Optical nerve regulation and control system based on ultrasonic technology

Abstract

The invention provides an optic nerve regulation and control system based on an ultrasonic technology, a first ultrasonic module is deployed on the outer side of a target human eyelid and is provided with an Archimedes spiral type first ultrasonic array facing the human eye, and the first ultrasonic array is configured to generate a gathered ultrasonic beam which acts on a human eye retina ganglion cell area in a targeted manner; the positioning module is deployed on the outer side of a target human eyelid and is configured to measure three-dimensional geometric data when a human eyeball horizontally faces the front and measure the depth distance of a human eye retina ganglion cell area in the human eye; the optic nerve response prediction module is pre-trained with an optic nerve response prediction model facing a target human body, and the optic nerve response prediction model is configured to adjust sound intensity output of the first ultrasonic array based on stimulation feedback prediction of human eye retina ganglion cells. According to the invention, the optic nerve regulation and control system satisfies non-intrusive, high-precision and self-adaptive requirements, and has optic nerve regulation and control and repair functions.

Description

Technical Field

[0001] This invention belongs to the field of ultrasound system technology, and particularly relates to an optic nerve modulation system based on ultrasound technology. Background Technology

[0002] Optic nerve injury refers to damage to the optic nerve (the second cranial nerve) caused by various factors, resulting in partial or complete blockage of the transmission of visual signals from the eyes to the brain, which in turn causes visual dysfunction such as decreased vision and visual field defects.

[0003] Under current technology, traditional neurotrophic drugs (such as neuroprotective agents, antioxidants, and nutrient factors) have limited ability to regenerate and restore function in damaged optic nerves. They can usually only delay degeneration or alleviate symptoms, but cannot significantly restore the structure and function of the optic nerve. While optical nerve stimulation (ONS) systems can modulate optic nerve function through electrical stimulation, they have issues with safety, accuracy, and tolerability, and may also adversely interfere with the electrophysiological activity of the visual cortex and the overall visual system. Therefore, current technology lacks a system that can achieve non-invasive / low-invasiveness, high spatial specificity, and dynamic adjustment of stimulation intensity / pattern / location according to individual differences to achieve a repairing effect on the optic nerve. Summary of the Invention

[0004] This invention provides an ultrasound-based optic nerve modulation system to address the technical problems of conventional optic nerve modulation systems in the prior art, which struggle to achieve non-invasive / low-invasiveness, high spatial specificity, and the ability to dynamically adjust stimulation intensity / pattern / location according to individual differences.

[0005] To solve the above problems, the technical solution of the present invention is: a visual nerve modulation system based on ultrasound technology, comprising: The first ultrasound module is deployed on the outer side of the eyelid of the target human body and is equipped with an Archimedes spiral first ultrasound array facing the human eye. The first ultrasound array is configured to generate a focused ultrasound beam that targets the retinal ganglion cell region of the human eye. The positioning module is deployed on the outer side of the eyelid of the target human body and is configured to measure the three-dimensional geometric data of the human eyeball when looking straight ahead horizontally, and to measure the depth distance of the retinal ganglion cell region of the human eye in the human eye. The optic nerve response prediction module is pre-trained with an optic nerve response prediction model for the target human body. The optic nerve response prediction model is configured to adjust the sound intensity output of the first ultrasound array based on the stimulation feedback prediction of the retinal ganglion cells of the human eye.

[0006] Preferably, the positioning module includes a head support mechanism and a second ultrasound module; The head support mechanism has an internal accommodating space that conforms to the contour of the human head. The bottom of the accommodating space forms a support part. The head support mechanism is configured to allow the target human's chin to fit against the surface of the support part, maintaining the human head in a relatively fixed posture within the accommodating space. The second ultrasound module is deployed on the outer side of the target human eyelid and is equipped with a second ultrasound array and a geometric calculation unit. The second ultrasound array is arranged coaxially with the first ultrasound array. The second ultrasound module is configured to emit ultrasound waves to the human eyeball. The geometric calculation unit calculates the three-dimensional geometric data of the human eyeball based on the echoes reflected from different tissue interfaces of the human eyeball. Furthermore, the geometric calculation unit obtains the fundus blood flow distribution map of the human eyeball based on the ultrasound echoes, indirectly obtains the retinal ganglion cell region of the human eyeball by identifying the acoustic signal characteristics of the retinal vascular layer, and calculates the depth distance of the retinal ganglion cell region of the human eyeball in the same coordinate system as the human eyeball.

[0007] Preferably, the positioning module further includes a positioning ring and an auxiliary positioning screen. The positioning ring has a through hole in the center, and the auxiliary positioning screen is located in the through hole area in the center of the positioning ring. The auxiliary positioning screen displays a fixed view icon in the center. When the human eye is focused on the fixed view icon, the human eyeball is horizontally looking straight ahead. The first ultrasonic array, the second ultrasonic array, and the positioning ring are arranged coaxially, and the fixed view marker is located at the center of the first ultrasonic array, the second ultrasonic array, and the positioning ring; The positioning ring can be moved horizontally towards the human eyeball and abut against the outer side of the human eyelid, so that the positioning ring and the human eyeball maintain a relatively fixed posture.

