allows for improved spatial specificity of the focusing region and increased intensity through the constructive superposition of beam profiles generated by each array. The approach described in this example is: calculating for each individual transducer... I_spta Furthermore, we assume linearity to calculate the results for the entire system.

[0532] from Figure 26As seen in the simulation modeling, the electrical power budget is limited to 325 mW. Considering the electrical-to-acoustic power conversion efficiency of 70%, the acoustic power (Pa) generated by the CMUT array can be calculated as follows: Pa = efficiency × Pe = 0.7 × 0.325 W = 0.2275 W The total effective area (A) of the CMUT array is: A = (Number of elements along one edge × Spacing) 2 = (112 components × 150e) -4 ) 2 = 2.82 cm 2 The resulting spatial peak time-averaged intensity at the transducer surface ( Is_spta )for: Is_spta = Pa / A = 0.2275 W / 2.82 cm 2 = 0.0806 W / cm 2 The above calculations assume a typical duty cycle of approximately 50%. Therefore, I_spta This corresponds to 0.16 W / cm 2 of I_sppa The measured surface pressure is 0.071 MPa, which is far below the technical capabilities of commercial CMUT arrays. For example, a standard CMUT array is assumed to be able to achieve a surface pressure of 1 MPa when driven by an RF voltage of 50 V.

[0533] For neuromodulation, ultrasonic energy can be focused to a specific location using the transmission timing of individual elements. This focusing results in constructive interference, which increases acoustic pressure (and therefore intensity), characterized by the system's focusing gain. To characterize this focusing gain, numerical simulations were performed on our 112×112 150 μm pitch transducer array. A homogeneous medium with an attenuation coefficient of 0.5 dB / cm / MHz and a transmission frequency of 4 MHz was assumed. The focus gains in terms of acoustic pressure for depths of 4, 6, and 8 cm were calculated to be 9.72, 6.15, and 4.38, respectively. At a 6 cm focus, the focus gain indicates that the pressure at the focus is 6.15 times the pressure at the transducer surface. Figure 28 The cross-section of the simulated output is shown, illustrating the intensity profile of a single transducer at a depth of focus of 6 cm.

[0534] The intensity of ultrasound is proportional to the square of the pressure. Therefore, assuming the focusing gain for pressure is G, the focal intensity ( If_spta ) can be calculated as If_spta = G 2 × Is_spta ,in I = P 2 / (2 × ρ × c) ,in I For strength, P For pressure, ρ For density, and c Let be the speed of sound. If the following terms are defined as the focusing gain for pressure and the focusing gain for intensity, G_P = P_focused / P_surface G_I = I_focused / I_surface Then we can substitute the strength formula into G_I From the definition, we get: G_I = (P_focused 2 / 2ρv) / (P_surface 2 / 2ρv) G_I = P_focused 2 / P_surface 2 G_I = G_P 2 therefore: If_spta (4 cm) = G 2 × Is_spta = 9.72 2 × 0.0806 W / cm 2 = 7.6 W / cm 2 If_spta (6 cm) = G 2 × Is_spta = 6.15 2 × 0.0806 W / cm 2 = 3.1 W / cm 2 If_spta (8 cm) = G 2 × Is_spta = 4.38 2 × 0.0806 W / cm 2 = 1.5 W / cm 2 Given that the system operating in this example may involve three such arrays, the total focal intensity at these distances ( If_spta It can be approximated as: If_spta (4 cm) = 68.6 If_spta (6 cm) = 27.4 If_spta (8 cm) = 13.9 Given that when multiple pressure waves are incident on the same point and their phases are aligned, the resulting total pressure is a linear superposition of the individual pressure amplitudes (by allowing the use of the same constructive interference focused by a phased array), the above approximation is possible. Based on this relationship, the net intensity of the three transducers can be calculated as follows: I_total=(P_total) 2 / (2 × ρ × c) in P_total The total pressure caused by the three transducers, and based on the linear assumption, is equal to... P_trans1 +P_trans2+P_trans3 Furthermore, assuming that the pressure caused by each transducer is equal, the relationship can be written as follows: 3×P_ single_transducer It should be noted that in practice, it is unlikely that equal pressures will be assumed, as transducers are likely located at different distances and can travel through different media; however, this unlikely assumption represents a reasonable first-order approximation. Based on the above considerations: The relationship between pressure and intensity at a single transducer is as follows: P_single_trasnducer 2 =I_single_ tansducer×2×rho×c After substituting and canceling, we get: I_total=9×I_single_transducer Or more generally: I_total=num_transducers 2 ×I_single_transducer .

[0535] Comparing these values ​​to our effective neuromodulation parameter table shows that at least twice the maximum intensity can be achieved even at a depth of 8 cm. Therefore, the thermal budget of the implant allows for effective neuromodulation spatial peak time-averaged intensity. Other factors, such as the heterogeneity of the medium and the relative spatial position of the transducer with respect to the focal point, may affect these values, but given our considerable margin of error, these factors will not prove to be obstacles.

[0536] Another aspect of feasibility is whether sufficient sustained power can be provided to the transducer to maintain neuromodulation intensity within the pulse duration (e.g., approximately 0.5 ms to 30 ms, using duty cycle mode) and the total treatment window. For this indicative feasibility calculation, the highest power neuromodulation parameter set was used at our focal location, a challenging scenario with a pulse duration of 30 ms, a duty cycle of 30%, a total treatment window of 40 s, and a spatial peak pulse-to-average intensity (PMI). I_sppa ) is 25 W / cm 2Reversing the above calculations (e.g., shifting from desired focal intensity to electrical power level) yields a target peak power of 584 mW for each cranial implant. The electrical power per transducer is: If_sppa=If_sppa_total / num_trasducers 2 Each transducer surface If_sppa for: I_surface=If_sppa / Gp 2 Sound power is expressed as: Pa = I_surface × A The electrical power is: Pe = Pa / Eff; When using 25 W / cm 2 goal If_sppa At that time, the required electrical power was 584 mW.

[0537] A single Li-cell battery with a capacity of 3000 mAh to 4000 mAh, powering up to three cranial implants, can provide neuromodulation current without significant voltage drop or capacity degradation. Finally, additional parameters can affect ultrasound neuromodulation. For example, tissue heterogeneity can distort individual beams, thus affecting cross-transducer focusing gain and beam alignment. This latter effect may also depend on the multi-transducer arrangement. Importantly, imaging and neuromodulation can be performed using the same array, so any distortion introduced by heterogeneous acoustic paths can be mapped to the imaging space. Therefore, distortion is effectively and automatically compensated when targeting a focal location defined in the imaging space using the same array for imaging. In conclusion, the feasibility analysis presented in this paper demonstrates that the disclosed system can even be matched to the most stringent acoustic parameters that have been proven to provide effective neuromodulation.

[0538] Figure 29A and Figure 29B A color map of pressure is provided when one or more ultrasound transducers apply ultrasound waves to the skull of a human subject. Figure 29A A simulated quantitative 2D graph is provided, which depicts the pressure exerted on the human skull by ultrasound waves emitted from the ultrasonic transducer. Figure 29B Provides a way to generate Figure 29A The material mapping shown in the simulation results makes Figure 29B The charts shown can be overlaid. Figure 29A As shown in the chart. For Figure 29A and Figure 29B Both depict the skull as a 2D projection viewed from above, and depict a single 2D slice along the z-axis. For Figure 29A and Figure 29BBoth charts have their y-axis plotting x-dimensions in meters, and their x-axis plotting y-dimensions in meters, such that the 0.00 m position of the y-dimension in meters is centered at the midpoint of the y-dimension in meters. For Figure 29A The right side of the chart shows color bars that quantify pressure across space as the color changes, with units of Pa. The color bars range from less than 50,000 Pa to greater than 400,000 Pa. Figure 29A The pressure field was depicted when three ultrasound transducers applied ultrasound waves to the subject's brain during ultrasound neuromodulation of the subject's brain. Figure 29B Depicts ultrasonic waves being transmitted through to generate Figure 29A The pressure field diagram shows three different media: water, the subject's skull, and the subject's brain. The circle in the center of the diagram with two quarter circles (2901) represents the target on which pressure is applied.

[0539] Figure 30A A simulated 3D graph of the pressure exerted when ultrasound is applied to a human subject from a non-implantable ultrasound transducer device is provided. Figure 30A It depicts the intensity of the pressure relative to other areas of the chart. Figure 30A Three ultrasonic wave emission points 3002 are shown, each corresponding to the position of one of the three ultrasonic transducers. The emitted ultrasonic waves intersect at a hot spot 3004, where constructive interference occurs. Figure 15D This intersection of ultrasound waves is also depicted in the figure, which refers to the intersection of ultrasound waves as beam intersection. The location of hotspot 3004 is deep inside the subject's skull, indicating the ability of each ultrasound transducer to deliver ultrasound waves deep into the subject's brain. Figure 30A The simulated 3D stress chart shown was generated using the open-source simulation software Neurotech Development Kit (Los Angeles, California).

[0540] Figure 30B A simulated 3D graph of the pressure exerted when ultrasound is applied to a subject from an implanted transducer device is provided. Figure 30B It depicts the intensity of the pressure relative to other areas of the chart. Figure 30B The pressure in the internal and external regions of the subject's skull (the brain is located inside the subject's skull) was depicted. Figure 30B Three implantable ultrasound transducers 3006 are shown implanted in the skull of a subject. The spatial distribution of pressure values ​​from ultrasound waves emitted from the three implantable ultrasound transducers is plotted. These three implantable transducers may include, for example, […]. Figure 5The miniature implantable ultrasonic transducer is depicted. The simulated strongest pressure value was observed at hotspot 3008, where each of the three hotspots corresponds to each of the three implantable transducers, thereby achieving constructive interference across the ultrasonic waves and generating hotspot 3008. Figure 15D This intersection of ultrasound waves is also depicted in the figure, which refers to the intersection of ultrasound waves as beam intersection. The location of hotspot 3008 is deep inside the subject's skull, indicating the ability of each ultrasound transducer to deliver ultrasound waves deep into the subject's brain. Figure 30B The simulated 3D stress chart shown was generated using the open-source simulation software Neurotech Development Kit (Los Angeles, California).

[0541] Example 5 - Imaging compatibility analysis of acoustic transmission window Figures 31A to 31G Data are provided describing the ultrasound B-mode imaging quality and transparency of an ultrasound transducer that includes one of several acoustically transparent materials. Figures 31A to 31G Each figure in the diagram shows the imaging quality transparency of the ultrasonic transducer through a lens made of a specific acoustically transparent material. Figures 31A to 31G The data shown are images of an imaging phantom configured for B-mode ultrasound imaging. The imaging phantom can serve as an object configured to operate as a substitute for human tissue when imaged by an imaging modality such as ultrasound imaging, e.g., focused ultrasound imaging. Therefore, Figures 31A to 31G Each graph in the image illustrates the effectiveness of each type of acoustically transparent material in maintaining clear ultrasonic signal transmission. (Used to generate...) Figures 31A to 31G The ultrasound transducer depicted includes an hermetically sealed housing designed to fit within the cranial region, with dimensions ranging from 3 mm to 20 mm in thickness and from 14 mm to 50 mm in width or length. The housing is designed to accommodate cables for unidirectional power and bidirectional data transmission.

