Electromagnetic-optical-acoustic synchronized brain functional imaging method and device

Abstract

This invention proposes an integrated electromagnetic-optical-acoustic brain functional imaging method and device, relating to the field of multimodal brain functional imaging research. This invention integrates electroencephalography (EEG), magnetoencephalography (MEG), functional near-infrared spectroscopy (FIR), and functional ultrasound imaging technologies, forming a significant complementary advantage in signal specificity and spatiotemporal resolution. It can simultaneously characterize neural activity from the perspective of human brain electrophysiology and observe neural metabolism from the perspective of hemodynamic information, achieving simultaneous recording of the brain's electrophysiological and hemodynamic activities.

Description

Technical Field

[0001] This invention relates to the field of multimodal brain functional imaging research, specifically to an integrated electromagnetic-optical-acoustic brain functional imaging method and device that integrates electroencephalography (EEG), magnetoencephalography (MEG), functional near-infrared spectroscopy (FIR), and functional ultrasound to simultaneously record the electrophysiological and hemodynamic activities of the brain and study neurovascular coupling and metabolic activities. Background Technology

[0002] Multimodal brain functional imaging, utilizing complementary signal specificities, analyzes human brain neural activity from multiple perspectives and has become a cutting-edge brain imaging technique that has attracted much attention in recent years. Given the rapid and dynamic changes in the working states of various brain regions, simultaneous multimodal brain functional imaging can capture highly sporadic and random brain activity, serving as a key foundation for cognitive neuroscience, neuropathology, and brain-computer interfaces. Neurovascular coupling describes the interrelationship between brain neurophysiological activity and hemodynamic changes, and is one of the crucial mechanisms for maintaining brain activity, associated with various brain diseases such as hypertension, stroke, and Alzheimer's disease. The simultaneous measurement and analysis of neurophysiological and hemodynamic signals is a crucial cornerstone for in-depth research on neurovascular coupling, contributing to a more comprehensive understanding of the patterns of brain neural activity and advancing cognitive neuroscience and improving the diagnosis and treatment of brain diseases.

[0003] Magnetoencephalography (MEG) and electroencephalography (EEG) are non-invasive electrophysiological measurement techniques that can measure the magnetic or electric fields generated by neuronal currents on the scalp with millisecond-level temporal resolution, thus directly recording brain electrophysiological activity. Although both MEG and EEG can directly measure neural activity in the brain, their sensitivity to neuronal currents varies across different directions and depths. Therefore, simultaneously recording and integrating MEG and EEG signals can more accurately characterize brain neural activity. Furthermore, compared to EEG source imaging, MEG source imaging is less affected by distortion caused by differences in conductivity among different tissues of the skull, allowing for more accurate localization of brain activity. However, the core component of traditional MEG, the superconducting quantum interference device (SQUID), requires a large cryogenic cooling system to operate, increasing the difficulty of combining MEG with other brain functional imaging modalities. In recent years, the rapidly developing optically pumped magnetometer (OPM) has provided a very effective alternative to SQUID MEG technology. OPM can measure extremely weak magnetic fields at room temperature without a cryogenic cooling system, significantly reducing equipment procurement and maintenance costs, and enabling more flexible detector array arrangements. Compared to the SQUID magnetoencephalography (MEG) detector, the OPM MEG detector can be placed closer to the scalp, significantly improving the signal-to-noise ratio by reducing the distance to intracranial cortical sources. The flexible detector arrangement enhances the compatibility of OPM MEG with other imaging modalities. Specifically, OPM detectors can be interleaved with other modal sensors on a helmet or customized to the individual subject's head shape and sensor arrangement requirements.

[0004] Functional near-infrared spectroscopy (FINS) is a non-ionizing, safe, low-cost, portable, and wearable neuroimaging technique for oxygenated hemodynamics. Utilizing the neurovascular coupling effect, FINS indirectly reflects brain neural activity by measuring functional hemodynamic responses in the brain (such as changes in the concentrations of oxyhemoglobin and deoxyhemoglobin). FINS uses near-infrared light to illuminate the scalp, placing a detector a few centimeters from the light source to detect changes in the intensity of the emitted light from the scalp, and calculating brain hemoglobin concentration based on different models. Currently, magnetoencephalography (MEG) and electroencephalography (EEG) have been combined with FINS to study neurovascular coupling effects, such as exercise-induced neurovascular coupling and the habitual effect of neurovascular coupling, thus advancing physiological research on neurovascular coupling effects (e.g., in sleep and normal aging) and pathological research (e.g., in Alzheimer's disease, hypertension, and stroke).

[0005] Functional ultrasound imaging (FUE) uses ultrasound to monitor changes in cerebral blood volume and blood flow velocity in real time. It is an indirect neuroimaging method for neuronal activity at high spatiotemporal resolution. Over the past decade, the development of new concepts such as ultrafast ultrasound has increased Doppler sensitivity by several orders of magnitude, paving the way for ultrasound-based functional neuroimaging. FUE can image awake and freely moving animals, as well as clinical neuroimaging of human newborns, and can be used for functional connectivity mapping in brain connectomics. 3D ultrafast ultrasound imaging has great potential in 3D mapping of human stiffness, tissue motion, and flow, offering new prospects for reducing invasive diagnostic procedures and expanding clinical ultrasound applications. Multimodal brain functional imaging technology has developed rapidly over the past thirty years, and disease diagnosis assisted by multimodal data analysis has achieved some clinical success. However, these clinical practices largely rely on patients visiting multiple departments and using multiple imaging instruments. A four-modal synchronous integrated brain functional imaging system that combines electroencephalography (EEG), magnetoencephalography (MEG), functional near-infrared spectroscopy (FIR), and functional ultrasound imaging is still a foreign field. Summary of the Invention

[0006] The purpose of this invention is to propose a non-invasive, integrated electromagnetic-optical-acoustic brain functional imaging method and device. This method combines electroencephalography (EEG), magnetoencephalography (MEG), functional near-infrared spectroscopy (FIR), and functional ultrasound imaging technologies, achieving significant complementary advantages in signal specificity and spatiotemporal resolution. It can simultaneously characterize neural activity from the perspective of human brain electrophysiology and observe neural metabolism from hemodynamic information. This device is a novel scientific instrument urgently needed in the fields of brain science research and the diagnosis and treatment of brain diseases, and will contribute to exploring the neural principles of brain cognition and the diagnosis and treatment of major brain diseases.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] An integrated electromagnetic-optical-acoustic brain functional imaging method includes the following steps:

[0009] 1) Head model construction steps: Based on the subject's magnetic resonance or CT images of the head structure, construct a head model including the scalp, skull, cerebrospinal fluid, and cortical gray matter;

[0010] 2) Four-modal data acquisition steps: Collect EEG signals, magnetoencephalogram (MEG) signals, near-infrared blood oxygenation signals, and ultrasound blood flow signals from the whole brain of the subject;

[0011] 3) Brain activity signal extraction steps: Brain activity signals are extracted from preprocessed EEG signals, magnetoencephalogram (MEG) signals, near-infrared blood oxygenation signals, and ultrasound blood flow signals to obtain four types of signal data: event-related electric and magnetic fields, event-induced blood oxygenation response, and blood flow response.