[0008] Preferably, an ultrasound-based optic nerve modulation system further includes a calibration module, which comprises a simulation coupling medium unit and a calibration calculation unit; The simulated coupling medium unit is configured such that the first ultrasound module faces the coupling medium therein, which is used to simulate the first ultrasound module facing the human eye tissue, and to make the first ultrasound array generate a focused ultrasound beam with preset requirements, and to measure and acquire the array element phase offset data, array element geometric coordinates and emission pointing offset data, array element ultrasound signal attenuation and ultrasound signal noise data of the first ultrasound array. The calibration calculation unit is configured to calculate the phase compensation at the target focal point of the first ultrasonic array. The calculation expression for the phase compensation is as follows: in, For the first Each element at frequency The following delay compensation time; and Represent the three-dimensional spatial coordinates of the target focus and the first... Each array element is based on spatially calibrated three-dimensional coordinates. Speed ​​of sound; Representing the Each element at frequency Phase offset below; Angular frequency; Furthermore, the calibration calculation unit is configured to calculate the amplitude weight correction of the first ultrasonic array, and the calculation expression for the amplitude weight correction is: in, For the first Each element at frequency The amplitude after correction; The value of the array element amplitude weighted window function; This is the amplitude-frequency response matrix.

[0009] Preferably, an ultrasound-based optic nerve modulation system further includes a simulation module, which is equipped with an acoustic simulation model of the eye. The element driving signal of the first ultrasonic array in the simulated input of the ocular acoustic simulation model is: in, Number the array elements. For array element In frequency The magnitude weighting of the values ​​is as follows: For the phase of the array element, Angular frequency, For transmission frequency; The Westervelt nonlinear acoustic equation used in the eye acoustic simulation model is as follows: in, For pressure sound field, The speed of sound is spatially dependent. For tissue density, These are nonlinear coefficients. The range is limited to [3.5, 7], and the nonlinear coefficient is... Used to indicate the nonlinear effect of human eye soft tissue on the focused ultrasound beam generated by the first ultrasound array; The expression for calculating sound intensity attenuation in the human eye using the eye acoustic simulation model is as follows: Among them, targeting the cornea and sclera tissue of the human eye, Limited to Targeting the lens and vitreous tissue of the human eye, Limited to ; ; The ocular acoustic simulation model is configured to simulate and verify the focused ultrasonic beam of the first ultrasonic array, thereby obtaining the three-dimensional sound pressure simulation distribution of the retinal ganglion cell region of the human eye. Simulation data on focal size, main lobe energy ratio, and side lobe suppression rate.

[0010] Preferably, the optic nerve response prediction model is configured such that, during the model training phase, based on the simulation module, simulation data generated corresponding to the ocular acoustic simulation model is acquired using a focused ultrasound beam under preset conditions. The simulation data set of the ocular acoustic simulation model is as follows: in, For the first The driving amplitude of each transducer unit For phase, For carrier frequency, Indicates pulse width or pulse sequence parameters; Furthermore, multimodal physiological feedback data was collected from the subjects, including intraocular pressure. Nerve discharge signals Electroretinography (ERG) and electroencephalography (EEG) indicators; The optic nerve response prediction model generates a training sample sequence based on simulation data corresponding to the ocular acoustic simulation model and multimodal physiological feedback data of the human body. And, generate a set of human physiological effect indicators: in, This refers to the intensity of short-term neural stimulation. The rate of enhanced expression of brain-derived neurotrophic factor; This refers to the change in intraocular pressure. This is a safety dosage indicator.

[0011] Preferably, the optic nerve response prediction model is further configured to perform time-series data modeling on the simulation data corresponding to the ocular acoustic simulation model and the multimodal physiological feedback data of the human body during the model training phase, thereby capturing long-term dependencies: in, The hidden state vector; These are network weight parameters; For long short-term memory units, it is a non-linear mapping function; The training objective of the optic nerve response prediction model is to minimize the predicted output. Compared with experimental measurements Mean square error: The optic nerve response prediction model is configured to establish a mapping relationship between the parameters of the focused ultrasound beam output by the first ultrasound array and human multimodal physiological feedback data.

[0012] Preferably, the prediction and modulation algorithm of the optic nerve response prediction model is as follows: in, This is the cost function, used to evaluate the effects of neural stimulation on safety constraints; To predict the length of the time domain; The predicted future effect value output by the optic nerve response prediction model; Safety constraints include thermal effects and mechanical safety constraints on human eye tissue. Smoothness of power variation in the first ultrasonic array The focal sound pressure range of the first ultrasonic array .

[0013] Preferably, the optic nerve response prediction model combines a long short-term memory network (LSTM) with model predictive control (MPC).

[0014] Because the present invention adopts the above technical solution, it has the following advantages and positive effects compared with the prior art: This invention provides an ultrasound-based optic nerve modulation system. A first ultrasound module is deployed on the outer side of the target human eyelid and is equipped with an Archimedean spiral-shaped first ultrasound array facing the human eye. The first ultrasound array is used to generate a focused ultrasound beam that targets the retinal ganglion cell region of the human eye. A positioning module is also deployed on the outer side of the target human eyelid and is used to measure the three-dimensional geometric data of the human eye when looking straight ahead horizontally, and to measure the depth distance of the retinal ganglion cell region of the human eye within the human eye. An optic nerve response prediction module is pre-trained with an optic nerve response prediction model facing the target human body. The optic nerve response prediction model can adjust the sound intensity output of the first ultrasound array based on the stimulation feedback prediction results of the retinal ganglion cells of the human eye. In this invention, firstly, a first ultrasonic array is used to generate low-intensity focused ultrasound to activate the human optic nerve through mechanical effects. In conjunction with a positioning module, the ultrasonic energy can be highly confined to the retinal ganglion cell region of the human eye, without interfering with the normal electrical activity of the visual cortex, significantly improving the safety of the optic nerve modulation system. Secondly, based on the optic nerve response prediction module, the optic nerve modulation can be adaptively optimized according to individual differences and physiological feedback predictions. Ultimately, the optic nerve modulation system meets the requirements of being non-invasive, highly precise, adaptive, and possessing both optic nerve modulation and repair functions. Attached Figure Description

