Nerve electrophysiological signal real-time electronic processing system in brain surgery

Abstract

The invention discloses a neural electrophysiological signal real-time electronic processing system in brain surgery, which relates to the technical field of neural electrophysiological signal electronic processing and comprises an electrode sensing module, a pre-amplification module, a spectrum analysis module, a mode recognition module, a dynamic feedback module and a surgery guide module. The electrode sensing module is used for performing high-sensitivity contact type acquisition on cerebral cortex neuron electric signals by adopting a nanoscale composite material electrode array to obtain original nerve electrophysiological signals; the pre-amplification module is used for preprocessing the original neural electrophysiological signal by adopting a low-noise differential amplification technology to obtain an enhanced neural electric signal; the spectrum analysis module is used for performing multi-band decomposition and energy distribution analysis on the enhanced electroneurographic signals by adopting a self-adaptive wavelet packet transformation method to obtain characteristic band power spectrum data of neural activity; and carrying out multi-band decomposition and energy distribution analysis on the enhanced electroneurographic signals by adopting a self-adaptive wavelet packet transformation method.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic processing of neural electrophysiological signals, in particular to a real-time electronic processing system of neural electrophysiological signals in brain surgery. Background Art

[0002] Electronic processing of neuroelectrophysiological signals is a collection of technologies specifically designed to record, amplify, filter, analyze, and interpret electrical signals generated by the nervous system. These signals include, but are not limited to, electroencephalograms (EEGs), electromyograms (EMGs), and the action potentials of individual neurons. Therefore, utilizing advanced technologies to enhance the intelligence and safety of electronic processing of neuroelectrophysiological signals has become a pressing issue.

[0003] In the field of electronic processing of neuroelectrophysiological signals, traditional neuroelectrophysiological signal acquisition methods are difficult to achieve high-sensitivity contact acquisition of electrical signals from cerebral cortical neurons due to limitations in electrode materials and design. In addition, existing pattern recognition technologies have problems with delays and high misjudgment rates in classifying dynamically changing neural activity types, and cannot meet the needs of rapid response during surgery. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a real-time electronic processing system for neuroelectrophysiological signals during brain surgery to solve the problem that existing pattern recognition technology has delays and high misjudgment rates in classifying dynamically changing neural activity types, and cannot meet the demand for rapid response during surgery.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a real-time electronic processing system for neurophysiological signals during brain surgery, comprising:

[0008] Electrode sensing module, preamplification module, spectrum analysis module, pattern recognition module, dynamic feedback module and surgical guidance module;

[0009] The electrode sensing module is used to perform high-sensitivity contact acquisition of cerebral cortical neuron electrical signals using a nano-scale composite material electrode array to obtain original neuroelectrophysiological signals;

[0010] The preamplifier module is used to pre-process the original neural electrophysiological signal using low-noise differential amplification technology to obtain an enhanced neural electrical signal;

[0011] The spectrum analysis module is used to perform multi-band decomposition and energy distribution analysis on the enhanced neural electrical signal using an adaptive wavelet packet transform method to obtain characteristic frequency band power spectrum data of neural activity;

[0012] The pattern recognition module is used to perform time series modeling and neural activity type classification on the characteristic frequency band power spectrum data of neural activity using a deep learning model based on a long short-term memory network (LSTM) to obtain neural function area state recognition results;

[0013] The dynamic feedback module is used to integrate the results of neurological functional area status recognition using an edge computing architecture and send them to the surgical operation terminal via a 5G / fiber optic transmission channel to obtain visual feedback information;

[0014] The surgical guidance module is used to generate nerve protection path planning suggestions based on visual feedback information, combined with preoperative imaging data and real-time positioning coordinates, and superimpose AR auxiliary markers in the field of view of the surgical microscope to guide doctors to avoid key nerve functional areas and perform precise resection operations.