[0012] 4) Brain activity image data processing steps;

[0013] A source space for brain neural activity is constructed using a head model. The transfer matrix from the source space to the EEG and MEG sensor arrays is calculated by combining the relative position of the subject's head with the EEG and MEG sensor arrays. The noise covariance matrix of the MEG signal is calculated based on the MEG data collected without a subject. The transfer matrix is ​​then combined with minimum norm estimation to perform joint EEG-MEG signal source tracing, resulting in the cortical gray matter neural intensity matrix E. The rows of E represent the time series of neural activity at a certain cortical gray matter grid, and the columns represent different cortical gray matter grids.

[0014] The event-induced blood oxygenation response is projected onto the cortical gray matter of the head model and interpolated to obtain the cortical gray matter blood oxygenation activity matrix H, where the rows of H represent the time series of blood oxygenation activity at a certain cortical gray matter grid and the columns represent different cortical gray matter grids.

[0015] The event-induced blood flow response is projected onto the gray matter of the head model to obtain the blood flow activity matrix B, where the rows of B represent the time series of blood flow activity at a certain cortical gray matter grid, and the columns represent different cortical gray matter grids.

[0016] 5) Four-modal image fusion steps: The cortical gray matter neural intensity matrix E, the cortical gray matter blood oxygenation activity matrix H, and the blood flow activity matrix B are fused on the head model to obtain the four-modal imaging results.

[0017] Preferably, in step 1), a head tissue segmentation algorithm is used to construct a head model including the scalp, skull, cerebrospinal fluid, and cortical gray matter.

[0018] Preferably, the head tissue segmentation algorithm employs one or more of the following methods: gray-level gradient method, gray-level threshold method, and region segmentation method.

[0019] Preferably, in step 2), the collected EEG signals, MRI signals, near-infrared blood oxygenation signals, and ultrasound blood flow signals are preprocessed to remove noise and correct signals.

[0020] Preferably, step 2) of preprocessing the EEG signal includes:

[0021] Band-stop filtering to remove power frequency in the time domain, and band-pass filtering to extract the desired frequency components;

[0022] Independent component analysis was used to remove noise from eye movements, electrocardiograms, and electromyography (EMG) signals from probe attachments, and head movement EEG signals were corrected.

[0023] Preferably, step 2) of preprocessing the magnetoencephalogram (MEG) signal includes:

[0024] The collected brain magnetic signals are spatially separated to remove environmental interference signals other than brain neural activity;

[0025] Band-stop filtering to remove power frequency in the time domain, and band-pass filtering to extract the desired frequency components;

[0026] Independent component analysis was used to remove noise from eye movements, electrocardiograms, and electromyography (EMG) signals from probe attachments, and head movement and magnetoencephalography (MEG) signals were corrected.

[0027] Preferably, step 2) of preprocessing the near-infrared blood oxygenation signal includes:

[0028] The expression for the relative change in concentration of oxyhemoglobin and deoxyhemoglobin obtained using the Modified Beer-Lambert Law is used to convert the change in optical intensity into a relative change in the concentration of oxyhemoglobin and deoxyhemoglobin.

[0029] Head motion correction is applied to the relative change data, respiratory and heartbeat artifacts are removed, and bandpass filtering is used to extract the required frequency components.

[0030] Preferably, step 2) of preprocessing the ultrasound blood flow signal includes:

[0031] The distortion of the received ultrasound blood flow signal is corrected using a transcranial echo correction algorithm.

[0032] A spatiotemporal filtering algorithm based on singular value decomposition is used to filter ultrasound images in both the temporal and spatial domains to remove motion artifacts from the image reconstruction, thereby obtaining images of cerebral cortex blood flow velocity and blood volume, and reconstructing the ultrasound image.

[0033] Preferably, in step 3), a generalized linear model is used to extract brain activity signals from the preprocessed electroencephalogram (EEG) signals, magnetoencephalogram (MEG) signals, near-infrared blood oxygenation signals, and ultrasound blood flow signals, respectively.

[0034] Preferably, in step 5), the Laplace pyramid and wavelet pyramid are used to fuse the cortical gray matter neural intensity matrix E, the cortical gray matter blood oxygenation activity matrix H, and the blood flow activity matrix B on the head model.

[0035] Preferably, the method further includes the steps of calculating and characterizing coupling coefficients: calculating the coupling coefficients C1, C2, and D between any two modal data according to the following three formulas and the minimum norm estimation method, and plotting the coupling coefficient diagram on the head model based on the calculated coupling coefficients C1, C2, and D;

[0036] E = C1H T +Noise

[0037] E = C2B T +Noise

[0038] H=DB T +Noise

[0039] Where T represents matrix transpose and Noise represents noise.

[0040] An integrated electromagnetic-optical-acoustic brain functional imaging device includes:

[0041] Magnetic field shielding device is used to shield the Earth's magnetic field and external magnetic field interference, and reduce the residual magnetic field, magnetic field gradient and magnetic noise in the imaging area;

[0042] The modal data processing module includes an electroencephalography (EEG) module, a magnetoencephalography (MEG) module, a near-infrared spectroscopy (NIRS) module, and an ultrasound module; among them,

[0043] The electroencephalography (EEG) module is used to acquire whole-brain EEG signals and provide source imaging of brain electrical activity.

[0044] The magnetoencephalography (MEG) module is used to acquire whole-brain MEG signals and provide source imaging of brain magnetoencephalography.

[0045] The near-infrared module employs a multi-channel functional near-infrared spectroscopy system to measure changes in whole-brain blood oxygenation and provide a map of cerebral cortex activation.

[0046] The ultrasound module employs a functional ultrasound imaging system to measure images of blood flow velocity and blood volume in the cerebral cortex.

[0047] The four-modal synchronization module is used to synchronize the modal signals of electroencephalography (EEG), magnetoencephalography (MEG), functional near-infrared spectroscopy (FIR), and functional ultrasound through clock signals, clock feedback signals, and multi-frequency synchronization signals from the same source, and to compensate and correct each clock based on clock feedback.

[0048] The data acquisition module uses a four-modal sensor array to collect electroencephalogram (EEG), magnetic resonance imaging (MRI), near-infrared blood oxygenation, and ultrasound blood flow signals from the subject's entire brain.

[0049] Preferably, the magnetic field shielding device includes a multi-layer unequal-spacing passive shielding device, a multi-stage fingerprint coil active shielding device, and supporting mechanical devices; the multi-layer unequal-spacing passive shielding device is a semi-open 4-6 layer unequal-spacing permalloy shielding cylinder, with one end of the shielding cylinder open and the open end integrated with a non-magnetic sliding bed; the multi-stage fingerprint coil active shielding device includes a magnetic field compensation coil and a high-precision low-noise current source, and adopts an adapted dynamic compensation algorithm.

[0050] Preferably, the EEG module includes EEG electrodes and an EEG electronic module. The EEG electronic module includes processing units such as electrostatic protection, multi-stage filtering, multi-stage amplification, analog-to-digital conversion, and a microcontroller. The multi-stage amplification uses an amplifier with a high common-mode rejection ratio to perform high-pass and low-pass filtering and notch filtering on the acquired signal to filter out noise from power frequency, DC, and high frequency. A three-stage amplification circuit is used to amplify the EEG signal to the millivolt level.