[0015] Figure 1 A schematic diagram of the optic nerve modulation system based on ultrasound technology provided by this invention; Figure 2 A schematic diagram of the anteroposterior structure of the optic nerve modulation system based on ultrasound technology provided by this invention; Figure 3 A schematic diagram of the focusing of the first ultrasound array under different axial depth conditions provided by the present invention; Figure 4 A schematic diagram of the head support mechanism provided by this invention; Figure 5 A schematic diagram of the module composition of the optic nerve modulation system based on ultrasound technology provided by this invention.

[0016] Explanation of reference numerals in the attached drawings: 1: First ultrasonic array; 2: Head support mechanism; 201: Accommodation space; 202: Support part; 3: Positioning ring; 4: Fixing reference numeral. Detailed Implementation

[0017] The present invention provides a optic nerve modulation system based on ultrasound technology, which will be further described in detail below with reference to the accompanying drawings and specific embodiments. The advantages and features of the present invention will become clearer from the following description and claims.

[0018] See Figure 1 , Figure 2 , Figure 5This embodiment provides an ultrasound-based optic nerve modulation system for controlling the human optic nerve through ultrasound technology. The main structure of the optic nerve modulation system includes a first ultrasound module, a positioning module, and an optic nerve response prediction module. In one embodiment, the optic nerve modulation system can also be integrated into an ultrasound transducer device.

[0019] The first ultrasound module is deployed on the outer side of the target human eyelid. The first ultrasound module forms an array structure through the combination of multiple independent array elements. Delay control is used to achieve phase calibration of the ultrasound beam, and the ultrasound beams are superimposed in the focal region to form a focused beam. In this embodiment, specifically, an Archimedean spiral first ultrasound array 1 is formed facing the human eye. The first ultrasound array 1 can generate a focused ultrasound beam that targets the retinal ganglion cell region of the human eye.

[0020] In the structural design of the first ultrasound array 1, it is necessary to optimize the number and layout of array elements to suppress adverse conditions such as secondary focus and sidelobe energy leakage. For example, a conventional discrete array arrangement will form grating lobes, causing acoustic energy to leak into non-target tissue areas, leading to local heat deposition and increasing safety risks. On the other hand, an excessive increase in the number of array elements will increase the complexity of the drive system and manufacturing costs. Conventional standard array forms include periodic arrays and random arrays. Periodic arrays are prone to producing obvious side lobes and grating lobes, limiting focal deflection capabilities. Although random arrays can reduce side lobes, the sparse array elements will reduce focusing energy. Therefore, in this embodiment, the first ultrasound array 1 is designed as an Archimedean spiral structure. Spiral arrays are pseudo-random layouts that can balance high-energy focusing and sidelobe suppression capabilities. Moreover, the structure is simple and easy to manufacture, with more flexible array arrangement capabilities and a larger focal deflection range, making it more suitable for adapting to the control needs of different axial depths and target areas.

[0021] Also, please refer to Figure 3 In this embodiment, the first ultrasound array 1 is a thin-film piezoelectric device (PMUT) made of aluminum scandium nitride (AlScN) material. Combined with phased array technology, it can achieve electronically adjustable and rapid switching of the depth of focus, enabling the ultrasound focus to be precisely aligned with the target point under different axial length conditions, thereby improving operational efficiency and clinical safety. The effective area of ​​a single ultrasound focusing is limited to 1.5-1.7mm * 1.5-1.7mm.

[0022] Similarly, the positioning module is deployed on the outer side of the target human eyelid to measure the three-dimensional geometric data of the human eyeball when looking straight ahead, and to measure the depth distance of the retinal ganglion cell region of the human eye in the human eye.

[0023] In the structure of the human eye, the retinal ganglion cell region is located in the innermost layer of the retina at the back of the eyeball, and the relative position between the retinal ganglion cell region and the geometric structure of the human eyeball is fixed. Therefore, in this embodiment, the positioning module can determine the specific distribution location of the retinal ganglion cell region of the human eye by keeping the human eyeball in a preset posture and measuring the depth distance of the retinal ganglion cell region of the human eye in the human eye.

[0024] The optic nerve response prediction module is pre-trained with an optic nerve response prediction model for the target human body. The optic nerve response prediction model can predict the stimulation feedback of the retinal ganglion cells of the human eye and adjust the sound intensity output of the first ultrasound array 1. This allows the first ultrasound array 1 to generate a personalized ultrasound energy control scheme based on the theoretical physiological feedback of different human bodies, thereby achieving safe, precise, and adaptive ultrasound stimulation of the optic nerve.

[0025] Therefore, this embodiment provides a visual nerve modulation system based on ultrasound technology. All components of the visual nerve modulation system are deployed on the outer side of the human eyelids and do not directly contact the human eyeball. First, the positioning module can measure and acquire the position data of the retinal ganglion cell region of the human eye. Then, the first ultrasound module generates a focused ultrasound beam based on the Archimedes spiral first ultrasound array 1 and applies ultrasound stimulation to the retinal ganglion cell region of the human eye. During this process, the visual nerve response prediction module predicts the stimulation feedback of the retinal ganglion cells of the human eye in real time and dynamically adjusts the ultrasound energy output of the first ultrasound array 1, so that the visual nerve modulation system meets the requirements of non-invasiveness, high precision, self-adaptation, and has both nerve modulation and repair functions.