[0542] Figure 31A Imaging data of an imaging phantom are shown when an ultrasonic transducer emits ultrasonic waves through ultra-high molecular weight polyethylene (UHMW-PE). Figure 31B The image data of the imaging phantom is shown when the ultrasonic transducer emits ultrasonic waves through PTFE. Figure 31C Imaging data of an imaging phantom are shown when an ultrasonic transducer emits ultrasonic waves through polymethyl methacrylate (PMMA). Figure 31D Imaging data of an imaging phantom is shown when an ultrasonic transducer emits ultrasonic waves through polyethylene terephthalate (PET). Figure 31E Imaging data of an imaging phantom is shown when an ultrasonic transducer emits ultrasonic waves through polyetheretherketone (PEEK). Figure 31FImaging data of an imaging phantom are shown when an ultrasonic transducer emits ultrasonic waves through polyether block amide (PEBAX). Figure 31G Imaging data of an imaging phantom is shown when an ultrasonic transducer emits ultrasonic waves through high-density polyethylene (HDPE).

[0543] Figure 32 A tabular table describing the physical properties of ultrasound waves passing through one of several acoustically transparent materials is provided. Both theoretical and experimental data are shown for the following acoustically transparent materials: UHMW-PE, PTFE, PMMA, PET, PEEK, PEBAX, LDPE, and HDPE. For each acoustically transparent material, the table provides density, velocity of sound (e.g., the velocity of sound as a sound wave, such as ultrasound, is transmitted through the material), attenuation coefficient, impedance (e.g., impedance coefficient), impedance ratio, reflection (e.g., reflection coefficient), transmission (e.g., transmission coefficient), and total attenuation of ultrasound waves at a frequency of 5 MHz. Values ​​provided for each of these properties include: a) a minimum value minus 50% of the range, where the range is the difference between the minimum and maximum values ​​of a given property; b) the minimum value; c) the mean value; d) the maximum value; and e) the maximum value plus 50% of the range, where the range is the difference between the minimum and maximum values ​​of a given property. Figure 32 The “N / A” value in the table data means that the value is not applicable. Figure 32 The impedance ratio shown can represent the acoustic impedance of the material relative to the reference medium (in Figure 32 The ratio of acoustic impedance to that of water (in the case of water). The total attenuation of a 5 MHz ultrasonic wave refers to the degree of attenuation that occurs when the ultrasonic wave travels a certain distance across a sound-transmitting material. The thickness of the sound-transmitting material is not taken into account when determining the total attenuation of the ultrasonic wave.

[0544] Example 6 - Feasibility Analysis of Regulating the Nervous System This example describes an experiment for validating the use of an ultrasound transducer configured for imaging and / or modulating the nervous system. Validation may include, for example, quantifying the physical properties of ultrasound waves emitted from one or more ultrasound transducers. Quantifying the physical properties of the ultrasound waves may be directed at the ultrasound waves emitted from the ultrasound transducers when one or more ultrasound transducers are configured for imaging or modulating the nervous system of a subject. Ultrasound waves from an ultrasound transducer configured to neuromodulate a subject may include delivered focused ultrasound waves.

[0545] Figure 33 An image of a laboratory is depicted, in which an ultrasonic transducer is configured to apply ultrasound waves to water tank 3306. This is achieved, in part, by adding an additional capacitor not present in the circuitry of the ultrasonic transducer configured for imaging the nervous system. Figure 33The depicted ultrasound transducer is modified for modulating the nervous system, rather than for imaging it. The ultrasound transducer is not designed for implantation into the subject's skull, but rather configured for use outside the skull. The ultrasound transducer can be manually held and / or positioned by a clinician (e.g., a surgeon or surgical assistant), or by a mechanical device (such as...) Figure 33 The described clamping bracket is used for holding and / or positioning. Figure 33 An ultrasonic transducer 3302, secured by a clamp 3304, is depicted. The ultrasonic transducer 3302 is oriented such that the MEMS ultrasonic array points towards a water tank 3306. The water tank can be used to estimate the impedance encountered by the ultrasonic waves when applied to the human brain. The ultrasonic transducer may contact the surface of the water tank, such as... Figure 33 The description.

[0546] Figures 34A to 34E This demonstrates the ability of ultrasound transducers to deliver ultrasound waves that can modulate the nervous system of a subject. Figures 34A to 34E The data depicted is based on laboratories used to test the characteristics of ultrasonic transducers, such as... Figure 33 The laboratory shown. Furthermore, in part by adding additional capacitors not present in the circuitry of the ultrasound transducer configured for imaging the nervous system, the ultrasound transducer was modified to modulate the nervous system, rather than to image it. A target set of physical ultrasound parameters for modulating the subject's nervous system was selected. This target set of ultrasound parameters may include a pulse length of at least approximately 300 ms, a focal pressure of at least approximately 1 MPa, and a frequency of at least approximately 5 MHz. For example... Figure 33 As shown, physical ultrasonic parameters from a modified ultrasonic transducer were measured using a hydrophone surface pressure measurement system with an unfocused transmission at 3 MHz, from a laboratory used for testing ultrasound.

[0547] Figure 34A The long pulse duration of ultrasound waves from an ultrasound transducer configured for neuromodulation of a subject is shown. Figure 34A The duration of long pulses suitable for real-time (e.g., online) neural modulation is shown, as indicated in part by sustained pressure over an extended period. Figure 34B An ultrasound sequence with a long total duration is shown that is suitable for non-real-time (e.g., offline) neural modulation. Figure 34C Peak negative focal pressure in MPa is shown as a function of distance in cm. The phased array of the ultrasound transducer can focus energy to achieve high focal pressure varying with depth at a given attenuation of 0.7 dB / MHz / cm. The ability of the ultrasound transducer to deliver high focal pressure allows for effective neuromodulation of tissue at specific depths, such as those indicating the distance between the implanted ultrasound transducer and the lower regions of the human brain. Figure 34D Normalized pressures of ultrasound waves emitted from an ultrasound transducer configured for neuromodulation of a subject were depicted at different phased array swivel angles. The ultrasound transducer can deliver ultrasound waves, such as ultrasound beams, at swivel angles of -22.5 degrees, 0 degrees, and 22.5 degrees. Normalized pressures were measured based on axial position in mm. Figure 34E The bandwidth under which ultrasound waves can operate from an ultrasound transducer configured for neuromodulation of a subject is depicted. Surface pressures in kPa are described for both Positive Peak Pressure (PPP) and Negative Peak Pressure (PNP) across a transmission frequency range from 2 MHz to 10 MHz. Figure 34E This indicates that a wide range of ultrasonic transmission frequencies can be used to achieve high surface pressure.

[0548] Figures 35A to 35C Provides spatially varying ultrasonic waves emitted from a modified ultrasonic transducer ( For example, a 2D color map of ultrasonic pressure, where space is mapped onto volume. Figures 35A to 35C The data depicted is derived from hydrophone data experimentally collected in the water tank. The color-mapped values ​​represent the pressure values ​​of the delivered ultrasound waves, for example, in Pascals. Figures 35A to 35C The data depicted, when combined, provide a 3D depiction of the spatial distribution of ultrasonic pressure as ultrasound waves are emitted from the ultrasonic transducer. Figure 35A The spatial distribution of ultrasonic pressure in the volume was depicted, with a lateral dimension of 11 mm and a depth dimension of 21 mm. Figure 35B The spatial distribution of ultrasonic pressure in the volume was depicted, with a height of 11 mm and a depth of 21 mm. Figure 35C The spatial distribution of ultrasonic pressure within the volume was depicted, with a lateral dimension of 11 m and an elevation dimension of 11 mm. Let's take a look. Figures 35A to 35C This indicates that the spatial distribution of the highest pressure value of ultrasound is similar to a rectangular 3D sheet structure, where the 'thickness' of the sheet can be viewed across the lateral dimension, the 'length' of the sheet can be viewed across the depth dimension, and the 'width' of the sheet structure can be viewed across the elevation dimension. Figures 35A to 35C The hydrophone data shown not only provides a feasibility analysis for ultrasound used for neuromodulation of subjects, but also for ultrasound used for imaging subjects.

[0549] Figures 36A to 36C Simulated data are provided for one or more ultrasonic pulses that vary spatially during delivery by an implanted ultrasonic transducer device. For Figures 36A to 36C Each of the attached figures shows the spatial peak pulse average of the ultrasound in dB. ISPPA or I_sppa The spatial distribution of intensity. Figures 36A to 36CThe simulation data shown not only provide a feasibility analysis for ultrasound-based neuromodulation of subjects, but also for ultrasound-based imaging of subjects. Figure 36A The spatial distribution of ISPPA intensity across the y and z length dimensions is depicted. Figure 36B The spatial distribution of ISPPA intensity across x and z length dimensions is depicted. Figure 36C The spatial distribution of ISPPA intensity across length dimensions y and x is depicted.

[0550] Example 7 - Correlating Subject Stimulus Responses with Ultrasound-Based Neural Activity Figure 37A An activity graph was depicted by ultrasound imaging of a coronal section in the brain of a female Long-Evans rat while visual stimuli were presented to the rat. Figure 37A The results demonstrate the illustrative advantages of the methods and systems disclosed herein over conventional neurotechniques for imaging neural activity. The brain regions indicated by pixels on the activity map illustrate the correlation between visual stimulation and ultrasound imaging in the subject. As illustrated, the x and y axes in the figure are in millimeters, and they depict the ultrasound imaging regions of interest in the subject's brain. Color bars describe statistical t-scores. A generalized linear model (GLM) was used to ...

Claims

1. An implantable transducer, comprising: a housing; an acoustically transparent window disposed at least partially at a first end of the housing; and an ultrasound array disposed within the housing proximate the first end, the ultrasound array configured to emit ultrasound waves through the acoustically transparent window to an external environment. one or more circuit boards disposed within the housing, the one or more circuit boards comprising one or more electronic components disposed thereon, the one or more electronic components configured to communicate one or more signals to the ultrasound array.

2. The implantable transducer of claim 1, further comprising:

3. The implantable transducer of claim 2, wherein the one or more electronic components disposed on the circuit board are configured to process data received from the ultrasound array, the data indicative of brain function in a subject.

4. The implantable transducer of claim 2 or 3, wherein the data comprises image data indicative of anatomical features of a subject.