[0051] Preferably, the magnetoencephalography (MEG) module includes an atomic magnetometer and a MEG electronic control and acquisition system; wherein, the MEG electronic control and acquisition system includes physiological electrodes, drivers, air chamber temperature control circuit, laser closed-loop control circuit, acquisition circuit, physiological electrical acquisition amplifier and system main controller; the system main controller adopts a programmable array logic gate circuit (FPGA) architecture to coordinate timing synchronization and data protocols between multiple channels.

[0052] Preferably, the near-infrared module includes a near-infrared light pole, a near-infrared emitting module, a near-infrared receiving module, and a main control circuit. The near-infrared emitting module emits near-infrared light into the brain region to be tested using frequency division multiplexing (FDM). The near-infrared receiving module collects the light emitted from the brain, performs Fourier transform using an FPGA to obtain the amplitude of the near-infrared light at each frequency, achieves frequency demodulation, pairs the near-infrared emitting and receiving modules according to the frequency encoding method, and identifies the imaging channel. The main control circuit includes a light source driving module for controlling the near-infrared emitting module, an analog acquisition circuit for acquiring the voltage signal of the near-infrared receiving module, a power supply module, a control module, and a communication module responsible for data aggregation and uploading.

[0053] Preferably, the ultrasound module employs a functional ultrasound imaging system, including an ultrasound transmission control and acquisition module, an ultrasound transducer array, and a functional ultrasound signal processing module; the ultrasound transducer array includes a 2MHz transducer array and a 6MHz transducer array, used to obtain the highest signal-to-noise ratio and resolution near the cerebral cortex; the functional ultrasound signal processing module employs a pre-corrected ultrasound transmission sequence, a transcranial echo correction algorithm, a singular value-based spatiotemporal filtering algorithm, and a generalized linear model.

[0054] Preferably, the four-modal synchronization module uses the phase-locked loop of the FPGA to generate two co-source synchronization signals with strict phase and fixed frequency, namely a high-frequency signal and a low-frequency signal. The high-frequency signal is connected to the magnetoencephalography (MEG) module and the electroencephalography (EEG) module, and the low-frequency signal is connected to the near-infrared spectroscopy (NIRS) module and the ultrasound module. These two co-source dual synchronization signals are used to align the data of each modality.

[0055] The present invention discloses an integrated electromagnetic-optical-acoustic brain functional imaging method and device that combines electroencephalography (EEG), magnetoencephalography (MEG), functional near-infrared spectroscopy (FIR), and functional ultrasound imaging technologies. This method and device enable simultaneous recording of the brain's electrophysiological and hemodynamic activities and offer the following advantages:

[0056] 1. This device achieves significant complementary advantages in signal specificity, temporal resolution, and spatial resolution, simultaneously characterizing neural activity from a neurophysiological perspective and observing neural metabolism from a hemodynamic perspective. Specifically, magnetoencephalography (MEG) and electroencephalography (EEG) are neuromagnetic and neuroelectrophysiological imaging, respectively, sensitive to neurophysiological signals. Functional near-infrared spectroscopy (FIR) is a blood oxygenation imaging technique, sensitive to changes in the concentrations of oxyhemoglobin and deoxyhemoglobin in the brain. Functional ultrasound (FUS) is a blood flow imaging technique, sensitive to cerebral blood volume and flow. These four modalities have different signal specificities, observing brain activity from both neurophysiological and hemodynamic perspectives, with their signal specificities complementing each other. MEG and EEG have millisecond-level temporal resolution and millimeter-level and centimeter-level spatial resolutions, respectively, generally used to observe neural activity in the cerebral cortex. FIR and FUS both have sub-second-level temporal resolution; the former has centimeter-level spatial resolution, covering the entire brain but with imaging depth limited to the cerebral cortex; the latter has sub-millimeter-level spatial resolution, covering part of the brain region and with greater imaging depth. Therefore, this invention integrates the four sub-modalities of electromagnetic-optical-acoustic synchronous brain functional imaging in terms of signal specificity, spatiotemporal resolution, and imaging area.

[0057] 2. The simultaneous measurement and analysis of neurophysiological signals and hemodynamic signals is a necessary approach to in-depth research on the neurovascular coupling effect. It helps to gain a more comprehensive understanding of the laws governing brain neural activity and is a necessary foundation for advancing cognitive neuroscience and diagnosing brain diseases.

[0058] 3. Currently, a non-invasive, simultaneous integrated brain functional imaging system encompassing electromagnetic, optical, and acoustic modalities is still lacking internationally. Simultaneous quadmodal brain functional imaging can simultaneously capture highly sporadic and random brain activity, making it an important research platform for cognitive neuroscience, brain pathology, brain-computer interfaces, and other fields. Attached Figure Description

[0059] Figure 1 This is a schematic diagram of the magnetoencephalography (MEG) control and acquisition module of the present invention;

[0060] Figure 2 This is a schematic diagram of the multi-channel functional near-infrared spectroscopy module structure of the present invention;

[0061] Figure 3 This is a schematic diagram of the near-infrared optical pole and optical fiber of the present invention;

[0062] Figure 4 This is a schematic diagram of the functional ultrasound signal processing module of the present invention;

[0063] Figure 5 This is a schematic diagram of the four-modal synchronization module of the present invention;

[0064] Figure 6This is a schematic diagram of the four-modal sensor array of the present invention. Detailed Implementation

[0065] To make the various technical features, advantages, or effects of the present invention more apparent and understandable, a detailed description is provided below in conjunction with the accompanying drawings.

[0066] This invention proposes an integrated electromagnetic-optical-acoustic brain functional imaging method and device that combines electroencephalography (EEG), magnetoencephalography (MEG), functional near-infrared spectroscopy (FIR), and functional ultrasound imaging technologies. The device includes a magnetic field shielding device, a modal data processing module, a four-modal synchronization module, and a data acquisition module. The modal data processing module includes an EEG module, a MEG module, a near-infrared spectroscopy module, and an ultrasound module, as detailed below:

[0067] 1. Magnetic field shielding device

[0068] The magnetic field shielding device is used to shield the Earth's magnetic field and external magnetic field interference, reducing residual magnetic fields, magnetic field gradients, and magnetic noise within the imaging area. This magnetic field shielding device includes a multi-layered, unequal-spaced passive shielding device, a multi-stage fingerprint coil active shielding device, and supporting mechanical devices. The multi-layered, unequal-spaced passive shielding device is a semi-open 4-6 layer unequal-spaced permalloy shielding cylinder with an internal diameter of not less than 650mm. One end of the shielding cylinder is open, and a non-magnetic sliding bed is integrated at the open end. The subject can lie on the non-magnetic bed and slide into the magnetic shielding cylinder from the open end along the track. The multi-stage fingerprint coil active shielding device includes a magnetic field compensation coil and a high-precision, low-noise current source, employing a suitable dynamic compensation algorithm.