[0026] The specific structure and functions of the ultrasound-based optic nerve modulation system provided in this embodiment will be described in further detail below: Preferably, in one embodiment, see Figure 4 The positioning module includes a head support mechanism 2 and a second ultrasound module.

[0027] The head support mechanism 2 has an open accommodating space 201 that adapts to the contour of the human head. The bottom of the accommodating space 201 forms a support part 202. When the head support mechanism 2 is in use, the chin of the target human body is naturally placed on the surface of the support part 202 to maintain the human head in a relatively fixed posture within the accommodating space 201 and ensure the stability of the human eye position.

[0028] The second ultrasound module is deployed on the outer side of the target human eyelid. The second ultrasound module includes a second ultrasound array and a geometric calculation unit. The second ultrasound array is arranged coaxially with the first ultrasound array 1. In one embodiment, the second ultrasound array is arranged around the outer periphery of the first ultrasound array 1. The first ultrasound module, the second ultrasound module, the positioning module, or an integrated ultrasound transducer device of the three can all move within the accommodating space 201. After moving to a preset position, the position can be fixed by a locking mechanism and a head-bearing mechanism 2.

[0029] In this embodiment, the second ultrasound module emits ultrasound waves into the human eye via a second ultrasound array. The geometric calculation unit calculates the three-dimensional geometric data of the human eye based on the echoes reflected from different tissue interfaces (cornea, sclera, anterior chamber, lens, vitreous body, etc.). Furthermore, the geometric calculation unit acquires a fundus blood flow distribution map of the human eye based on the ultrasound echoes. Since the retinal vascular layer and the retinal ganglion cell layer are anatomically closely adjacent, they can be used as biomarkers for locating the retinal plane. By identifying the specific acoustic signal characteristics of the retinal vascular layer, the retinal ganglion cell region of the human eye can be indirectly obtained. Finally, in the same coordinate system as the human eye, the depth distance of the retinal ganglion cell region within the human eye is calculated, obtaining the specific location data of the retinal ganglion cell region.

[0030] Furthermore, in one embodiment, the positioning module also includes a positioning ring 3 and an auxiliary positioning screen. The positioning ring 3 has a through hole in the center, and the auxiliary positioning screen is located in the through hole area in the center of the positioning ring 3. A fixed view icon 4 is displayed in the center of the auxiliary positioning screen. When the human eye is focused on the fixed view icon 4, the center point of the pupil of the human eye is arranged horizontally and coaxially with the fixed view icon 4, that is, the human eyeball is guided to look horizontally forward.

[0031] The first ultrasonic array 1, the second ultrasonic array, and the positioning ring 3 are arranged coaxially, and the fixed view marker 4 is located at the center of the first ultrasonic array 1, the second ultrasonic array, and the positioning ring 3.

[0032] Meanwhile, once the user's eyes are looking straight ahead, the user maintains their eye posture and closes their eyelids. The positioning ring 3 can then move horizontally towards the user's eyeballs and come into contact with the outer side of the user's eyelids, so that the positioning ring 3 and the user's eyeballs maintain a relatively fixed posture.

[0033] The positioning ring 3 has a flexible pad on the side facing the human eyeball. The positioning ring 3 is driven by a micro stepper motor to achieve precise movement. The positioning ring 3 has a central through-hole area with a circular structure. The human eyeball has an ellipsoidal structure, and the anteroposterior diameter of the human eyeball is slightly larger than its horizontal and vertical diameters. That is, the human cornea has a certain curvature of protrusion in the human eyeball. Therefore, when the positioning ring 3 abuts against the outer side of the human eyelid in the first posture, the protruding tissue of the human cornea will be locked in the central through-hole area of ​​the positioning ring 3. The positioning ring 3 can achieve the initial limitation of the human eyeball, preventing the human eyeball from rotating excessively during the subsequent intraocular pressure adjustment process, which would cause the ultrasonic energy application position to deviate.

[0034] Furthermore, in one embodiment, a plurality of pressure sensors are provided in the central through-hole area or circumferential sidewall of the positioning ring 3. The pressure sensors are configured to detect the long-term reaction force generated by the human eyeball as it passes the human eyelid towards the positioning ring 3. Since the human cornea has a convex structure, the reaction force generated by different positions of the human eyeball on the corresponding positions of the positioning ring 3 is different. Compared with other positions of the human eyeball, the human cornea generates a stronger reaction force on the pressure sensors in the positioning ring 3. When the human eyeball rotates, the reaction force exerted by the human cornea on each pressure sensor in the positioning ring 3 changes. When the reaction force at any moment exceeds the preset threshold range, it indicates that the human eyeball rotates too much and the human eyeball is marked as deviating from the horizontal forward-looking posture. At this time, the human eyeball needs to be repositioned.

[0035] Preferably, in one embodiment, the optic nerve modulation system based on ultrasound technology further includes a calibration module, which is equipped with a simulation coupling medium unit and a calibration calculation unit.