5. The implantable transducer of any one of claims 1 to 4, wherein the implantable transducer is configured to be disposed in a hole in a skull of a subject.

6. The implantable transducer of any one of claims 1 to 5, wherein the implantable transducer is positioned in contact with soft tissue of a subject.

7. The implantable transducer of claim 5 or 6, wherein the implantable transducer is configured to elevate local body temperature by less than 2 °C.

8. The implantable transducer of claim 5 or 6, wherein the implantable transducer is configured to limit absolute local brain temperature to less than 39 °C.

9. The implantable transducer of claim 5 or 6, wherein the implantable transducer is positioned in contact with the dura mater of the subject.

10. The implantable transducer of claim 5 or 6, wherein the implantable transducer is located outside the brain parenchyma of the subject.

11. The implantable transducer of any one of claims 1 to 10, wherein the housing comprises a lip disposed at a second end of the housing, the lip configured to be mounted onto an outer surface of a skull of a subject.

12. The implantable transducer of any one of claims 1 to 11, wherein the implantable transducer comprises a cable configured to transmit power or data to or from the implantable transducer.

13. The implantable transducer of claim 12, wherein the cable extends out through the housing of the implantable transducer.

14. The implantable transducer of any one of claims 1 to 13, wherein the acoustically transparent window comprises a biocompatible polymer.

15. The implantable transducer of claim 14, wherein the biocompatible polymer is polymethyl methacrylate (PMMA), or polyether ether ketone (PEEK), polychlorotrifluoroethylene (PCTFE), polytetrafluoroethylene (PTFE), ultra-high-molecular-weight polyethylene (UHMWPE), polyethylene terephthalate (PET), low-density polyethylene (LDPE), polyether block amide (PEBAX), and / or high-density polyethylene (HDPE). ​ 16. The implantable transducer of any one of claims 1 to 15, wherein the ultrasound array is fabricated on a complementary metal-oxide-semiconductor (CMOS) application-specific integrated circuit (ASIC).

17. The implantable transducer of any one of claims 1 to 16, wherein the ultrasound array comprises a capacitive micromachined ultrasonic transducer (CMUT), a piezoelectric micromachined ultrasonic transducer (PMUT) array, or a lead zirconate titanate (PZT) array.

18. The implantable transducer of any one of claims 1 to 17, wherein the implantable transducer is configured to be coupled to one or more electrical wires, the implantable transducer being configured to transmit data through the one or more electrical wires and further configured to receive data through the one or more electrical wires.

19. The implantable transducer of any one of claims 1 to 18, wherein the implantable transducer is configured to receive a plurality of ultrasound waves.

20. The implantable transducer of any one of claims 1 to 19, wherein the ultrasound array comprises a plurality of transducer elements.

21. The implantable transducer of claim 20, wherein the plurality of transducer elements comprises 100 to 199, 200 to 399, 400 to 999, 1,000 to 1,499, 1,500 to 9,999, 10,000 to 11,999, 12,000 to 99,000, or 100,000 to 120,000 transducer elements.

22. The implantable transducer of claim 20, wherein the ultrasound array comprises an n x m matrix, where n is in a range of 16 to 256 transducer elements, and m is in a range of 1 to 256 transducer elements.

23. A system for monitoring or modulating a physiological activity of a subject, comprising: one or more implantable transducers, wherein an implantable transducer of the one or more implantable transducers corresponds to the implantable transducer of any one of claims 1 to 22; and a controller coupled to each implantable transducer of the one or more implantable transducers, the controller comprising a power source and a processor, wherein the power source is configured to power each implantable transducer of the one or more implantable transducers, and wherein the processor is configured to perform a method comprising: transmitting one or more signals to the one or more implantable transducers; and receiving data from the one or more implantable transducers.

24. The system of claim 23, wherein the method further comprises: emitting ultrasound waves by the one or more implantable transducers, wherein the ultrasound waves are configured to modify the physiological activity of the subject.

25. The system of claim 23 or 24, wherein the one or more signals are configured to specify an amplitude or timing of one or more transducer elements of the plurality of transducer elements. between one and ten implantable transducers. ​ 26. The system of any one of claims 1 to 25, comprising: ​ 27. The system of any one of claims 1 to 26, further comprising: a remote center configured to receive the data from the controller and further configured to send external data to the controller.

28. The system of claim 27, wherein the remote center is configured to communicate with a display to provide a user interface for controlling the system.

29. The system of any one of claims 1 to 28, wherein the implantable transducer and the controller are configured to communicate wirelessly.

30. The system of claim 29, wherein the implantable transducer and the controller are configured to communicate through Bluetooth, Bluetooth Low Energy, WiFi, or a combination thereof.

31. The system of any one of claims 23 to 30, wherein the one or more signals are configured to coordinate transmission of ultrasound waves by the implantable transducer and further configured to coordinate reception of the ultrasound waves.

32. The system of any one of claims 23 to 31, wherein the controller comprises a clock, and wherein the one or more signals are transmitted based on a predetermined interval associated with the clock.

33. The system of claim 32, wherein the controller comprises a central clock, and the one or more implantable transducers each comprise the clock, and wherein the signals correspond to a reset signal associated with the central clock.

34. The system of any one of claims 1 to 33, wherein the subject is performing a clinically relevant behavior while the implantable transducer obtains data.

35. The system of claim 34, wherein the clinically relevant behavior comprises an activity of daily living, motion estimation, motion capture, facial expression and reaction time, self-reported mood, self-reported cognitive state, heart rate, heart rate variability, respiratory rate, oxygenation, galvanic skin response, inertial monitoring, or a combination thereof.

36. The system of any one of claims 21 to 35, wherein the system is configured to modify the physiological activity of the subject based on the data.

37. The system of claim 36, wherein modifying the physiological activity of the subject based on the data occurs in real time.

38. The system of claim 36, wherein real time includes a reaction time of 5 seconds or less after the data is received.

39. The system of claim 36, wherein modifying the physiological activity of the subject occurs at regular predetermined intervals.

40. The system of any one of claims 21 to 39, wherein the physiological activity of the subject is neural activity.

41. The system of claim 40, wherein the neural activity of the subject is neural activity of the central nervous system.

42. The system of claim 40, wherein the neural activity of the subject is neural activity of the brain.

43. The system of claim 42, wherein the neural activity of the brain comprises neural activity from a distributed neural network of the brain.

44. The system of any one of claims 1-43, wherein the subject has or is suspected of having a neurological dysfunction.

45. The system of claim 44, wherein the neurological dysfunction is clinical depression, clinical anxiety, neuropathic pain, or a combination thereof.

46. A method for monitoring a physiological activity of a subject, the method comprising: transmitting, by a controller, one or more signals to one or more implantable transducers, wherein the controller is located remotely from the one or more implantable transducers, and wherein the one or more implantable transducers are mounted to the skull of the subject; and receiving data from the one or more implantable transducers.

47. The method of claim 46, wherein the method further comprises: emitting, by the one or more implantable transducers, ultrasound waves based on the one or more signals, wherein the ultrasound waves are configured to modify the physiological activity of the subject.

48. The method of claim 46 or 47, wherein the method further comprises: modifying a physiological activity of the subject based on the ultrasound waves.

49. A method for determining instructions for modulating neural activity of a nervous system of a subject, comprising: receiving ultrasound data of the nervous system from an implantable transducer, wherein the ultrasound data is indicative of a physiological state of the nervous system; processing the ultrasound data of the nervous system; and sending processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data.

50. The method of claim 49, wherein modulating the neural activity comprises: stimulating one or more regions of the nervous system.

51. The method of claim 49 or 50, wherein: stimulating the one or more regions of the nervous system comprises electrically stimulating through one or more electrodes, and the instructions for modulating the neural activity comprise instructions for controlling the electrical stimulation through the one or more electrodes.

52. The method of any one of claims 49-51, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the method further comprises: receiving, by the implantable transducer, the instructions for modulating the neural activity; and performing, by the implantable transducer, the ultrasound neuromodulation.

53. The method of any one of claims 49-52, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the method further comprises: receiving, by the implantable transducer, the ultrasound neuromodulation.

54. A method for determining instructions for modulating neural activity of a nervous system of a subject, comprising: receiving ultrasound data of the nervous system from an implantable transducer of any one of claims 1-53, wherein the ultrasound data is indicative of a physiological state of the nervous system; processing the ultrasound data of the nervous system; sending processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data; receiving, by the implantable transducer, the instructions for modulating the neural activity; and performing, by the implantable transducer, the ultrasound neuromodulation on the subject.

55. A method of assembling an implantable transducer, comprising: placing an ultrasound array as claimed in any one of claims 1 to 54 proximate to a sound transparent window; and joining the housing to the placed ultrasound array, wherein the placed ultrasound array is at least partially disposed in the housing.

56. The method as claimed in claim 55, wherein the housing comprises one or more housing components.

57. The method of claim 55 or 56, wherein joining the housing to the placed ultrasound array comprises: joining the one or more of the housing components to the placed ultrasound array.

58. A method of assembling an implantable transducer as claimed in any one of claims 1 to 57, comprising: casting a one-piece sound transparent housing, wherein the one-piece sound transparent housing comprises a sound transparent window and an ultrasound array, and the ultrasound array is disposed within the cast one-piece sound transparent housing.

59. A method of training a machine learning model, comprising: receiving one or more ultrasound data of one or more samples from one or more subjects obtained from an implantable transducer as claimed in any one of claims 1 to 58, and one or more functional ultrasound image data corresponding to the one or more ultrasound data; converting the one or more ultrasound data to one or more ultrasound arrays; converting the one or more functional ultrasound image data to one or more functional ultrasound arrays; and training the machine learning model with the one or more ultrasound arrays and the one or more functional ultrasound arrays to predict one or more inferred functional ultrasound arrays from an inputted one or more ultrasound data or an inputted one or more ultrasound arrays.

60. The method as claimed in claim 59, wherein the machine learning model is retrained one or more training iterations based on one or more additional ultrasound data, one or more additional ultrasound arrays, one or more additional functional ultrasound image data, or one or more additional functional ultrasound arrays.

61. The method of claim 60, wherein the retraining of the machine learning model comprises: tuning the machine learning model based on the one or more additional ultrasound data, the one or more additional ultrasound arrays, the one or more additional functional ultrasound image data, or the one or more additional functional ultrasound arrays.

62. A method of inferring a functional ultrasound array from one or more ultrasound data, comprising: receiving the one or more ultrasound data of one or more samples from one or more subjects; converting the one or more ultrasound data to one or more ultrasound arrays; providing the one or more ultrasound arrays to a trained machine learning model; and outputting one or more inferred functional ultrasound arrays based on the received one or more ultrasound data.

63. The method of claim 62, further comprising: converting the one or more inferred functional ultrasound arrays to one or more inferred functional ultrasound image data.