[0069] 2. Electroencephalogram (EEG) module

[0070] The electroencephalography (EEG) module is used to acquire whole-brain EEG signals, providing source imaging of EEG signals. This EEG module includes EEG electrodes and an EEG electronic module. The EEG electrodes are gel-based passive electrodes, fabricated using a silver / silver chloride powder sintering process (but not limited to this). The EEG electronic module includes processing units such as electrostatic protection, multi-stage filtering, multi-stage amplification, analog-to-digital conversion, and a microcontroller. The multi-stage amplification uses a high common-mode rejection ratio (CMRR) amplifier to perform high-pass and low-pass filtering and notch filtering on the acquired signals to remove noise from power frequency, DC, and high-frequency sources. A three-stage amplification circuit amplifies the EEG signals to the millivolt level. Specifically, the first-stage amplification circuit uses a high input impedance design to facilitate the coupling of EEG signals into the electronic system; the second-stage amplification circuit uses a high CMRR to reduce interference from common-mode signals such as power frequency noise; and the third-stage amplification circuit uses a dynamically adjustable gain coefficient design to expand the dynamic range of the acquisition system.

[0071] 3. Magnetoencephalography (MEG) module

[0072] The magnetoencephalography (MEG) module is used to acquire whole-brain MEG signals and provide source imaging of brain magnetoencephalography; the MEG module includes an atomic magnetometer and a MEG electronic control and acquisition system.

[0073] Among them, the atomic magnetometer uses alkali metal rubidium atoms as the working substance and is based on the principle of spinless exchange relaxation. This atomic magnetometer is a magnetically sensitive sensor for magnetoencephalography (MEG), and its core component is a device filled with... 87 A cubic borate glass chamber containing Rb and a buffer gas (nitrogen). During sensor operation, this chamber is heated to ~130°C, causing the atoms within to enter a spin-free exchange relaxation state. A small semiconductor laser emits a pulse tuned to [missing value] from bottom to top. 87 The circularly polarized pump light of the Rb D1 line; another small semiconductor laser emits a beam tuned to the direction perpendicular to the pump light. 87 The Rb D2 line is detuned to a linearly polarized detection beam of ~0.1 nm. Based on this principle, the atomic magnetometer is sensitive to the magnetic field perpendicular to the plane formed by the pump and detection beams. By designing the directions of the pump and detection beams entering the gas chamber, the magnetic field sensitivity direction of the atomic magnetometer is set along the longitudinal direction of the sensor. The detection beam passing through the gas chamber is detected by a differential photodetector after passing through a polarizing beam splitter. The output signal is pre-amplified and transmitted to the controller unit of the atomic magnetometer. A small current coil is placed close to the gas chamber to calibrate the relationship between the magnetic field and the output voltage. A flexible printed circuit mounted on the bottom of the sensor is used for data communication and control. Multiple shielded twisted-pair cables are used to transmit signals between the flexible printed circuit and the atomic magnetometer electronics. The atomic magnetometer electronics automatically optimize the sensor control parameters (including heating temperature, laser voltage, and current) based on sensor feedback, complete the calibration process, and output parameters, analog signals, and digital signals indicating the sensor status. The control computer can monitor multiple channels in real time, control all sensor parameters, and calibrate the atomic magnetometer based on the calibration information measured before the experiment.

[0074] Designed to address the characteristics of the atomic magnetometer's output signal and the overall requirements of magnetoencephalography (MEG), as follows: Figure 1 The magnetoencephalography (MEG) electronic control and acquisition system shown includes physiological electrodes, drivers, an air chamber temperature control circuit, a laser closed-loop control circuit, an acquisition circuit, a physiological electroencephalography (EEG) acquisition amplifier, and a system main controller. The system uses Ethernet as the communication method to meet the needs of large-scale data exchange from atomic magnetometers. To address the requirement for independent control of each atomic magnetometer, each atomic magnetometer is equipped with a dedicated... Figure 1The drive control circuit is shown. Based on the characteristics of multi-channel coordination and synchronization, the system's main controller adopts a programmable gate array (FPGA) architecture to coordinate timing synchronization and data protocols between multiple channels. This magnetoencephalography (MEG) control and acquisition system employs a modular design, consisting of multiple cascaded chassis and a physiological electrode acquisition module. Each cascaded chassis contains 32 MEG acquisition control cards and one main control communication card. One MEG acquisition control card is responsible for operating one atomic magnetometer and acquiring its magnetic field. The main control communication card is responsible for receiving and parsing commands from the host computer and forwarding them to the 32 MEG acquisition control cards. Simultaneously, it collects the status, magnetic field data, and task signals of each atomic magnetometer and uploads them to the host computer. This modular design enables the MEG system to be rapidly assembled, debugged, and repaired, and it possesses strong scalability.

[0075] 4. Near-infrared module,

[0076] The near-infrared module employs a multi-channel functional near-infrared spectroscopy system to measure changes in whole-brain blood oxygenation and provide a map of cerebral cortex activation. The structure of this near-infrared module is as follows: Figure 2 As shown, the system includes a near-infrared light pole, a near-infrared emitting module, a near-infrared receiving module, and a main control circuit. The near-infrared emitting module uses frequency division multiplexing (FDM) to modulate the current driving the laser diode (or other near-infrared light emitting element) to emit near-infrared light towards the brain region to be tested. The modulation range is 1.2kHz to 13.8kHz, with a frequency interval of 200Hz between each emitting module and a constant amplitude. The near-infrared receiving module collects the brain's emitted light through an avalanche photodiode (or other near-infrared light receiving element), and uses an FPGA to perform Fourier transform to obtain the amplitude of the near-infrared light encoded at each frequency, achieving frequency-division demodulation. This FDM demodulation pairs the near-infrared emitting and receiving modules according to the frequency encoding method, identifies the imaging channel, and avoids contamination of the imaging fiber by low-frequency ambient light (such as DC natural light and 50Hz illumination light). The main control circuit includes a light source driving module to control the near-infrared emitting module, an analog acquisition circuit to collect the voltage signal of the near-infrared receiving module, a power supply module, a control module, and a communication module responsible for data aggregation and uploading.

[0077] This near-infrared module uses laser frequency division modulation and demodulation to emit and receive near-dual-band near-infrared light, such as 690nm and 830nm, but it can also be used for three-band and higher near-infrared light. Taking 690nm and 830nm wavelength light as an example... Figure 3As shown, the transmitting fiber 1 is connected to the transmitting electrode 2, emitting light at wavelengths of 690nm and 830nm; the receiving fiber 3 is connected to the receiving electrode 4, receiving the emitted light; both the transmitting electrode 2 and the receiving electrode 4 are mounted on the helmet 5 and contact the scalp 6; the light signal passes through the scalp 6 and skull 7, detecting blood oxygenation data in the cerebral cortex 8. This near-infrared module uses the Modified Beer-Lambert Law (MBLL) to convert the received near-infrared light intensity fluctuations into relative changes in the concentrations of oxyhemoglobin and deoxyhemoglobin.