[0036] Because array-type ultrasonic transducers inevitably suffer from element size deviations and spatial positioning offsets during manufacturing and assembly, and because their ultrasonic performance accuracy requires secondary verification after prolonged use, using an uncalibrated array-type ultrasonic transducer directly can lead to phase inconsistencies between elements and deviations in effective radiation area. This can cause problems such as focus shift, sidelobe lift, main lobe distortion, and acoustic field energy leakage, affecting the precise control of ultrasonic energy in the retinal ganglion cell region of the human eye. Therefore, in this embodiment, a calibration module is used to calibrate the phase offset and spatial position of each element. Subsequently, these calibration data are used in the beamforming algorithm for compensation, realizing the calibration function of the array-type ultrasonic transducer.

[0037] The simulated coupling medium unit includes a coupling medium whose material properties are consistent with those of human eye tissue, simulating the ultrasound focusing environment of human eye tissue. In this embodiment, for array element phase offset calibration, the first ultrasound module can be oriented towards the coupling medium of the simulated coupling medium unit, enabling the first ultrasound array 1 to generate a pre-defined focused ultrasound beam. The simulated coupling medium unit, based on three-dimensional sound field scanning technology, acquires the sound pressure amplitude and phase spectrum of the first ultrasound array 1 at a reference test point. By comparing with the theoretical propagation model, the inherent phase offset of each array element is extracted. and frequency-dependent transmission amplitude-frequency characteristics The data is stored in the array element characteristic library to achieve pre-correction of the subsequent transmitted waveform, so that the array elements can achieve phase consistency and amplitude equalization within the working bandwidth.

[0038] For the spatial coordinates and emission pointing calibration of array elements, under the multi-point measurement of the simulated coupling medium element, the actual geometric coordinates of each array element are estimated by triangulation and acoustic time delay inversion algorithms. The emission pointing offset is also considered. A geometric error matrix is ​​constructed between the actual array structure and the theoretical design model, serving as the basis for propagation path compensation in dynamic focusing calculations, thus enabling the ultrasonic transducer to achieve higher spatial accuracy when focusing in the optic nerve region.

[0039] To address the issues of array element consistency and operational status screening, the simulation of coupled dielectric elements, combined with amplitude-frequency response and resonance characteristics, identifies array elements with severe attenuation or excessive noise, assigns attenuation weights to them, or removes them to improve system robustness.

[0040] Subsequently, the calibration calculation unit uses algorithm compensation to enable the first ultrasonic module to load correction factors in real time during transmit beamforming and echo reception imaging by using the array element phase offset data, array element geometric coordinates and transmit pointing offset data, array element ultrasonic signal attenuation and ultrasonic signal noise data of the first ultrasonic array 1.

[0041] Specifically, the calibration calculation unit calculates the phase compensation at the target focal point of the first ultrasonic array 1. In the propagation delay calculation, the actual position and phase offset of the array elements are introduced. The expression for phase compensation is as follows: in, For the first Each element at frequency Delay compensation time (unit: seconds); and Represent the three-dimensional spatial coordinates of the target focus and the first... Each array element is based on spatially calibrated three-dimensional coordinates. Speed ​​of sound; Representing the Each element at frequency Phase offset (unit: rad); Angular frequency (unit: rad / s).

[0042] The calibration calculation unit calculates the amplitude weight correction of the first ultrasonic array 1, based on... Pre-equalization is performed to ensure consistent output sound intensity across all array elements and suppress sidelobe formation. The calculation expression for amplitude weighting correction is as follows: in, For the first Each element at frequency The amplitude after correction; The value of the array element amplitude weighted window function; This is the amplitude-frequency response matrix.

[0043] In this embodiment, after the first ultrasonic array 1 is phase compensated and amplitude weighted based on the calibration module, the initial beam of the first ultrasonic array 1 is generated, wherein the initial target frequency is selected as 2.5MHz and the sound velocity is selected as 1500m / s.

[0044] Preferably, in one embodiment, the optic nerve modulation system based on ultrasound technology further includes a simulation module, which is equipped with an acoustic simulation model of the eye.

[0045] In this embodiment, to ensure that the ultrasonic transducer forms a sufficiently spatially specific focused sound field in the retinal ganglion cell region, a simulation module (which can be used with time-domain acoustic propagation simulation tools such as k-Wave) is used to perform a full-process numerical verification simulation of the sound field characteristics. The verification simulation process is based on the real acoustic parameters (sound velocity, density, attenuation coefficient, acoustic impedance) considering the multi-layered structure of human eye tissues, including the cornea, sclera, aqueous humor, lens, vitreous body, etc., thereby maximally simulating and reproducing the attenuation and phase shift of the real focused sound field through human eye tissues.

[0046] The element driving signal of the first ultrasound array 1 in the ocular acoustic simulation model is as follows: in, Number the array elements. For array element In frequency The magnitude weighting of the values ​​is as follows: For the phase of the array element, Angular frequency, The transmission frequency is selected as 2.5 MHz in this embodiment.

[0047] The Westervelt nonlinear acoustic equation used in the eye acoustic simulation model is as follows: in, For pressure sound field, The speed of sound is spatially dependent. For tissue density, These are nonlinear coefficients.

[0048] Nonlinear coefficients The calculation expression is: Wherein, B / A is the nonlinear parameter of the medium. Different media have different B / A values. In this embodiment, human eye tissue is mainly composed of water and some elastic components. The B / A value of water is approximately 5.2, therefore the nonlinear coefficient of water is... The B / A ratio is approximately 3.6, while the B / A ratio of human ocular soft tissue is approximately 6–9. Therefore, the nonlinear coefficient of human ocular soft tissue is... Approximately 4-5.5. In this embodiment, redundancy and nonlinear coefficients are considered. The range is limited to [3.5, 7], preferably... That is, nonlinear coefficient This is used to indicate the nonlinear effect of the soft tissue of the human eye on the focused ultrasound beam generated by the first ultrasound array 1.