64. The method as claimed in any one of claims 59 to 63, wherein the one or more functional ultrasound arrays comprise power Doppler images.

65. The method as claimed in any one of claims 59 to 64, wherein the one or more ultrasound data comprise ultrasound data sequences.

66. The method of any one of claims 59-65, wherein the ultrasound data comprises radio frequency data, in-phase and quadrature (IQ) data, B-mode image data, composite image data, or any combination thereof.

67. The method of any one of claims 59-66, wherein the ultrasound data comprises radio frequency data, in-phase and quadrature (IQ) data, composite image data, or any combination thereof, but does not comprise B-mode image data.

68. A method for determining instructions for modulating neural activity of a nervous system of a subject, comprising: receiving ultrasound data of the nervous system from an implantable transducer, wherein the ultrasound data is indicative of a physiological state of the nervous system; processing the ultrasound data of the nervous system; and sending processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data.

69. The method of claim 68, wherein a region of the neural activity being modulated is determined based on the ultrasound data. receiving one or more of data associated with the physiological state and data associated with the neural activity.

70. The method of claim 68 or 69, further comprising:

71. The method of any one of claims 68-70, wherein the physiological state comprises a neurophysiological state.

72. The method of claim 71, wherein the neurophysiological state comprises a hemodynamic activity.

73. The method of claim 72, wherein the hemodynamic activity is indicated by a power Doppler intensity associated with the ultrasound data.

74. The method of claim 72 or 73, wherein the hemodynamic activity comprises a cerebral blood volume (CBV) activity, and a change in the CBV activity is directly proportional to a change in the power Doppler intensity. stimulating one or more regions of the nervous system.

75. The method of any one of claims 68-74, wherein modulating the neural activity comprises:

76. The method of claim 75, wherein the one or more regions of the nervous system comprise one or more regions of a peripheral nervous system, one or more regions of a central nervous system, or a combination thereof.

77. The method of claim 76, wherein the one or more regions of the central nervous system comprise a brain.

78. The method of any one of claims 75-77, wherein: stimulating the one or more regions of the nervous system comprises electrically stimulating through one or more electrodes, and the instructions for modulating the neural activity comprise instructions for controlling the electrical stimulation through the one or more electrodes.

79. The method of claim 78, wherein the electrical stimulation is controlled through electrical modulation parameters comprising an amplitude, a frequency, a pulse width, an intensity, a waveform, a polarity, an acoustic pressure, or any combination thereof. ​ 80. The method of claim 78 or 79, wherein the electrical stimulation comprises deep brain stimulation (DBS), transcranial magnetic stimulation (TMS), repetitive TMS (rTMS), vagus nerve stimulation (VNS), transcranial direct current stimulation (tDCS), electrocorticography (ECoG), or any combination thereof.

81. The method of any one of claims 75-80, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the method further comprises: receiving, by the implantable transducer, the instructions for modulating the neural activity; and performing, by the implantable transducer, the ultrasound neuromodulation.

82. The method of any one of claims 68-81, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the method further comprises: receiving, by the implantable transducer, the ultrasound neuromodulation.

83. The method of any one of claims 68-82, wherein the method is performed over a period of seconds, minutes, hours, days, weeks, months, or years.

84. The method of any one of claims 68-83, wherein the instructions for modulating the neural activity are associated with a longitudinal therapy or longitudinal study.

85. The method of any one of claims 68 to 84, wherein: the instructions for modulating the neural activity are determined further based on pre-trial physiological state information.

86. The method of claim 85, wherein the pre-trial physiological state information comprises pre-trial ultrasound information, functional magnetic resonance imaging (fMRI) information, electrophysiological recordings, structural magnetic resonance imaging scans, diffusion tensor imaging (DTI) information, computed tomography (CT) scan information, or any combination thereof.

87. The method of any one of claims 68-86, wherein the instructions for modulating the neural activity are determined based on an output of a machine learning algorithm.

88. The method of claim 87, wherein the output of the machine learning algorithm is based on the ultrasound data provided to the machine learning algorithm.

89. The method of claim 87 or 88, wherein the machine learning algorithm is trained by pre-trial physiological state information.

90. The method of claim 89, wherein the pre-trial physiological state information comprises ultrasound information, functional magnetic resonance imaging (fMRI) information, electrophysiological recordings, structural magnetic resonance imaging scans, diffusion tensor imaging (DTI) information, computed tomography (CT) scan information, or any combination thereof.

91. The method of any one of claims 87-90, wherein the machine learning algorithm comprises reinforcement learning, Bayesian optimization, generalized linear models, support vector machines, deep neural networks, or any combination thereof.

92. The method of any one of claims 87-91, wherein the machine learning algorithm is trained offline, tested offline, validated offline, or any combination thereof.

93. The method of any one of claims 68-92, wherein the ultrasound data comprises radio frequency (RF) data or in-phase and quadrature (IQ) data.

94. The method of any one of claims 68-93, wherein the ultrasound data comprises one or more ultrasound images.

95. The method of claim 94, wherein the one or more ultrasound images comprise two- dimensional images, three-dimensional images, or any combination thereof.

96. The method of claim 95, wherein the one or more ultrasound images have a resolution of 100 pm to 4 mm.

97. The method of any one of claims 94-96, wherein an imaging volume of the one or more ultrasound images comprises a spherical sector having a conical radius.

98. The method of claims 94-97, wherein the one or more ultrasound images are received at 10 Hz to 257 kHz.

99. The method of any one of claims 68-98, wherein the instructions for modulating the neural activity are determined based on target neural activity.

100. The method of claim 99, wherein the target neural activity is determined by ultrasound imaging, fMRI imaging, electrophysiological recording, structural magnetic resonance imaging scan, diffusion tensor imaging (DTI), or any combination thereof.

101. The method of claim 99, wherein the target neural activity is determined by the ultrasound data.

102. The method of any one of claims 99-101, wherein the target neural activity is determined based on an output of a transfer learning algorithm.

103. The method of any one of claims 99-102, wherein the target neural activity is represented as a composite time-independent state.

104. The method of any one of claims 99-103, wherein the target neural activity is represented as multi-dimensional time-series data.

105. The method of claim 104, wherein a temporal resolution or a spatial resolution of the multi-dimensional time-series data is equal to or less than a temporal resolution or a spatial resolution of the ultrasound data.

106. The method of any one of claims 68-105, further comprising: receiving second ultrasound data of the nervous system from the implantable transducer, wherein the second ultrasound data is indicative of a second physiological state of the nervous system; processing the second ultrasound data; and transmitting the processed second ultrasound data, wherein second instructions for modulating a second neural activity of the nervous system are determined based on the processed second ultrasound data.

107. The method of claim 106, wherein the second instructions for modulating the second neural activity comprise adjusted first instructions for modulating the first neural activity.

108. The method of claim 107, wherein adjusting the first instructions for modulating the first neural activity comprises adjusting an electrical modulation parameter, a spatial modulation parameter, a temporal modulation parameter, or any combination thereof. ​ 109. The method of claim 108, wherein the electrical modulation parameters comprise amplitude, frequency, pulse width, intensity, waveform, polarity, acoustic pressure, or any combination thereof.

110. The method of claim 108 or 109, wherein the spatial modulation parameters comprise electrode configuration, electrode position, electrode size, electrode placement, directivity, coil orientation, coil position, stimulation focus, stimulation bilaterality, montage, focus size, target location, or any combination thereof.

111. The method of any one of claims 108 to 110, wherein the temporal modulation parameters comprise burst, cycle, ramp, frequency, pulse duration, train duration, inter-train interval, total number of pulses, stimulation pattern, duration, inter-stimulation interval, session frequency, pulse repetition frequency, duty cycle, or any combination thereof.

112. The method of any one of claims 68, further comprising: iteratively performing the receiving step, the processing step, and the sending step, wherein: the instructions for modulating the respective neural activity of the nervous system are determined based on respective ultrasound data, and the method stops after a predetermined number of iterations is reached.

113. The method of any one of claims 68 to 112, further comprising: iteratively performing the receiving step, the processing step, and the sending step, wherein: the instructions for modulating the respective neural activity of the nervous system are determined based on respective ultrasound data, and the method stops after a predetermined number of iterations is reached.

114. The method of claim 113, wherein the method stops in accordance with a determination that the subject exhibits target neural activity for at least a predetermined duration.

115. The method of any one of claims 68 to 114, further comprising: receiving, from the implantable transducer, second ultrasound data of the nervous system in response to modulating the neural activity.

116. The method of any one of claims 68 to 115, further comprising: associating the second ultrasound data of the nervous system with modulated neural activity.

117. The method of claim 115 or 116, wherein the second ultrasound data of the nervous system is received a predetermined period of time after modulating the neural activity.

118. The method of any one of claims 106 to 117, wherein: the neural activity is modulated at a first region of the nervous system, and in response to modulating the first region of the nervous system, determining second instructions for modulating a second neural activity of the nervous system at a second region of the nervous system.

119. The method of any one of claims 106 to 118, wherein the second instructions are determined a predetermined period of time after modulating the neural activity.

120. The method of any one of claims 106 to 119, wherein the second instructions are executed a second predetermined period of time after determining the second instructions.

121. The method of any one of claims 68 to 120, wherein determining the instructions for modulating the neural activity comprises: determining, based on the ultrasound data, a region of the nervous system for the modulation.

122. The method of any one of claims 68 to 121, wherein the instructions for modulating the region of the neural activity are further based on a second physiological state.

123. The method of claim 122, wherein the second physiological state is received with the first physiological state of the nervous system.

124. The method of claim 122 or 123, wherein the second physiological state comprises a behavior of the subject, an ocular measurement of the subject, a hematological measurement of the subject, or any combination thereof.

125. The method of claim 124, wherein the behavior of the subject is determined based on the subject’s responses to a survey questionnaire, an emotional assessment, or both.

126. The method of claim 124 or 125, wherein the ocular measurement of the subject comprises an eye tracking or a pupil dilation measurement.

127. The method of any one of claims 124 to 126, wherein the hematological measurement of the subject comprises a blood pressure, a blood sugar level, a blood cholesterol level, a blood hormone level, or any combination thereof.

128. The method of any one of claims 122 to 127, wherein the second physiological state is determined by a camera, a microphone, a wearable device, or any combination thereof.

129. The method of claim 128, wherein the wearable device comprises an electronic watch, an electronic ring, or electronic glasses.

130. The method of any one of claims 122 to 129, wherein the second physiological state is associated with a positive valence or a negative valence.

131. The method of claim 130, wherein the positive valence or the negative valence is determined based on pre-test physiological state observations, the ultrasound data, or both.

132. The method of claim 130 or 131, wherein the positive valence or the negative valence is determined experimentally.

133. The method of any one of claims 130 to 132, wherein the positive valence or the negative valence is used in part to determine a target neural activity.