[0078] 5. Ultrasonic module

[0079] The ultrasound module employs a functional ultrasound imaging system, including an ultrasound transmission control and acquisition module, an ultrasound transducer array, and a functional ultrasound signal processing module. The ultrasound transducer array includes a 2MHz center frequency transducer array and a 6MHz transducer array, with an axial focusing depth set to 2-3cm to achieve the highest signal-to-noise ratio and resolution near the cerebral cortex. The 2MHz transducer array is suitable for healthy subjects; the 6MHz transducer array is suitable for subjects undergoing bone flap surgery. The functional ultrasound signal processing module employs a pre-corrected ultrasound transmission sequence, a transcranial echo correction algorithm, a singular value-based spatiotemporal filtering algorithm, and a generalized linear model, such as... Figure 4 As shown.

[0080] The specific processing procedure is as follows: Combining the pre-modeled digital skull model, a self-developed transcranial pre-corrected ultrasound transmission sequence and transcranial echo correction algorithm are used to achieve accurate focusing of the transmitted signal and distortion correction of the received signal. Specifically, the finite element method is used to simulate the propagation process of the ultrasound signal emitted by the transducer in the head, obtaining the intensity loss and phase delay caused by the skull to the ultrasound transmission signal. These losses and delays are compensated during the ultrasound signal transmission stage to obtain the transcranial pre-corrected ultrasound transmission sequence, achieving accurate focusing of the transmitted signal. These losses and delays are corrected during the ultrasound signal reception stage to obtain the transcranial echo correction algorithm, achieving distortion correction of the received signal. The reconstructed image is then processed using a spatiotemporal filtering algorithm based on Singular Value Decomposition (SVD) to obtain clear images of cerebral cortical blood flow velocity and blood volume. The spatiotemporal filtering algorithm based on Singular Value Decomposition can be used to remove the influence of motion artifacts on image reconstruction, especially the influence on low-flow-velocity small vessel imaging, which is crucial for neurovascular coupling. The reconstructed cerebral cortex image will be used to extract functional information through a generalized linear model (GLM) to obtain an activation map of the cerebral cortex.

[0081] 6. Four-modal synchronization module

[0082] The four-modal synchronization module achieves synchronization of four modal signals—EEG, MEG, functional near-infrared spectroscopy, and functional ultrasound—through clock signals, clock feedback signals, and multi-frequency synchronization signals from the same source. It also compensates and corrects each clock signal based on the clock feedback. Figure 5 As shown, the phase-locked loop of an FPGA generates two synchronization signals with strict phase and fixed frequency (e.g., 1kHz and 10Hz). The high-frequency signal is connected to the magnetoencephalography (MEG) and electroencephalography (EEG) systems, while the low-frequency signal is connected to the functional near-infrared spectroscopy (FIR) and functional ultrasound systems. These two co-originating dual synchronization signals are used to align the data of each modality. The above also applies to three or more synchronization signals. In multi-signal scenarios, it is necessary to ensure that the multiple signals have strict phase and fixed frequency. One of the multiple signals is used as the master signal, and the remaining signals are aligned with the master signal in pairs.

[0083] Specifically, neurophysiological signals are typically generated tens of milliseconds after nerve activation and last for hundreds of milliseconds. Therefore, the typical frequency of magnetoencephalography (MEG) and electroencephalography (EEG) signals is 0-100 Hz, with a few signals reaching several hundred Hz. MEG and EEG can acquire neurophysiological signals using sampling rates above 1 kHz, and using synchronization signals of 1 kHz or higher can meet the data synchronization requirements of both. Hemodynamic signals are typically generated within seconds after nerve activity and last for tens of seconds. Therefore, the typical frequency of signals from functional near-infrared spectroscopy (FIR) and functional ultrasound (FUS) systems is below 1 Hz. Both instruments can acquire hemodynamic signals using sampling rates above 5 Hz, and using synchronization signals of 10 Hz or higher can meet the data synchronization requirements of both. To address the different signal synchronization accuracy requirements for the four modal measurements of neurophysiological and hemodynamic signals, a dual-synchronization method is adopted as the four-modal data synchronization approach. This method uses a high-frequency, high-precision, and high-stability clock-driven data synchronization unit to transmit two synchronization signals with consistent phase: a high-frequency signal and a low-frequency signal (e.g., 1 kHz, 10 Hz). High-frequency signals are input to the magnetoencephalography (MEG) and electroencephalography (EEG) systems, while low-frequency signals are input to the functional near-infrared spectroscopy (FIR) and functional ultrasound (FUS) systems. These two co-source dual-synchronization signals are used in the host computer to align the data from each modality to achieve simultaneous four-modal detection. Specifically, the FPGA's phase-locked loop (PLL) generates multiple clock signals with strict phase and fixed frequency. Each clock signal is compensated and corrected based on clock feedback, and a synchronization signal is output to achieve data synchronization across the various modal acquisition systems. A quartz crystal oscillator serves as the system clock source, generating seven clock signals and synchronization signals with fixed phase and delay at different frequencies through the FPGA's PLL. These clock signals serve as the system clocks for the EEG, MEG, and FUS systems. The aforementioned modalities provide local clock compensation for the system clocks through clock feedback signals, enabling high-speed acquisition and precise synchronization of the four modal signals.

[0084] 7. Data Acquisition Module

[0085] The data acquisition module employs a four-modal sensor array, allowing for flexible placement and alignment with the subject's head for four-modal data acquisition, including electroencephalography (EEG), magnetoencephalography (MEG), functional near-infrared spectroscopy (FIR), and functional ultrasound imaging sensors. Figure 6 As shown, the quadmodal sensor array is located on the helmet and consists of EEG electrodes 9, an atomic magnetometer 10, near-infrared electrodes 11, and an ultrasonic transducer array 12. A 3D model of the subject's head is created based on the pre-scanned magnetic resonance imaging information. The curved surface where the sensor array fits snugly against the subject's scalp is extracted, and the arrangement of the quadmodal sensors on the sensor array is optimized. Finally, a personalized quadmodal sensor array is fabricated using 3D printing technology. Specifically, the head model provides the relative positional relationship between the head and the sensors for manufacturing the quadmodal sensor array device. For example, the ultrasonic transducer is installed at the temporal window at the frontotemporal junction, and the near-infrared emitting and receiving electrodes are close to the scalp and approximately 3 cm apart. This is also linked to the quadmodal data (as described in step 4 below).

[0086] The steps and algorithms employed in this invention to achieve simultaneous electromagnetic-optical-acoustic integrated brain functional imaging are described below:

[0087] 1) Establish the head model.

[0088] For each subject's MRI or CT head structure images, head tissue segmentation algorithms such as gray-level gradient method, gray-level threshold method, and region segmentation method were used to construct a head model including scalp, skull, cerebrospinal fluid, and cortical gray matter (and white matter).

[0089] 2) Preprocessing of electromagnetic, optical, and acoustic four-mode data.

[0090] The acquired magnetoencephalogram (MEG) signals were subjected to signal-space separation to remove environmental interference signals other than brain neural activity. Further preprocessing of the EEG and MEG signals included band-stop filtering to remove the power frequency (50Hz) in the time domain, band-pass filtering to extract the desired frequency components, removal of eye movement, ECG, and probe-related electromyography (EMG) noise through independent component analysis, and head movement correction.