[0049] Meanwhile, the expression for calculating sound intensity attenuation in the human eye using the eye acoustic simulation model is as follows: Among them, targeting the cornea and sclera tissue of the human eye, Limited to Targeting the lens and vitreous tissue of the human eye, Limited to ; .

[0050] Therefore, in this embodiment, the focused ultrasound beam of the first ultrasound array 1 is simulated and verified using an ocular acoustic simulation model. By solving the above equations, the three-dimensional sound pressure simulation distribution of the retinal ganglion cell region of the human eye is obtained. Simulation data on focal size, main lobe energy ratio, and side lobe suppression rate were obtained. The simulation results can be used to guide the design of ultrasound transducer arrays and the optimization of beamforming parameters, thereby ensuring optimal spatial consistency of focused energy and safe intensity threshold control in the retinal ganglion cell layer.

[0051] Preferably, in one embodiment, since the sound field distribution inside the human eyeball is affected by the nonlinear coupling of multiple factors such as individual anatomical structure, intraocular pressure changes and tissue acoustic parameters (such as sound velocity and impedance), traditional optic nerve response prediction models based on linear assumptions or single parameter fitting are difficult to accurately describe the complex temporal relationship between "ultrasound stimulation parameters - optic nerve response", which can easily lead to the accumulation of focusing errors and uneven distribution of treatment energy.

[0052] Therefore, in this embodiment, the visual nerve response prediction model combines a long short-term memory network (LSTM) with model prediction control (MPC). In the time domain, it captures the long-term dependency features between stimulus parameters and neurophysiological responses, and achieves real-time parameter adaptive optimization in the control layer, thereby improving prediction accuracy and safety robustness.

[0053] Specifically, during the model training phase, the optic nerve response prediction model first uses a simulation module to collect simulation data corresponding to the ocular acoustic simulation model after the first ultrasonic array 1 generates a focused ultrasonic beam under preset conditions. The simulation data set of the ocular acoustic simulation model is as follows: in, For the first The driving amplitude (driving voltage or target sound pressure amplitude) of each transducer unit. For phase, For carrier frequency, This indicates the pulse width or pulse sequence parameter.

[0054] Simultaneously, the optic nerve response prediction model collects multimodal physiological feedback data of the subject under pre-defined conditions of focused ultrasound beam stimulation. This multimodal physiological feedback data includes intraocular pressure. Nerve discharge signals Electroretinography (ERG) and electroencephalography (EEG) indicators.

[0055] In the model training phase of the optic nerve response prediction model, the measurement method for the subject's multimodal physiological feedback data is as follows: Regarding intraocular pressure... Measurements can be taken using existing equipment such as the Goldmann applanation tonometer, non-contact tonometer, and dynamic intraocular pressure monitoring system; targeting neural discharge signals. Animal experiments can be conducted by placing the tip of a microelectrode close to or inserting it into the optic nerve to record the action potentials of single or multiple neurons, or by using a multi-electrode array to simultaneously record the neural discharge activity at dozens to hundreds of sites, providing information about the population of neurons. For electroretinography (ERG), a clinical ERG system can be used, employing corneal contact lens electrodes or skin electrodes to record the total electrical response of the retina under light stimulation. For electroencephalography (EEG), a clinical EEG system can be used, after applying specific ultrasound stimulation, to measure the specific waveform data corresponding to the stimulation, extracted from the average of the EEG signal.

[0056] The optic nerve response prediction model uses simulation data from the ocular acoustic simulation model and multimodal physiological feedback data from the human body. After normalization, bandpass filtering, and temporal synchronization, a unified training sample sequence is formed. And, generate a set of human physiological effect indicators: in, This refers to the intensity of short-term neural stimulation. The rate of enhanced expression of brain-derived neurotrophic factor (BDNF); This refers to the change in intraocular pressure. The safe dosage index is determined mainly based on the local temperature rise and power spatial distribution. In this embodiment, the surface temperature change measured by infrared thermography is used as the reference standard.

[0057] Furthermore, in one embodiment, the optic nerve response prediction model further performs time-series data modeling on the simulation data corresponding to the ocular acoustic simulation model and the multimodal physiological feedback data of the human body during the model training phase, capturing long-term dependencies: in, The hidden state vector; These are network weight parameters; For long short-term memory units, it is a non-linear mapping function; The training objective of the visual neural response prediction model is to minimize the predicted output. Compared with experimental measurements Mean square error: In this embodiment, during the model training phase, the optic nerve response prediction model iteratively learns the cumulative modulation law of the ultrasound stimulation parameter sequence and the human nerve response, and establishes a mapping relationship between the parameters of the focused ultrasound beam output by the first ultrasound array 1 and the human multimodal physiological feedback data.

[0058] Furthermore, the prediction and conditioning algorithm of the visual nerve response prediction model is as follows: in, This is the cost function, used to evaluate the effects of neural stimulation on safety constraints; To predict the length of the time domain; This represents the predicted future effects output by the optic nerve response prediction model.

[0059] Safety constraints include thermal effects and mechanical safety constraints on human eye tissue. Smoothness of power variation in the first ultrasonic array 1 The focal sound pressure range of the first ultrasonic array 1 .

[0060] Based on this, the optic nerve response prediction model provided in this embodiment forms a "prediction-optimization-feedback" closed loop. The optimization adopts rolling time domain calculation, and the state prediction is updated in real time according to the latest feedback signal each time. The millisecond-level parameter adaptive adjustment is achieved through gradient descent or QP solver to ensure the dynamic safety and precise focusing of ultrasound stimulation energy distribution.