134. The method of any one of claims 68 to 133, wherein modulating the neural activity is associated with treating chronic pain, depression and anxiety, obsessive-compulsive disorder, Parkinson’s disease, essential tremor, epilepsy, post-traumatic stress disorder, memory impairment, or any combination thereof.

135. The method of claim 134, wherein the obsessive-compulsive disorder is obsessive-compulsive disorder, substance abuse disorder, or both.

136. The method of any one of claims 68 to 135, wherein the subject is a human.

137. The method of any one of claims 68 to 136, wherein the instructions for modulating the neural activity are sent to a neural modulation system through a docking device.

138. The method of any one of claims 68 to 137, wherein the instructions for modulating the neural activity are sent to a neural modulation system through a communication protocol.

139. The method of claim 138, wherein the communication protocol comprises USB.

140. A method for determining instructions for modulating neural activity of a nervous system of a subject, comprising: receiving ultrasound data of the nervous system from an implantable transducer, wherein the ultrasound data is indicative of a physiological state of the nervous system; processing the ultrasound data of the nervous system; transmitting the processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data; and receiving, by the implantable transducer, the instructions for modulating the neural activity; and performing, by the implantable transducer, the ultrasound neuromodulation on the subject.

141. The method of any one of claims 68-140, wherein the implantable transducer comprises: a housing; an acoustically transparent window disposed at least partially at a first end of the housing; and an ultrasound array disposed within the housing proximate to the first end, the ultrasound array configured to emit ultrasound waves to an external environment through the acoustically transparent window.

142. A method of assembling an implantable transducer, comprising: placing an ultrasound array proximate to an acoustically transparent window; and joining the housing to the placed ultrasound array, wherein the placed ultrasound array is disposed at least partially in the housing.

143. The method of claim 142, wherein the housing comprises one or more housing components.

144. The method of claim 142 or 143, wherein joining the housing to the placed ultrasound array comprises: joining the one or more of the housing components to the placed ultrasound array.

145. A method of assembling the implantable transducer of any one of claims 68-144, comprising: casting a one-piece acoustically transparent housing, wherein the one-piece acoustically transparent housing comprises an acoustically transparent window and an ultrasound array, and the ultrasound array is disposed within the cast one-piece acoustically transparent housing.

146. A method of training a machine learning model, comprising: receiving one or more ultrasound data of one or more samples from one or more subjects obtained from the implantable transducer of any one of claims 68-145, and one or more functional ultrasound image data corresponding to the one or more ultrasound data; converting the one or more ultrasound data to one or more ultrasound arrays; converting the one or more functional ultrasound image data to one or more functional ultrasound arrays; and training the machine learning model with the one or more ultrasound arrays and the one or more functional ultrasound arrays to predict one or more inferred functional ultrasound arrays from an inputted one or more ultrasound data or an inputted one or more ultrasound arrays.

147. The method of claim 146, wherein the machine learning model is retrained one or more training iterations based on one or more additional ultrasound data, one or more additional ultrasound arrays, one or more additional functional ultrasound image data, or one or more additional functional ultrasound arrays.

148. The method of claim 147, wherein the retraining of the machine learning model comprises: tuning the machine learning model based on the one or more additional ultrasound data, the one or more additional ultrasound arrays, the one or more additional functional ultrasound image data, or the one or more additional functional ultrasound arrays.

149. A method of inferring functional ultrasound arrays from one or more ultrasound data, comprising: receiving the one or more ultrasound data for one or more samples from one or more subjects; converting the one or more ultrasound data to one or more ultrasound arrays; providing the one or more ultrasound arrays to a trained machine learning model; and outputting one or more inferred functional ultrasound arrays based on the received one or more ultrasound data.

150. The method of claim 149, further comprising: converting the one or more inferred functional ultrasound arrays to one or more inferred functional ultrasound image data.

151. The method of any one of claims 146-150, wherein the one or more functional ultrasound arrays comprise power Doppler images.

152. The method of any one of claims 146-151, wherein the one or more ultrasound data comprise ultrasound data sequences.

153. The method of any one of claims 146-152, wherein the ultrasound data comprise radio frequency data, in-phase and quadrature (IQ) data, B-mode image data, composite image data, or any combination thereof.

154. The method of any one of claims 146-153, wherein the ultrasound data comprise radio frequency data, in-phase and quadrature (IQ) data, composite image data, or any combination thereof, but do not comprise B-mode image data.

155. An acoustically transparent window adjacent to an ultrasound array, comprising: a biocompatible polymer, wherein the acoustically transparent window is configured to allow transmission of ultrasound waves through the acoustically transparent window.

156. The acoustically transparent window of claim 155, wherein the biocompatible polymer comprises polymethyl methacrylate (PMMA), polyether ether ketone (PEEK), polychlorotrifluoroethylene (PCTFE), polytetrafluoroethylene (PTFE), ultra-high molecular weight polyethylene (UHMWPE), polyethylene terephthalate (PET), low-density polyethylene (LDPE), polyether block amide (PEBAX), and / or high-density polyethylene (HDPE).

157. The acoustically transparent window of any one of claims 155 or 156, wherein the biocompatible polymer has a density greater than or equal to a predetermined lower density and less than or equal to a predetermined higher density.

158. The acoustically transparent window of claim 157, wherein the lower density is about 0.31 g / cm3. 3 .

159. The acoustic window of claim 157 or 158, wherein the higher density is about 2.75 g / cm 3 .

160. The acoustically transparent window of any one of claims 155-159, wherein the biocompatible polymer is configured to allow transmission of ultrasound waves at a speed greater than or equal to a predetermined lower speed of sound and less than or equal to a predetermined higher speed of sound.

161. The acoustically transparent window of claim 160, wherein the predetermined lower speed of sound is approximately 896 meters per second.

162. The acoustically transparent window of claim 160, wherein the predetermined higher speed of sound is approximately 3680 meters per second.

163. The acoustically transparent window of any one of claims 155-162, wherein the biocompatible polymer has an attenuation coefficient greater than or equal to a predetermined lower attenuation coefficient and less than or equal to a predetermined higher attenuation coefficient.

164. The acoustic window of claim 163, wherein the predetermined lower attenuation coefficient is about 0.15 dB / cm / MHz.

165. The acoustic window of claim 163, wherein the predetermined higher attenuation coefficient is about 9.27 dB / cm / MHz.

166. The acoustic window of any one of claims 155 to 165, wherein the impedance of the biocompatible polymer is greater than or equal to a lower impedance and less than or equal to a higher impedance.

167. The acoustic window of claim 166, wherein the lower impedance is about 0.685 MRayl.

168. The acoustic window of claim 166, wherein the higher impedance is about 2.765 MRayl.

169. The acoustic window of any one of claims 155 to 168, wherein the impedance ratio of the biocompatible polymer is greater than or equal to a predetermined lower impedance ratio and less than or equal to a predetermined higher impedance ratio.

170. The acoustic window of claim 169, wherein the predetermined lower impedance ratio is about 0.

625.

171. The acoustic window of claim 169, wherein the predetermined higher impedance ratio is about 2.

765.

172. The acoustic window of any one of claims 155 to 171, wherein the reflection coefficient of the biocompatible polymer is greater than or equal to a lower reflection coefficient and less than or equal to a higher reflection coefficient.

173. The acoustic window of claim 172, wherein the lower reflection coefficient is about 0.

005.

174. The acoustic window of claim 172, wherein the higher reflection coefficient is about 0.

215.

175. The acoustic window of any one of claims 155 to 174, wherein the transmission coefficient of the biocompatible polymer is greater than or equal to a lower transmission coefficient and less than or equal to a higher transmission coefficient.

176. The acoustic window of claim 175, wherein the lower transmission coefficient is about 0.

785.

177. The acoustic window of claim 175, wherein the higher transmission coefficient is about 0.

995.

178. The acoustic window of any one of claims 155 to 177, wherein the total attenuation of the biocompatible polymer at a predetermined frequency is greater than or equal to a predetermined lower total attenuation at the predetermined frequency and less than or equal to a predetermined higher total attenuation at the predetermined frequency.

179. The acoustic window of claim 178, wherein the predetermined lower total attenuation is about 0.01 dB / cm.

180. The acoustic window of claim 178, wherein the predetermined higher total attenuation is about 40.95 dB / cm.

181. The acoustic window of any one of claims 178 to 180, wherein the predetermined frequency is about 5 MHz.

182. The acoustic window of any one of claims 155 to 181, wherein the enclosed ultrasound array is proximate to the acoustic window comprises the enclosed ultrasound array is adjacent to the acoustic window.

183. The acoustic window of any of claims 155-182, wherein an implantable transducer comprises the acoustic window, the ultrasound array, and a housing.

184. The acoustic window of claim 183, wherein the implantable transducer comprises a cable configured to send power or data to or from the implantable transducer.

185. An implantable transducer comprising: a housing; an acoustic window of any of claims 155-184 disposed at least partially at a first end of the housing; and the ultrasound array disposed within the housing proximate the first end, the ultrasound array configured to emit ultrasound waves through the acoustic window to an external environment.

186. A method for determining instructions for modulating neural activity of a nervous system of a subject, comprising: receiving ultrasound data of the nervous system from an implantable transducer comprising an acoustic window of any of claims 155-185, wherein the ultrasound data is indicative of a physiological state of the nervous system; processing the ultrasound data of the nervous system; and sending processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data. stimulating one or more regions of the nervous system.

187. The method of claim 186, wherein modulating the neural activity comprises:

188. The method of claim 186 or 187, wherein: stimulating the one or more regions of the nervous system comprises electrically stimulating through one or more electrodes, and the instructions for modulating the neural activity comprise instructions for controlling the electrical stimulation through the one or more electrodes.

189. The method of any of claims 186-188, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the method further comprises: receiving, by the implantable transducer, the instructions for modulating the neural activity; and performing, by the implantable transducer, the ultrasound neuromodulation.

190. The method of any of claims 186-189, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the method further comprises: receiving, by the implantable transducer, the ultrasound neuromodulation.

191. A method for determining instructions for modulating neural activity of a nervous system of a subject, comprising: receiving ultrasound data of the nervous system from an implantable transducer comprising an acoustic window of any of claims 155-190, wherein the ultrasound data is indicative of a physiological state of the nervous system; processing the ultrasound data of the nervous system; sending processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data; receiving, by the implantable transducer, the instructions for modulating the neural activity; and performing, by the implantable transducer, the ultrasound neuromodulation on the subject. ​ 192. A method of training a machine learning model, comprising: receiving one or more ultrasound data from one or more samples of one or more subjects obtained from an implantable transducer comprising a transonic window as in any of claims 155 to 191, and one or more functional ultrasound image data corresponding to the one or more ultrasound data; converting the one or more ultrasound data to one or more ultrasound arrays; converting the one or more functional ultrasound image data to one or more functional ultrasound arrays; and training the machine learning model with the one or more ultrasound arrays and the one or more functional ultrasound arrays to predict one or more inferred functional ultrasound arrays from an inputted one or more ultrasound data or an inputted one or more ultrasound arrays.