[0091] Near-infrared brain imaging data preprocessed by using the expression for the relative change in concentration of oxyhemoglobin and deoxyhemoglobin obtained by the Modified Beer-Lambert Law (MBLL) (as shown in the formula below) to convert the optical intensity change into the relative change in concentration of oxyhemoglobin and deoxyhemoglobin; further, head motion correction, respiratory and heartbeat artifact removal, and bandpass filtering were performed on the data to extract the required frequency components.

[0092]

[0093] Wherein, Δ[HbO] is the relative change in oxyhemoglobin concentration, Δ[HbR] is the relative change in deoxyhemoglobin concentration, ΔOD is the relative change in optical density, DPF is the differential path length factor, and wavelengths λ1 and λ2 are 690 nm and 830 nm, respectively; ε represents the molar extinction coefficient, in units of M. -1 cm -1 ; d represents the light source detector spacing, unit: cm.

[0094] For ultrasound imaging data preprocessing, a transcranial echo correction algorithm was used to correct the distortion of the received signal and reconstruct the ultrasound image. Then, a spatiotemporal filtering algorithm based on Singular Value Decomposition (SVD) was used to obtain images of cerebral cortical blood flow velocity and blood volume. The SVD-based spatiotemporal filtering algorithm can be used to remove the influence of motion artifacts on image reconstruction, especially on the imaging of small vessels with low blood flow velocities, which is crucial for neurovascular coupling.

[0095] 3) Extracting brain activity signals

[0096] Combining experimental paradigms, a generalized linear model (GLM) was used to extract functional information from preprocessed electroencephalogram (EEG), magnetoencephalogram (MEG), near-infrared oxygenation (NIO), and ultrasound blood flow signals, respectively, to obtain event-related electric field, event-related magnetic field, event-induced oxygenation response, and event-induced blood flow response.

[0097] 4) Brain activity images

[0098] 4-1) Combined EEG-MEG / MEG Tracing

[0099] A source space for brain neural activity was constructed using a head model, and the transfer matrix from the source space to the EEG and MEG sensor arrays was calculated by combining the relative positions of the subject's head and the EEG and MEG sensor arrays. The noise covariance matrix of the MEG signal was calculated based on MEG data collected without a subject, and the transfer matrix was combined with minimum norm estimation to perform joint EEG-MEG signal source tracing, yielding the cortical gray matter neural intensity matrix E. The rows of the cortical gray matter neural intensity matrix E represent the time series of neural activity at a specific cortical gray matter grid, and the columns represent different cortical gray matter grids.

[0100] 4-2) Near-infrared functional imaging

[0101] The event-induced blood oxygenation response is projected onto the cortical gray matter of the head model and interpolated to obtain the cortical gray matter blood oxygenation activity matrix H, where the rows of H represent the time series of blood oxygenation activity at a certain cortical gray matter grid, and the columns represent different cortical gray matter grids.

[0102] 4-3) Functional ultrasound imaging

[0103] Using the structural information of the 3D printed helmet, the head model and the ultrasonic transducer are registered. The event-induced blood flow response is projected onto the gray matter (and white matter) of the head model to obtain the blood flow activity matrix B, where the rows of B represent the time series of blood flow activity at a certain cortical gray matter grid, and the columns represent different cortical gray matter grids.

[0104] 5) Four-modal image fusion

[0105] Using Laplacian pyramids and wavelet pyramids, the cortical gray matter neural intensity matrix E, cortical gray matter blood oxygenation activity matrix H, and blood flow activity matrix B are fused on the head model to obtain four-modal imaging results.

[0106] As a further processing result, the following step 6) may also be included:

[0107] 6) Calculation and characterization of coupling coefficients

[0108] The neurovascular coupling column vector C represents the neurovascular coupling strength in the cortical gray matter. The blood flow and oxygenation coupling column vector D represents the coupling strength between blood flow and oxygenation in the cortical gray matter. The values ​​at each position in C and D correspond to a grid point in the cortical gray matter.

[0109] E = C1H T +Noise

[0110] E = C2B T +Noise

[0111] H=DB T +Noise

[0112] C1, C2, and D are calculated using the above formulas and the minimum norm estimation method, where T represents matrix transpose and Noise represents noise.

[0113] Based on the above calculation results, coupling coefficient diagrams were plotted on the head model. These coupling coefficient diagrams represent the relationships between E, H, and B, with different relationships occurring at different locations in the brain.

[0114] The following is a specific embodiment of the electromagnetic-optical-acoustic four-modal synchronous integrated brain functional imaging method and device provided by the present invention, which realizes multimodal brain functional imaging applications by simultaneously acquiring electrophysiological and hemodynamic activity signals of the brain. The specific implementation includes two parts: preparation stage and experimental stage.

[0115] Preparation stage

[0116] First, an MRI scan of the subject's head was performed to obtain T1-weighted structural images and arterial spin labeling images. Then, different experimental tasks were prepared based on the subject's physical condition and personal preferences. The alternative tasks included somatosensory, auditory, and visual tasks, as well as higher-level tasks such as language, motor, and memory tasks. Based on the subject's condition and experimental design, and combined with the subject's MRI results, a personalized multimodal integrated helmet was fabricated.

[0117] Experimental phase

[0118] A four-modal synchronous integrated brain functional imaging device (electromagnetic, photoacoustic, and electromagnetic) was used to simultaneously acquire whole-brain magnetoencephalography (MEG), electroencephalography (EEG), functional near-infrared spectroscopy (FIR), and ultrasound imaging signals from subjects under selected tasks, while also recording physiological signals such as electrooculography (EOG) and electrocardiography (ECG). After the experiment, various analytical methods, including brain source imaging and brain network studies, were used to investigate the subjects' brain function.

[0119] The specific experimental procedure is as follows:

[0120] 1) Before the day of the experiment, the participants were informed of the requirements: shave their heads, do not wear makeup, and do not wear underwear with metal parts.

[0121] 2) Before the experiment, remove any metal objects from the subject, such as belts, watches, mobile phones, keys, wallets, and glasses (if any). If a visual experiment is being conducted, the subject should wear special non-magnetic glasses.

[0122] 3) The subject is fitted with a multimodal integrated helmet (including an EEG cap, a near-infrared fiber optic head, and an ultrasound probe). The EEG cap is connected to the EEG electrodes using EEG ointment, and the connection is agitated with a cotton swab to reduce the impedance to below 10 kΩ. Ultrasound coupling gel is applied to the subject before installing the ultrasound probe.

[0123] 4) Fit the subject with electrooculogram (EOG) electrodes, electrocardiogram (ECG) electrodes, and other physiological monitoring sensors. To reduce skin resistance, clean the skin with an alcohol swab before placing the electrodes, apply a small amount of conductive paste to the electrodes, and then fix them in the aforementioned positions.

[0124] 5) Subjects should sit or lie down as designed in the experiment, and adjust the chair or bed so that the top of the subject's head is as close as possible to the measuring helmet without causing discomfort.

[0125] 6) Install and adjust task devices (such as screens and backdrops for visual tasks, and headphones for auditory tasks), auxiliary measurement equipment, and modal acquisition devices as needed.