[0061] Meanwhile, to further ensure the confidence and robustness of the optic nerve response prediction model's prediction results, this embodiment introduces an uncertainty estimation module at the LSTM network output. This module utilizes methods such as model ensemble, Gaussian process (GP), or Monte Carlo Dropout (MC-Dropout) to estimate the uncertainty of the predicted output. The confidence intervals are estimated to quantify the uncertainty of the optic nerve response prediction model. .

[0062] Specifically, to determine whether the ultrasound stimulation effect has reached the target threshold, a confidence perception threshold determination is introduced. The system selects actions based on the "confidence perception threshold determination" principle: given a prediction... confidence interval If a conservative judgment rule is adopted, When the lower confidence limit is higher than the target threshold, the ultrasound stimulation is considered satisfactory, and the power can be maintained or fine-tuned; if If the ultrasound stimulation is deemed insufficient, energy compensation is triggered; otherwise, the system enters the gray zone, triggering the conservative exploration mechanism of MPC (such as reducing the power increment or extending the stimulation period) or directly reverting to the safe power. The confidence factor is set at 1-2. All control and data flows (sampling, triggering, and sending) are synchronized with a hardware clock or TTL to ensure consistency and traceability of training / online inference. This uncertainty-based threshold judgment mechanism can effectively avoid false triggering caused by overfitting of the optic nerve response prediction model or signal fluctuations, thus improving the safety of the optic nerve modulation process.

[0063] In summary, this invention provides a visual nerve modulation system based on ultrasound technology. A first ultrasound module is deployed on the outer side of the target human eyelid and is equipped with an Archimedean spiral first ultrasound array 1 facing the human eye. The first ultrasound array 1 is used to generate a focused ultrasound beam that targets the retinal ganglion cell region of the human eye. A positioning module is also deployed on the outer side of the target human eyelid and is used to measure the three-dimensional geometric data of the human eye when looking straight ahead horizontally, and to measure the depth distance of the retinal ganglion cell region of the human eye within the human eye. A visual nerve response prediction module is pre-trained with a visual nerve response prediction model facing the target human body. The visual nerve response prediction model can adjust the sound intensity output of the first ultrasound array 1 based on the stimulation feedback prediction results of the retinal ganglion cells of the human eye. In this invention, firstly, the mechanical effect of low-intensity focused ultrasound generated by the first ultrasound array 1 is used to activate the human optic nerve. In conjunction with the positioning module, the ultrasound energy can be highly confined to the retinal ganglion cell region of the human eye, without interfering with the normal electrical activity of the visual cortex, which significantly improves the safety of the optic nerve modulation system. Secondly, based on the optic nerve response prediction module, the optic nerve modulation can be adaptively optimized according to individual differences and physiological feedback prediction. Finally, the optic nerve modulation system meets the requirements of being non-invasive, highly accurate, adaptive, and having both optic nerve modulation and repair functions.

[0064] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings, but the present invention is not limited to the above embodiments. Even if various changes are made to the present invention, if these changes fall within the scope of the claims of the present invention and their equivalents, they shall still fall within the protection scope of the present invention.

Claims

1. A visual nerve modulation system based on ultrasound technology, characterized in that, include: The first ultrasound module is deployed on the outer side of the eyelid of the target human body and is equipped with an Archimedes spiral first ultrasound array facing the human eye. The first ultrasound array is configured to generate a focused ultrasound beam that targets the retinal ganglion cell region of the human eye. The positioning module is deployed on the outer side of the eyelid of the target human body and is configured to measure the three-dimensional geometric data of the human eyeball when looking straight ahead horizontally, and to measure the depth distance of the retinal ganglion cell region of the human eye in the human eye. The optic nerve response prediction module is pre-trained with an optic nerve response prediction model for the target human body. The optic nerve response prediction model is configured to adjust the sound intensity output of the first ultrasound array based on the stimulation feedback prediction of the retinal ganglion cells of the human eye.

2. The optic nerve modulation system based on ultrasound technology as described in claim 1, characterized in that, The positioning module includes a head support mechanism and a second ultrasound module; The head support mechanism has an internal accommodating space that conforms to the contour of the human head. The bottom of the accommodating space forms a support part. The head support mechanism is configured to allow the target human's chin to fit against the surface of the support part, maintaining the human head in a relatively fixed posture within the accommodating space. The second ultrasound module is deployed on the outer side of the target human eyelid and is equipped with a second ultrasound array and a geometric calculation unit. The second ultrasound array is arranged coaxially with the first ultrasound array. The second ultrasound module is configured to emit ultrasound waves to the human eyeball. The geometric calculation unit calculates the three-dimensional geometric data of the human eyeball based on the echoes reflected from different tissue interfaces of the human eyeball. Furthermore, the geometric calculation unit obtains the fundus blood flow distribution map of the human eyeball based on the ultrasound echoes, indirectly obtains the retinal ganglion cell region of the human eyeball by identifying the acoustic signal characteristics of the retinal vascular layer, and calculates the depth distance of the retinal ganglion cell region of the human eyeball in the same coordinate system as the human eyeball.