193. The method of claim 192, wherein the machine learning model is retrained one or more training iterations based on one or more additional ultrasound data, one or more additional ultrasound arrays, one or more additional functional ultrasound image data, or one or more additional functional ultrasound arrays.

194. The method of claim 193, wherein the retraining of the machine learning model comprises: tuning the machine learning model based on the one or more additional ultrasound data, the one or more additional ultrasound arrays, the one or more additional functional ultrasound image data, or the one or more additional functional ultrasound arrays.

195. A method of inferring a functional ultrasound array from one or more ultrasound data, comprising: receiving the one or more ultrasound data from one or more samples of one or more subjects; converting the one or more ultrasound data to one or more ultrasound arrays; providing the one or more ultrasound arrays to a trained machine learning model; and outputting one or more inferred functional ultrasound arrays based on the received one or more ultrasound data.

196. The method of claim 195, further comprising: converting the one or more inferred functional ultrasound arrays to one or more inferred functional ultrasound image data.

197. The method of any of claims 192 to 196, wherein the one or more functional ultrasound arrays comprise power Doppler images.

198. The method of any of claims 192 to 197, wherein the one or more ultrasound data comprise ultrasound data sequences.

199. The method of any of claims 192 to 198, wherein the ultrasound data comprise radio frequency data, in-phase and quadrature (IQ) data, B-mode image data, composite image data, or any combination thereof.

200. The method of any of claims 192 to 199, wherein the ultrasound data comprise radio frequency data, in-phase and quadrature (IQ) data, composite image data, or any combination thereof, but not B-mode image data.

201. A method of assembling an implantable transducer, comprising: placing an ultrasound array proximate to a transonic window; and joining the housing to the placed ultrasound array, wherein the placed ultrasound array is at least partially disposed in the housing.

202. The method of claim 201, wherein the housing comprises one or more housing components.

203. The method of claim 201 or 202, wherein joining the housing to the placed ultrasound array comprises: joining the one or more of the housing components to the placed ultrasound array.

204. The method of any one of claims 201 to 203, wherein the assembling or the joining comprises using a bonding method.

205. The method of claim 204, wherein the bonding method comprises laser welding, electron beam welding, TIG welding, heat welding, epoxy sealing, or a combination thereof.

206. A method of assembling an implantable transducer, comprising: casting a one-piece acoustic housing, wherein the one-piece acoustic housing comprises an acoustic window and an ultrasound array, and the ultrasound array is disposed within the cast one-piece acoustic housing.

207. The method of any one of claims 201 to 206, wherein the acoustic housing or the acoustic window comprises a biocompatible polymer.

208. The method of claim 207, wherein the biocompatible polymer comprises polymethyl methacrylate (PMMA), polyether ether ketone (PEEK), polychlorotrifluoroethylene (PCTFE), polytetrafluoroethylene (PTFE), ultra-high-molecular-weight polyethylene (UHMWPE), polyethylene terephthalate (PET), low-density polyethylene (LDPE), polyether block amide (PEBAX), high-density polyethylene (HDPE), or any combination thereof.

209. The method of any one of claims 201 to 208, wherein the housing comprises an acoustic material that is not an acoustic window.

210. The method of any one of claims 201 to 209, wherein the housing comprises a non-acoustic material.

211. The method of any one of claims 201 to 210, wherein the assembling comprises assembling the implantable transducer in a dry gas environment.

212. The method of any one of claims 201 to 211, wherein the assembling comprises sterilizing the implantable transducer.

213. The method of claim 212, wherein the sterilizing comprises gamma irradiation, autoclaving, ethylene oxide treatment, and / or a combination thereof.

214. The implantable transducer of any one of claims 201 to 213, wherein the implantable transducer comprises: the housing; the acoustic window, the acoustic window being disposed at least partially at the first end of the housing; and the ultrasound array, the ultrasound array being disposed within the housing proximate to the first end, the ultrasound array being configured to emit ultrasound waves through the acoustic window to an external environment.

215. A method for determining instructions for modulating neural activity of a nervous system of a subject, comprising: receiving ultrasound data of the nervous system from the implantable transducer of any one of claims 201 to 214, wherein the ultrasound data is indicative of a physiological state of the nervous system; processing the ultrasound data of the nervous system; and and transmitting the processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data.

216. The method of claim 215, wherein modulating the neural activity comprises: stimulating one or more regions of the nervous system.

217. The method of claim 215 or 216, wherein: stimulating the one or more regions of the nervous system comprises electrical stimulation by one or more electrodes, and the instructions for modulating the neural activity comprise instructions for controlling the electrical stimulation by the one or more electrodes.

218. The method of any one of claims 215 to 217, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the method further comprises: receiving, by the implantable transducer, the instructions for modulating the neural activity; and performing, by the implantable transducer, the ultrasound neuromodulation.

219. The method of any one of claims 215 to 218, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the method further comprises: receiving, by the implantable transducer, the ultrasound neuromodulation.

220. A method for determining instructions for modulating neural activity of a nervous system of a subject, comprising: receiving, from an implantable transducer of any one of claims 201 to 219, ultrasound data of the nervous system, wherein the ultrasound data is indicative of a physiological state of the nervous system; processing the ultrasound data of the nervous system; transmitting the processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data; receiving, by the implantable transducer, the instructions for modulating the neural activity; and performing, by the implantable transducer, the ultrasound neuromodulation on the subject.

221. A method of training a machine learning model, comprising: receiving anatomical ultrasound images of one or more samples from one or more subjects obtained from an implantable transducer of any one of claims 201 to 220, and one or more functional ultrasound images corresponding to the one or more anatomical ultrasound images; converting the anatomical ultrasound images into an anatomical ultrasound matrix; converting the one or more functional ultrasound images into one or more functional ultrasound matrices; and training the machine learning model with the anatomical ultrasound matrix and the functional ultrasound matrices to predict an inferred functional ultrasound matrix from an input anatomical ultrasound image or an input anatomical ultrasound matrix.

222. The method of claim 221, wherein the machine learning model is retrained one or more training iterations based on additional anatomical ultrasound imaging, additional anatomical ultrasound matrices, additional functional ultrasound images, or additional functional ultrasound matrices. fine-tuning the machine learning model based on the additional anatomical ultrasound imaging, the additional anatomical ultrasound matrices, the additional functional ultrasound images, or the additional functional ultrasound matrices.

223. The method of claim 222, wherein the retraining of the machine learning model comprises: ​ 224. A method of inferring one or more functional ultrasound images from one or more anatomical ultrasound images, comprising: receiving the one or more anatomical ultrasound images of one or more samples from one or more subjects; converting the one or more anatomical ultrasound images into one or more anatomical ultrasound matrices; providing the one or more anatomical ultrasound matrices to a trained machine learning model, wherein the trained machine learning model is trained according to the method of training the machine learning model of any one of claims [-3] to [-1]; and outputting one or more inferred functional ultrasound matrices based on the received one or more anatomical ultrasound images.

225. The method of claim 224, further comprising: converting the inferred functional ultrasound matrices into inferred functional ultrasound images.

226. The method of any one of claims 221 to 225, wherein the functional ultrasound images comprise power Doppler images.

227. The method of any one of claims 221 to 226, wherein the anatomical ultrasound images comprise a sequence of anatomical ultrasound images.

228. The method of any one of claims 221 to 227, wherein the anatomical ultrasound images comprise radiofrequency images, in-phase and quadrature (IQ) images, B-mode images, composite images, or any combination thereof.

229. The method of any one of claims 221 to 228, wherein the anatomical ultrasound images comprise radiofrequency images and / or in-phase and quadrature (IQ) images, but not B-mode images.

230. A method of training a machine learning model, comprising: receiving one or more ultrasound data of one or more samples from one or more subjects obtained from an implanted transducer, and one or more functional ultrasound image data corresponding to the one or more ultrasound data; converting the one or more ultrasound data into one or more ultrasound arrays; converting the one or more functional ultrasound image data into one or more functional ultrasound arrays; and training the machine learning model with the one or more ultrasound arrays and the one or more functional ultrasound arrays to predict one or more inferred functional ultrasound arrays from the inputted one or more ultrasound data or the inputted one or more ultrasound arrays.

231. The method of claim 230, wherein the machine learning model is retrained one or more training iterations based on one or more additional ultrasound data, one or more additional ultrasound arrays, one or more additional functional ultrasound image data, or one or more additional functional ultrasound arrays. fine-tuning the machine learning model based on the one or more additional ultrasound data, the one or more additional ultrasound arrays, the one or more additional functional ultrasound image data, or the one or more additional functional ultrasound arrays.

232. The method of claim 231, wherein the retraining of the machine learning model comprises: ​ 233. The method of any one of claims 230-232, wherein training the machine learning model further comprises determining one or more metrics that describe a relationship between the one or more inferred functional ultrasound arrays and the behavioral data of the one or more subjects.

234. The method of claim 233, wherein the one or more metrics comprise a correlation metric, a regression metric, a classification metric, a model performance metric, an information theoretic metric, a temporal metric, or a cross-validated metric.

235. The method of claim 234, wherein the correlation metric comprises a Pearson correlation coefficient, a Spearman rank correlation coefficient, or a canonical correlation coefficient (CCA).

236. The method of claim 234, wherein the regression metric comprises an R-squared metric, an adjusted R-squared metric, a t-statistic from a generalized linear model (GLM), or an f-statistic from a GLM.

237. The method of claim 234, wherein the classification metric comprises a decoding accuracy metric, an accuracy metric, a precision metric, a recall metric, an F1 score metric, an area under the receiver operating characteristic curve (AUC-ROC) metric, an area under the precision-recall curve (AUC-PR) metric, or a confusion matrix metric.

238. The method of claim 234, wherein the model performance metric comprises a mean squared error (MSE) metric, a root mean squared error (RMSE) metric, a mean absolute error (MAE) metric, an explained variance metric, or a log loss metric.

239. The method of claim 234, wherein the information theoretic metric comprises a mutual information metric or a transfer entropy metric.

240. The method of claim 234, wherein the temporal metric comprises a temporal signal-to-noise ratio (tSNR) metric or a detection latency metric.

241. The method of any one of claims 234-240, wherein the cross-validated metric comprises a metric in which the metric is cross-validated.

242. The method of any one of claims 234-241, wherein training the machine learning model comprises jointly optimizing the metric and an error between the one or more inferred functional ultrasound arrays and the one or more functional ultrasound arrays.

243. The method of any one of claims 234-242, wherein the one or more metrics are used as part of a cost function during the training of the machine learning model.