[0126] 7) Explain the experimental procedure and precautions, as well as the time and requirements of the first experiment, to the subjects again.

[0127] 8) Start the task and record the first section of data.

[0128] 9) Rest and check the subject's condition.

[0129] 10) Repeat steps 7-9 to complete the experiments in the other sections.

[0130] In the foregoing description, specific embodiments are used to further illustrate the electromagnetic-optical-acoustic four-modal synchronous integrated brain functional imaging method and apparatus of the present invention, so as to enable those skilled in the art to have a more thorough understanding of the features and advantages of the present invention. It should be noted that the foregoing description is only a representative typical application. Obviously, the present invention is not limited to any specific structure, function, device, and method described herein, and may have other embodiments, or combinations of other embodiments. The software / hardware modules described in the present invention or shown in the accompanying drawings can also be flexibly adjusted as needed.

[0131] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can make modifications and changes to the above embodiments without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention should be as set forth in the claims.

Claims

1. A method for simultaneous electromagnetic-optical-acoustic brain functional imaging, characterized in that, Includes the following steps: 1) Head model construction steps: Based on the subject's magnetic resonance or CT images of the head structure, construct a head model including the scalp, skull, cerebrospinal fluid, and cortical gray matter; 2) Four-modal data acquisition steps: Collect EEG signals, magnetoencephalogram (MEG) signals, near-infrared blood oxygenation signals, and ultrasound blood flow signals from the whole brain of the subject; 3) Brain activity signal extraction steps: Brain activity signals are extracted from preprocessed EEG signals, magnetoencephalogram (MEG) signals, near-infrared blood oxygenation signals, and ultrasound blood flow signals to obtain four types of signal data: event-related electric and magnetic fields, event-induced blood oxygenation response, and blood flow response. 4) Brain activity image data processing steps; A source space for brain neural activity is constructed using a head model. The transfer matrix from the source space to the EEG and MEG sensor arrays is calculated by combining the relative position of the subject's head with the EEG and MEG sensor arrays. The noise covariance matrix of the MEG signal is calculated based on the MEG data collected without a subject. The transfer matrix is ​​then combined with minimum norm estimation to perform joint EEG-MEG signal source tracing, resulting in the cortical gray matter neural intensity matrix E. The rows of E represent the time series of neural activity at a certain cortical gray matter grid, and the columns represent different cortical gray matter grids. The event-induced blood oxygenation response is projected onto the cortical gray matter of the head model and interpolated to obtain the cortical gray matter blood oxygenation activity matrix H, where the rows of H represent the time series of blood oxygenation activity at a certain cortical gray matter grid and the columns represent different cortical gray matter grids. The event-induced blood flow response is projected onto the gray matter of the head model to obtain the blood flow activity matrix B, where the rows of B represent the time series of blood flow activity at a certain cortical gray matter grid, and the columns represent different cortical gray matter grids. 5) Four-modal image fusion steps: The cortical gray matter neural intensity matrix E, the cortical gray matter blood oxygenation activity matrix H, and the blood flow activity matrix B are fused on the head model to obtain the four-modal imaging results.

2. The method as described in claim 1, characterized in that, In step 1), a head tissue segmentation algorithm is used to construct a head model that includes the scalp, skull, cerebrospinal fluid, and cortical gray matter.

3. The method as described in claim 2, characterized in that, The head tissue segmentation algorithm employs one or more of the following methods: gray-level gradient method, gray-level threshold method, and region segmentation method.

4. The method as described in claim 1, characterized in that, In step 2), the collected EEG signals, MRI signals, near-infrared blood oxygenation signals, and ultrasound blood flow signals are preprocessed to remove noise and correct signals.

5. The method as described in claim 4, characterized in that, The steps for preprocessing EEG signals include: Band-stop filtering to remove power frequency in the time domain, and band-pass filtering to extract the desired frequency components; Independent component analysis was used to remove noise from eye movements, electrocardiograms, and electromyography (EMG) signals from probe attachments, and head movement EEG signals were corrected.

6. The method as described in claim 4, characterized in that, The steps for preprocessing magnetoencephalography (MEG) signals include: The collected brain magnetic signals are spatially separated to remove environmental interference signals other than brain neural activity; Band-stop filtering to remove power frequency in the time domain, and band-pass filtering to extract the desired frequency components; Independent component analysis was used to remove noise from eye movements, electrocardiograms, and electromyography (EMG) signals from probe attachments, and head movement and magnetoencephalography (MEG) signals were corrected.

7. The method as described in claim 4, characterized in that, The steps for preprocessing near-infrared blood oxygenation signals include: The expression for the relative change in concentration of oxyhemoglobin and deoxyhemoglobin obtained using the Modified Beer-Lambert Law is used to convert the change in optical intensity into a relative change in the concentration of oxyhemoglobin and deoxyhemoglobin. Head motion correction is applied to the relative change data, respiratory and heartbeat artifacts are removed, and bandpass filtering is used to extract the required frequency components.

8. The method as described in claim 4, characterized in that, The steps for preprocessing ultrasound blood flow signals include: The distortion of the received ultrasound blood flow signal is corrected using a transcranial echo correction algorithm. A spatiotemporal filtering algorithm based on singular value decomposition is used to filter ultrasound images in both the temporal and spatial domains to remove motion artifacts from the image reconstruction, thereby obtaining images of cerebral cortex blood flow velocity and blood volume, and reconstructing the ultrasound image.

9. The method as described in claim 1, characterized in that, In step 3), a generalized linear model is used to extract brain activity signals from the preprocessed electroencephalogram (EEG), magnetoencephalogram (MEG), near-infrared blood oxygenation, and ultrasound blood flow signals, respectively.

10. The method as described in claim 1, characterized in that, In step 5), the Laplace pyramid and wavelet pyramid are used to fuse the cortical gray matter neural intensity matrix E, the cortical gray matter blood oxygenation activity matrix H, and the blood flow activity matrix B on the head model.

11. The method according to any one of claims 1-10, characterized in that, It also includes the steps of calculating and characterizing coupling coefficients: calculate the coupling coefficients C1, C2 and D between any two modal data according to the following three formulas and the minimum norm estimation method, and draw the coupling coefficient diagram on the head model based on the calculated coupling coefficients C1, C2 and D; E=C1H T +Noise E=C2B T +Noise H=DB T +Noise Where T represents matrix transpose and Noise represents noise.