3. The optic nerve modulation system based on ultrasound technology as described in claim 2, characterized in that, The positioning module also includes a positioning ring and an auxiliary positioning screen. The positioning ring has a through hole in the center, and the auxiliary positioning screen is located in the through hole area in the center of the positioning ring. A fixed view icon is displayed in the center of the auxiliary positioning screen. When the human eye is looking at the fixed view icon, the human eyeball is horizontally looking straight ahead. The first ultrasonic array, the second ultrasonic array, and the positioning ring are arranged coaxially, and the fixed view marker is located at the center of the first ultrasonic array, the second ultrasonic array, and the positioning ring; The positioning ring can be moved horizontally towards the human eyeball and abut against the outer side of the human eyelid, so that the positioning ring and the human eyeball maintain a relatively fixed posture.

4. The optic nerve modulation system based on ultrasound technology as described in claim 1, characterized in that, It also includes a calibration module, which comprises a simulation coupling medium unit and a calibration calculation unit; The simulated coupling medium unit is configured such that the first ultrasound module faces the coupling medium therein, which is used to simulate the first ultrasound module facing the human eye tissue, and to make the first ultrasound array generate a focused ultrasound beam with preset requirements, and to measure and acquire the array element phase offset data, array element geometric coordinates and emission pointing offset data, array element ultrasound signal attenuation and ultrasound signal noise data of the first ultrasound array. The calibration calculation unit is configured to calculate the phase compensation at the target focal point of the first ultrasonic array. The calculation expression for the phase compensation is as follows: in, For the first Each element at frequency The following delay compensation time; and Represent the three-dimensional spatial coordinates of the target focus and the first... Each array element is based on spatially calibrated three-dimensional coordinates. Speed ​​of sound; Representing the Each element at frequency Phase offset below; Angular frequency; Furthermore, the calibration calculation unit is configured to calculate the amplitude weight correction of the first ultrasonic array, and the calculation expression for the amplitude weight correction is: in, For the first Each element at frequency The amplitude after correction; The value of the array element amplitude weighted window function; This is the amplitude-frequency response matrix.

5. The optic nerve modulation system based on ultrasound technology as described in claim 1, characterized in that, It also includes a simulation module, which is equipped with an acoustic simulation model of the eye; The element driving signal of the first ultrasonic array in the simulated input of the ocular acoustic simulation model is: in, Number the array elements. For array element In frequency The magnitude weighting of the values ​​is as follows: For the phase of the array element, Angular frequency, For transmission frequency; The Westervelt nonlinear acoustic equation used in the eye acoustic simulation model is as follows: in, For pressure sound field, The speed of sound is spatially dependent. For tissue density, These are nonlinear coefficients. The range is limited to [3.5, 7], and the nonlinear coefficient is... Used to indicate the nonlinear effect of human eye soft tissue on the focused ultrasound beam generated by the first ultrasound array; The expression for calculating sound intensity attenuation in the human eye using the eye acoustic simulation model is as follows: Among them, targeting the cornea and sclera tissue of the human eye, Limited to Targeting the lens and vitreous tissue of the human eye, Limited to ; ; The ocular acoustic simulation model is configured to simulate and verify the focused ultrasonic beam of the first ultrasonic array, thereby obtaining the three-dimensional sound pressure simulation distribution of the retinal ganglion cell region of the human eye. Simulation data on focal size, main lobe energy ratio, and side lobe suppression rate.

6. The optic nerve modulation system based on ultrasound technology as described in claim 5, characterized in that, The optic nerve response prediction model is configured such that, during the model training phase, based on the simulation module, it acquires simulation data corresponding to the ocular acoustic simulation model using a focused ultrasound beam under preset conditions. The simulation data set of the ocular acoustic simulation model is as follows: in, For the first The driving amplitude of each transducer unit For phase, For carrier frequency, Indicates pulse width or pulse sequence parameters; Furthermore, multimodal physiological feedback data was collected from the subjects, including intraocular pressure. Nerve discharge signals Electroretinography (ERG) and electroencephalography (EEG) indicators; The optic nerve response prediction model generates a training sample sequence based on simulation data corresponding to the ocular acoustic simulation model and multimodal physiological feedback data of the human body. And, generate a set of human physiological effect indicators: in, This refers to the intensity of short-term neural stimulation. The rate of enhanced expression of brain-derived neurotrophic factor; This refers to the change in intraocular pressure. This is a safety dosage indicator.

7. The optic nerve modulation system based on ultrasound technology as described in claim 6, characterized in that, The optic nerve response prediction model is further configured to perform time-series data modeling on the simulation data corresponding to the ocular acoustic simulation model and the multimodal physiological feedback data of the human body during the model training phase, thereby capturing long-term dependencies: in, The hidden state vector; These are network weight parameters; For long short-term memory units, it is a non-linear mapping function; The training objective of the optic nerve response prediction model is to minimize the predicted output. Compared with experimental measurements Mean square error: The optic nerve response prediction model is configured to establish a mapping relationship between the parameters of the focused ultrasound beam output by the first ultrasound array and human multimodal physiological feedback data.

8. The optic nerve modulation system based on ultrasound technology as described in claim 7, characterized in that, The prediction and modulation algorithm of the optic nerve response prediction model is as follows: in, This is the cost function, used to evaluate the effects of neural stimulation on safety constraints; To predict the length of the time domain; The predicted future effect value output by the optic nerve response prediction model; Safety constraints include thermal effects and mechanical safety constraints on human eye tissue. Smoothness of power variation in the first ultrasonic array The focal sound pressure range of the first ultrasonic array .

9. The optic nerve modulation system based on ultrasound technology as described in claim 6, characterized in that, The optic nerve response prediction model combines a long short-term memory network (LSTM) with model predictive control (MPC).

Images