244. The method of claim 243, wherein the cost function comprises a weighted sum of the one or more metrics.

245. The method of claim 244, wherein the weighted sum of the one or more metrics is dynamically adjusted during the training of the machine learning model.

246. The method of any one of claims 233-245, wherein the behavioral data comprises motor data, cognitive task performance data, affective state data, or any combination thereof.

247. The method of claim 246, wherein the motion data is obtained from an accelerometer, gyroscope, or motion capture system.

248. The method of claim 246, wherein the cognitive task performance data is based on reaction time, error rate, or task completion time.

249. The method of claim 246, wherein the affective state data is obtained from physiological signals such as heart rate, galvanic skin response, survey questionnaire, or facial expression.

250. The method of any one of claims 230-249, wherein training the machine learning model comprises using a regularization technique.

251. The method of claim 250, wherein the regularization technique comprises dropout, LI regularization, or L2 regularization.

252. The method of any one of claims 230-251, wherein training the machine learning model comprises a human-in-the-loop technique.

253. The method of claim 252, wherein the human-in-the-loop technique comprises evaluation of the inferred functional ultrasound array by a medical professional.

254. The method of any one of claims 230-253, wherein the machine learning model comprises an attention mechanism.

255. The method of any one of claims 230-254, wherein the ultrasound data or the functional ultrasound image data is subjected to image enhancement.

256. The method of claim 255, wherein the image enhancement comprises deconvolution or application of a super-resolution technique.

257. The method of any one of claims 230-256, wherein the ultrasound data or the functional ultrasound image data of the one or more subjects is paired with clinical metadata corresponding to the one or more subjects.

258. The method of claim 257, wherein the clinical metadata comprises age, gender, or medical history of the subject.

259. A method of inferring a functional ultrasound array from one or more ultrasound data, comprising: receiving the one or more ultrasound data of one or more samples from one or more subjects; converting the one or more ultrasound data to one or more ultrasound arrays; providing the one or more ultrasound arrays to a trained machine learning model; and outputting one or more inferred functional ultrasound arrays based on the received one or more ultrasound data.

260. The method of claim 259, further comprising: converting the one or more inferred functional ultrasound arrays to one or more inferred functional ultrasound image data.

261. The method of any one of claims 230-260, wherein the one or more functional ultrasound arrays comprise power Doppler images.

262. The method of any one of claims 230-261, wherein the one or more ultrasound data comprise ultrasound data sequences.

263. The method of any one of claims 230-262, wherein the ultrasound data comprise radio frequency data, in-phase and quadrature (IQ) data, B-mode image data, composite image data, or any combination thereof.

264. The method of any one of claims 230 to 263, wherein the ultrasound data comprises radio frequency data, in-phase and quadrature (IQ) data, complex image data, or any combination thereof, but does not comprise B-mode image data.

265. The method of any one of claims 230 to 264, wherein the trained machine learning model is trained based on training data comprising the ultrasound data and the functional ultrasound image data.

266. The method of any one of claims 230 to 265, wherein at least a portion of the training data comprises normalized image data or augmented image data.

267. The method of claim 266, wherein the normalized image data comprises color normalized image data.

268. The method of claim 266, wherein the augmented image data comprises image data that has been augmented by removing noise in the image data, increasing contrast of the image data, adjusting brightness of the image data, performing a convolution with an image kernel, and / or a geometric transformation.

269. The method of claim 268, wherein the convolution with an image kernel comprises a convolution with a Gaussian blur kernel, a box blur kernel, an edge detection kernel, a sharpening kernel, a non-sharpening masking kernel, or any combination thereof.

270. The method of claim 269, wherein the geometric transformation comprises an affine transformation, an elastic transformation, a flip, a crop, a grid distortion, an optical distortion, a perspective transformation, a transpose, or any combination thereof.

271. The method of claim 270, wherein the affine transformation comprises a translation, a rotation, a scaling, a shear, or any combination thereof.

272. The method of any one of claims 230 to 271, wherein the training data is divided into a first training data portion, a first test data portion, and a validation data portion.

273. The method of claim 272, wherein the first training data portion comprises 70%, 75%, 80%, 85%, or 90% of the training data, the first test data portion comprises 20%, 18%, 15%, 13%, 10%, or 5% of the training data, and the validation data portion comprises 20%, 18%, 15%, 13%, 10%, or 5% of the training data.

274. The method of claim 272 or 273, wherein the validation data portion comprises one or more training image patches, and the first training data portion comprises all training images except the one or more training images in the validation data portion.

275. The method of any one of claims 230 to 274, wherein the training data is divided into a second training data portion and a second test data portion.

276. The method of claim 275, wherein the second training data portion comprises 60%, 65%, 70%, 75%, or 80% of the training data, and the second test data portion comprises 40%, 35%, 30%, 25%, or 20% of the training data.

277. The method of any one of claims 230-276, wherein the training data is subjected to cross-validation.

278. The method of claim 277, wherein the cross-validation comprises k-fold cross-validation, leave-p-fold cross-validation, leave-one-out cross-validation, stratified k-fold cross-validation, repeated k-fold cross-validation, nested k-fold cross-validation, or Monte Carlo cross-validation.

279. The method of any one of claims 230-278, wherein the machine learning model comprises an encoder-decoder architecture.

280. The method of any one of claims 230-279, wherein the machine learning model comprises a 3D convolutional filter.

281. The method of claim 280, wherein the 3D convolutional filter is configured to extract one or more spatio-temporal features from the one or more ultrasound data, the one or more ultrasound arrays, the one or more functional ultrasound image data, or the one or more functional ultrasound arrays.

282. The method of any one of claims 230-281, wherein the machine learning model comprises a residual block.

283. The method of any one of claims 230-282, wherein the machine learning model comprises a convolutional neural network (CNN).

284. The method of claim 283, wherein the CNN comprises a convolution function, an activation function, a pooling function, or any combination thereof.

285. The method of claim 284, wherein the convolution function comprises convolving a matrix from the input with a kernel.

286. The method of claim 285, wherein the kernel is randomly initialized and is learned through training the neural network.

287. The method of claim 286, wherein the learning comprises backpropagation and optimization.

288. The method of claim 287, wherein the optimization comprises gradient descent, stochastic gradient descent, batch gradient descent, mini-batch gradient descent, Adam optimization, AdaGrad optimization, RMSprop optimization, momentum optimization, or any combination thereof.

289. The method of any one of claims 284-288, wherein the activation function is a rectified linear unit (ReLU) function, a leaky ReLU function, a linear activation function, a non-linear activation function, a sigmoid activation function, or a hyperbolic tangent activation function.

290. The method of any one of claims 284-289, wherein the pooling function is a max-pooling function, an average-pooling function, or an attention-based pooling function.

291. The method of any one of claims 230-290, wherein the machine learning model further comprises a softmax function or an argmax function.

292. An implantable transducer of any one of claims 230-291, comprising: a housing; an acoustic window disposed at least partially at a first end of the housing; and a an ultrasonic array disposed within the housing proximate the first end, the ultrasonic array configured to emit ultrasonic waves through the acoustically transparent window to an external environment.

293. A method for determining instructions for modulating neural activity of a nervous system of one or more subjects as claimed in any of claims 230 to 292, comprising: receiving, from the implantable transducer, ultrasound data of the nervous system, wherein the ultrasound data is indicative of a physiological state of the nervous system; processing the ultrasound data of the nervous system; and transmitting processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data.

294. The method of claim 293, wherein modulating the neural activity comprises: stimulating one or more regions of the nervous system.

295. The method of claim 293 or 294, wherein: stimulating the one or more regions of the nervous system comprises electrical stimulation by one or more electrodes, and the instructions for modulating the neural activity comprise instructions for controlling the electrical stimulation by the one or more electrodes.

296. The method of any of claims 293 to 295, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the method further comprises: receiving, by the implantable transducer, the instructions for modulating the neural activity; and performing, by the implantable transducer, the ultrasound neuromodulation.

297. The method of any of claims 293 to 296, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the method further comprises: receiving, by the implantable transducer, the ultrasound neuromodulation.

298. A method for determining instructions for modulating neural activity of a nervous system of one or more subjects as claimed in any of claims 230 to 297, comprising: receiving, from the implantable transducer, ultrasound data of the nervous system, wherein the ultrasound data is indicative of a physiological state of the nervous system; processing the ultrasound data of the nervous system; transmitting processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data; receiving, by the implantable transducer, the instructions for modulating the neural activity; and performing, by the implantable transducer, the ultrasound neuromodulation on the subject.

299. A method of assembling an implantable transducer as claimed in any of claims 230 to 298, comprising: placing an ultrasonic array proximate an acoustically transparent window; and joining the housing to the placed ultrasonic array, wherein the placed ultrasonic array is at least partially disposed in the housing.

300. The method of claim 299, wherein the housing comprises one or more housing components.

301. The method of claim 299 or 300, wherein joining the housing to the placed ultrasound array comprises: joining the one or more of the housing components to the placed ultrasonic array.

302. A method of assembling an implantable transducer, comprising: Casting an integral acoustic housing, wherein the integral acoustic housing includes an acoustic window and an ultrasound array, and the ultrasound array is disposed within the cast integral acoustic housing.

303. A system for determining instructions for modulating neural activity of a nervous system of a subject, comprising: one or more processors; and a memory communicatively coupled to the one or more processors and configured to store instructions that, when executed by the one or more processors, cause the system to: receive ultrasound data of the nervous system from an implantable transducer, wherein the ultrasound data is indicative of a physiological state of the nervous system; process the ultrasound data of the nervous system; and send processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data.

304. The system of claim 303, wherein modulating the neural activity comprises: stimulate one or more regions of the nervous system.

305. The system of claim 303 or 304, wherein: stimulating the one or more regions of the nervous system comprises electrical stimulation by one or more electrodes, and the instructions for modulating the neural activity comprise instructions for controlling the electrical stimulation by the one or more electrodes.

306. The system of any one of claims 303 to 305, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the system comprises further instructions that, when executed by the one or more processors, cause the system to: receive, by the implantable transducer, the instructions for modulating the neural activity; and perform, by the implantable transducer, the ultrasound neuromodulation.

307. The system of any one of claims 303 to 306, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the system comprises further instructions that, when executed by the one or more processors, cause the system to: receive, by the implantable transducer, the ultrasound neuromodulation.

308. A system for determining instructions for modulating neural activity of a nervous system of a subject, comprising: one or more processors; and a memory communicatively coupled to the one or more processors and configured to store instructions that, when executed by the one or more processors, cause the system to: receive ultrasound data of the nervous system from an implantable transducer, wherein the ultrasound data is indicative of a physiological state of the nervous system; process the ultrasound data of the nervous system; send processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data; receive, by the implantable transducer, the instructions for modulating the neural activity; and perform, by the implantable transducer, the ultrasound neuromodulation of the subject.