12. The method as described in claim 1, characterized in that, Step 1) uses a head tissue segmentation algorithm to construct a head model including the scalp, skull, cerebrospinal fluid, and cortical gray matter; the head tissue segmentation algorithm adopts one or more of the following methods: gray-level gradient method, gray-level threshold method, and region segmentation method. In step 2), the collected EEG signals, MRI signals, near-infrared blood oxygenation signals, and ultrasound blood flow signals are preprocessed to remove noise and correct signals. The steps for preprocessing EEG signals include: Band-stop filtering to remove power frequency in the time domain, and band-pass filtering to extract the desired frequency components; Independent component analysis was used to remove noise from eye movements, electrocardiograms, and electromyography (EMG) signals from probe attachments, and head movement EEG signals were corrected. The steps for preprocessing magnetoencephalography (MEG) signals include: The collected brain magnetic signals are spatially separated to remove environmental interference signals other than brain neural activity; Band-stop filtering to remove power frequency in the time domain, and band-pass filtering to extract the desired frequency components; Independent component analysis was used to remove noise from eye movements, electrocardiograms, and electromyography (EMG) signals from probe attachments, and head movement magnetoencephalography (MEG) signals were corrected. The steps for preprocessing near-infrared blood oxygenation signals include: The expression for the relative change in concentration of oxyhemoglobin and deoxyhemoglobin obtained using the Modified Beer-Lambert Law is used to convert the change in optical intensity into a relative change in the concentration of oxyhemoglobin and deoxyhemoglobin. Perform head motion correction on relative change data, remove breathing and heartbeat artifacts, and extract bandpass filtering for the required frequency components; The steps for preprocessing ultrasound blood flow signals include: The distortion of the received ultrasound blood flow signal is corrected using a transcranial echo correction algorithm. A spatiotemporal filtering algorithm based on singular value decomposition is used to filter ultrasound images in both the temporal and spatial domains to remove motion artifacts from the image reconstruction, thereby obtaining images of cerebral cortex blood flow velocity and blood volume, and reconstructing the ultrasound image. In step 3), a generalized linear model is used to extract brain activity signals from the preprocessed electroencephalogram (EEG), magnetoencephalogram (MEG), near-infrared blood oxygenation, and ultrasound blood flow signals, respectively. Step 5) uses Laplace's pyramid and wavelet pyramid to fuse the cortical gray matter neural intensity matrix E, cortical gray matter blood oxygenation activity matrix H, and blood flow activity matrix B on the head model; It also includes the steps of calculating and characterizing coupling coefficients: calculate the coupling coefficients C1, C2 and D between any two modal data according to the following three formulas and the minimum norm estimation method, and draw the coupling coefficient diagram on the head model based on the calculated coupling coefficients C1, C2 and D; E=C1H T +Noise E=C2B T +Noise H=DB T +Noise Where T represents matrix transpose and Noise represents noise.

13. An integrated electromagnetic-optical-acoustic brain functional imaging device, characterized in that, include: Magnetic field shielding device is used to shield the Earth's magnetic field and external magnetic field interference, and reduce the residual magnetic field, magnetic field gradient and magnetic noise in the imaging area; The modal data processing module includes an electroencephalography (EEG) module, a magnetoencephalography (MEG) module, a near-infrared spectroscopy (NIRS) module, and an ultrasound module; among them, The electroencephalography (EEG) module is used to acquire whole-brain EEG signals and provide source imaging of brain electrical activity. The magnetoencephalography (MEG) module is used to acquire whole-brain MEG signals and provide source imaging of brain magnetoencephalography. The near-infrared module employs a multi-channel functional near-infrared spectroscopy system to measure changes in whole-brain blood oxygenation and provide a map of cerebral cortex activation. The ultrasound module employs a functional ultrasound imaging system to measure images of blood flow velocity and blood volume in the cerebral cortex. The four-modal synchronization module is used to synchronize the modal signals of electroencephalography (EEG), magnetoencephalography (MEG), functional near-infrared spectroscopy (FIR), and functional ultrasound through clock signals, clock feedback signals, and multi-frequency synchronization signals from the same source, and to compensate and correct each clock based on clock feedback. The data acquisition module uses a four-modal sensor array to collect electroencephalogram (EEG), magnetic resonance imaging (MRI), near-infrared blood oxygenation, and ultrasound blood flow signals from the subject's entire brain.

14. The apparatus as claimed in claim 13, characterized in that, The magnetic field shielding device includes a multi-layer unequal-spacing passive shielding device, a multi-stage fingerprint coil active shielding device, and supporting mechanical devices. The multi-layer unequal-spacing passive shielding device is a semi-open 4-6 layer unequal-spacing permalloy shielding cylinder with one end open and a non-magnetic sliding bed integrated at the open end. The multi-stage fingerprint coil active shielding device includes a magnetic field compensation coil and a high-precision, low-noise current source, and adopts an adaptive dynamic compensation algorithm.

15. The apparatus as claimed in claim 13, characterized in that, The electroencephalography (EEG) module includes EEG electrodes and an EEG electronic module. The EEG electronic module includes processing units such as electrostatic protection, multi-stage filtering, multi-stage amplification, analog-to-digital conversion, and a microcontroller. The multi-stage amplification uses an amplifier with a high common-mode rejection ratio to perform high-pass and low-pass filtering and notch filtering on the acquired signal to filter out noise from power frequency, DC and high frequency. A three-stage amplification circuit is used to amplify the EEG signal to the millivolt level.

16. The apparatus as claimed in claim 13, characterized in that, The magnetoencephalography (MEG) module includes an atomic magnetometer and a MEG electronic control and acquisition system. The MEG electronic control and acquisition system includes physiological electrodes, drivers, air chamber temperature control circuits, laser closed-loop control circuits, acquisition circuits, physiological electrophysiological acquisition amplifiers, and a system main controller. The system main controller adopts a programmable array logic gate circuit architecture to coordinate the timing synchronization and data protocol between multiple channels.

17. The apparatus as claimed in claim 13, characterized in that, The near-infrared module includes a near-infrared photoelectric electrode, a near-infrared emitting module, a near-infrared receiving module, and a main control circuit. The near-infrared emitting module emits near-infrared light to the brain region to be tested using a frequency-division multiplexing current. The near-infrared receiving module collects the light emitted from the brain, performs Fourier transform using an FPGA to obtain the amplitude of the near-infrared light encoded at each frequency, realizes frequency demodulation, pairs the near-infrared emitting and receiving modules according to the frequency encoding method, and identifies the imaging channel. The main control circuit includes a light source driving module that controls the near-infrared emitting module, an analog acquisition circuit that collects the voltage signal of the near-infrared receiving module, a power supply module, a control module, and a communication module responsible for data aggregation and uploading.

18. The apparatus as claimed in claim 13, characterized in that, The ultrasound module employs a functional ultrasound imaging system, including an ultrasound transmission control and acquisition module, an ultrasound transducer array, and a functional ultrasound signal processing module. The ultrasound transducer array includes a 2MHz transducer array and a 6MHz transducer array, used to obtain the highest signal-to-noise ratio and resolution near the cerebral cortex. The functional ultrasound signal processing module employs a pre-corrected ultrasound transmission sequence, a transcranial echo correction algorithm, a spatiotemporal filtering algorithm based on singular values, and a generalized linear model.

19. The apparatus according to any one of claims 13-18, characterized in that, The four-modal synchronization module uses the phase-locked loop of the FPGA to generate two co-source dual synchronization signals with strict phase and fixed frequency: a high-frequency signal and a low-frequency signal. The high-frequency signal is connected to the magnetoencephalography (MEG) module and the electroencephalography (EEG) module, while the low-frequency signal is connected to the near-infrared spectroscopy (NIRS) module and the ultrasound module. These two co-source dual synchronization signals are used to align the data of each modality.

Images