309. A system for assembling an implantable transducer, comprising: one or more processors; and a memory communicatively coupled to the one or more processors and configured to store instructions that, when executed by the one or more processors, cause the system to: receive ultrasound data of the nervous system from an implantable transducer, wherein the ultrasound data is indicative of a physiological state of the nervous system; process the ultrasound data of the nervous system; and send processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data. stimulate one or more regions of the nervous system.

305. The system of claim 303 or 304, wherein: stimulating the one or more regions of the nervous system comprises electrical stimulation by one or more electrodes, and the instructions for modulating the neural activity comprise instructions for controlling the electrical stimulation by the one or more electrodes.

306. The system of any one of claims 303 to 305, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the system comprises further instructions that, when executed by the one or more processors, cause the system to: receive, by the implantable transducer, the instructions for modulating the neural activity; and perform, by the implantable transducer, the ultrasound neuromodulation.

307. The system of any one of claims 303 to 306, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the system comprises further instructions that, when executed by the one or more processors, cause the system to: receive, by the implantable transducer, the ultrasound neuromodulation.

308. A system for determining instructions for modulating neural activity of a nervous system of a subject, comprising: one or more processors; and a memory communicatively coupled to the one or more processors and configured to store instructions that, when executed by the one or more processors, cause the system to: receive ultrasound data of the nervous system from an implantable transducer, wherein the ultrasound data is indicative of a physiological state of the nervous system; process the ultrasound data of the nervous system; send processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data; receive, by the implantable transducer, the instructions for modulating the neural activity; and perform, by the implantable transducer, the ultrasound neuromodulation of the subject.

309. A system for assembling an implantable transducer, comprising: one or more processors; and a memory communicatively coupled to the one or more processors and configured to store instructions that, when executed by the one or more processors, cause the system to: position an ultrasound array proximate to a sound transparent window; and couple the housing to the positioned ultrasound array, wherein the positioned ultrasound array is at least partially disposed in the housing.

310. The system of claim 309, wherein the housing comprises one or more housing components.

311. The system of claim 309 or 310, wherein coupling the housing to the placed ultrasound array comprises: couple the one or more of the housing components to the positioned ultrasound array.

312. A system for assembling an implantable transducer, comprising: one or more processors; and a memory communicatively coupled to the one or more processors and configured to store instructions that, when executed by the one or more processors, cause the system to: cast an integral sound transparent housing, wherein the integral sound transparent housing comprises a sound transparent window and an ultrasound array, and the ultrasound array is disposed within the cast integral sound transparent housing.

313. A system, comprising: one or more processors; and a memory communicatively coupled to the one or more processors and configured to store instructions for training a machine learning model, which, when executed by the one or more processors, cause the system to: receive one or more ultrasound data obtained from one or more samples of one or more subjects from an implantable transducer, and one or more functional ultrasound image data corresponding to the one or more ultrasound data; convert the one or more ultrasound data to one or more ultrasound arrays; convert the one or more functional ultrasound image data to one or more functional ultrasound arrays; and train the machine learning model with the one or more ultrasound arrays and the one or more functional ultrasound arrays to predict one or more inferred functional ultrasound arrays from an inputted one or more ultrasound data or an inputted one or more ultrasound arrays.

314. The system of claim 313, wherein the machine learning model is retrained one or more training iterations based on one or more additional ultrasound data, one or more additional ultrasound arrays, one or more additional functional ultrasound image data, or one or more additional functional ultrasound arrays.

315. The system of claim 314, the retraining of the machine learning model comprising: tune the machine learning model based on the one or more additional ultrasound data, the one or more additional ultrasound arrays, the one or more additional functional ultrasound image data, or the one or more additional functional ultrasound arrays.

316. A system for inferring a functional ultrasound array from one or more ultrasound data, comprising: one or more processors; and a memory communicatively coupled to the one or more processors and configured to store instructions that, when executed by the one or more processors, cause the system to: receive the one or more ultrasound data of one or more samples of one or more subjects; convert the one or more ultrasound data to one or more ultrasound arrays; providing the one or more ultrasound arrays to a trained machine learning model; and outputting one or more inferred functional ultrasound arrays based on the received one or more ultrasound data.

317. The system of claim 316, further comprising instructions that, when executed by the one or more processors, cause the system to: convert the one or more inferred functional ultrasound arrays to one or more inferred functional ultrasound image data.

318. The system of any one of claims 313 to 317, wherein the one or more functional ultrasound arrays comprise power Doppler images.

319. The system of any one of claims 313 to 318, wherein the one or more ultrasound data comprise a sequence of ultrasound data.

320. The system of any one of claims 313 to 319, wherein the ultrasound data comprise radio frequency data, in-phase and quadrature (IQ) data, B-mode image data, composite image data, or any combination thereof.

321. The system of any one of claims 313 to 320, wherein the ultrasound data comprise radio frequency data, in-phase and quadrature (IQ) data, composite image data, or any combination thereof, but not B-mode image data.

322. A non-transitory computer-readable storage medium storing one or more programs, the one or more programs comprising instructions for modulating neural activity of a nervous system of a subject, which when executed by one or more processors of a system cause the system to: receive ultrasound data of the nervous system from an implanted transducer, wherein the ultrasound data is indicative of a physiological state of the nervous system; process the ultrasound data of the nervous system; and send processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data.

323. The system of claim 322, wherein modulating the neural activity comprises: stimulate one or more regions of the nervous system.

324. The non-transitory computer-readable storage medium of claim 322 or 323, wherein: stimulating the one or more regions of the nervous system comprises electrically stimulating through one or more electrodes, and the instructions for modulating the neural activity comprise instructions for controlling the electrical stimulation through the one or more electrodes.

325. The non-transitory computer-readable storage medium of any one of claims 322 to 324, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the non-transitory computer-readable storage medium comprises further instructions that, when executed by the one or more processors, cause the system to: receive, by the implanted transducer, the instructions for modulating the neural activity; and perform, by the implanted transducer, the ultrasound neuromodulation.

326. The non-transitory computer-readable storage medium of any one of claims 322 to 325, wherein: modulating the neural activity comprises ultrasound neuromodulation, and the non-transitory computer-readable storage medium comprises further instructions that, when executed by the one or more processors, cause the system to: receive, by the implanted transducer, the instructions for modulating the neural activity; and perform, by the implanted transducer, the ultrasound neuromodulation. The system includes further instructions that, when executed by the one or more processors, cause the non-transitory computer-readable storage medium to: receive, by the implantable transducer, the ultrasound neuromodulation.

327. A non-transitory computer-readable storage medium storing one or more programs, the one or more programs comprising instructions for modulating neural activity of a nervous system of a subject, the instructions, when executed by one or more processors of a system, cause the system to: receive, from an implantable transducer, ultrasound data of the nervous system, wherein the ultrasound data is indicative of a physiological state of the nervous system; process the ultrasound data of the nervous system; transmit the processed ultrasound data, wherein the instructions for modulating neural activity of the nervous system are determined based on the processed ultrasound data; receive, by the implantable transducer, the instructions for modulating the neural activity; and execute, by the implantable transducer, the ultrasound neuromodulation on the subject.

328. A non-transitory computer-readable storage medium storing one or more programs, the one or more programs comprising instructions for assembling an implantable transducer, the instructions, when executed by one or more processors of a system, cause the system to: position an ultrasound array proximate to a transmissive window; and couple the housing to the positioned ultrasound array, wherein the positioned ultrasound array is at least partially disposed in the housing.

329. The non-transitory computer-readable storage medium of claim 328, wherein the housing comprises one or more housing components.

330. The non-transitory computer-readable storage medium of claim 328 or 329, wherein joining the housing to the placed ultrasound array comprises: couple the one or more of the housing components to the positioned ultrasound array.

331. A non-transitory computer-readable storage medium storing one or more programs, the one or more programs comprising instructions for assembling an implantable transducer, the instructions, when executed by one or more processors of a system, cause the system to: cast a unitary transmissive housing, wherein the unitary transmissive housing comprises a transmissive window and an ultrasound array, and the ultrasound array is disposed within the cast unitary transmissive housing.

332. A non-transitory computer-readable storage medium storing one or more programs, the one or more programs comprising instructions for training a machine learning model, the instructions, when executed by one or more processors of a system, cause the system to: receive one or more ultrasound data of one or more samples from one or more subjects obtained from an implantable transducer, and one or more functional ultrasound image data corresponding to the one or more ultrasound data; convert the one or more ultrasound data to one or more ultrasound arrays; convert the one or more functional ultrasound image data to one or more functional ultrasound arrays; and train the machine learning model with the one or more ultrasound arrays and the one or more functional ultrasound arrays to predict one or more inferred functional ultrasound arrays from an inputted one or more ultrasound data or an inputted one or more ultrasound arrays.

333. The non-transitory computer-readable storage medium of claim 332, wherein the machine learning model is retrained one or more training iterations based on one or more additional ultrasound data, one or more additional ultrasound arrays, one or more additional functional ultrasound image data, or one or more additional functional ultrasound arrays.

334. The non-transitory computer-readable storage medium of claim 333, wherein the retraining of the machine learning model comprises: fine-tune the machine learning model based on the one or more additional ultrasound data, the one or more additional ultrasound arrays, the one or more additional functional ultrasound image data, or the one or more additional functional ultrasound arrays.

335. A non-transitory computer-readable storage medium storing one or more programs, the one or more programs comprising instructions for inferring functional ultrasound arrays from one or more ultrasound data, the instructions, when executed by one or more processors of a system, cause the system to: receive the one or more ultrasound data of one or more samples from one or more subjects; convert the one or more ultrasound data into one or more ultrasound arrays; provide the one or more ultrasound arrays to a trained machine learning model; and output one or more inferred functional ultrasound arrays based on the received one or more ultrasound data.

336. The non-transitory computer-readable storage medium of claim 335, further comprising instructions, which when executed by the one or more processors, cause the system to convert the one or more inferred functional ultrasound arrays into one or more inferred functional ultrasound image data.

337. The non-transitory computer-readable storage medium of any one of claims 332-336, wherein the one or more functional ultrasound arrays comprise power Doppler images.

338. The non-transitory computer-readable storage medium of any one of claims 332-337, wherein the one or more ultrasound data comprise ultrasound data sequences.

339. The non-transitory computer-readable storage medium of any one of claims 332-338, wherein the ultrasound data comprise radio frequency data, in-phase and quadrature (IQ) data, B-mode image data, composite image data, or any combination thereof.

340. The non-transitory computer-readable storage medium of any one of claims 332-339, wherein the ultrasound data comprise radio frequency data, in-phase and quadrature (IQ) data, composite image data, or any combination thereof, but not B-mode image data.

Images