[0015] As a preferred embodiment of the real-time electronic processing system for neuroelectrophysiological signals during brain surgery described in the present invention, the nanoscale composite material electrode array is used to perform high-sensitivity contact acquisition of cerebral cortical neuronal electrical signals to obtain raw neuroelectrophysiological signals, and the specific steps are as follows:

[0016] A flexible polyimide substrate is used to support a graphene-silicon carbide conductive layer. A vertically aligned zinc oxide nanopillar array is constructed on the substrate through micro-nanofabrication technology. Each nanopillar is 200nm high and 50nm in diameter, with a 100nm spacing between adjacent nanopillars. This creates a multi-point contact interface that enhances charge coupling efficiency with cerebral cortical tissue, resulting in a composite electrode structure with high specific surface area and low impedance.

[0017] The composite electrode structure is mounted on a surgical head cap or adjustable bracket, and is precisely positioned and bonded to the target cerebral cortex area using preoperative imaging data. A biocompatible gel is used to assist bonding, resulting in a tightly fitted electrode-tissue interface.

[0018] When neurons in the cerebral cortex discharge, the local field potential changes, resulting in dynamic changes in the charge distribution on the surface of the zinc oxide nanopillars. The acquisition algorithm based on spatial weighted integration is used to integrate and calculate the instantaneous current received by multiple nanopillars to obtain the original neuroelectrophysiological signal, which is expressed as:

[0019]

[0020] Among them, V out(t) is the voltage output by the electrode array at time t, R e is the equivalent resistance of the electrode-tissue interface, R i is the internal on-resistance of the electrode, I i (t) represents the instantaneous current intensity detected by the i-th nanopillar at the t-th moment, N is the total number of nanopillars in the electrode array, τ is the system time constant, d i is the spatial distance between the ith nanopillar and the current neural activity center, and D is the maximum radius of the effective coverage range of the electrode.

[0021] As a preferred embodiment of the real-time electronic processing system for neuroelectrophysiological signals in brain surgery of the present invention, the low-noise differential amplification technology is used to pre-process the original neuroelectrophysiological signals to obtain enhanced neuroelectrophysiological signals, and the specific steps are as follows:

[0022] The raw neural electrophysiological signals V obtained from the nanoscale composite electrode array were analyzed using a low-noise differential amplifier. out (t) performing preliminary gain adjustment by connecting the input of the differential amplifier to the reference electrode and the working electrode respectively to eliminate common-mode interference and obtain a preliminary amplified neural electrical signal;

[0023] The preliminarily amplified neural electrical signal is subjected to frequency band selection by a bandpass filter, components with frequencies below 0.5 Hz and above 500 Hz are removed, and effective signal components related to cerebral cortical neuronal activity are retained to obtain a filtered signal;

[0024] The filtered signal is again passed through the programmable gain amplifier (PGA) for secondary gain adjustment to adjust the amplification factor and obtain the enhanced neural electrical signal.

[0025] As a preferred embodiment of the real-time electronic processing system for neuroelectrophysiological signals during brain surgery of the present invention, the adaptive wavelet packet transform method is used to perform multi-band decomposition and energy distribution analysis on the enhanced neural electrical signals to obtain the characteristic frequency band power spectrum data of the neural activity, and the specific steps are as follows:

[0026] Adaptive wavelet packet transform is used to preliminarily decompose the enhanced neural electrical signal obtained by low-noise differential amplification technology to obtain sub-signals;

[0027] Calculate the energy distribution of each sub-signal and obtain the energy value of each sub-signal. The expression is:

[0028]

[0029] Among them, E j represents the energy value of the jth sub-signal, S j(t) represents the time series of the jth sub-signal; according to the energy distribution of the sub-signals, the Shannon entropy is used as the evaluation criterion to determine the optimal decomposition layer number,

[0030] The expression is:

[0031]

[0032] Among them, H(L) represents the overall entropy value under the decomposition layer number L, N L represents the number of sub-signals under the decomposition layer L, p j Indicates the energy proportion of the j-th sub-signal;

[0033] The enhanced neural electrical signal is subjected to the final wavelet packet transform according to the selected wavelet basis function and the optimal decomposition layer number L to obtain the detailed decomposition results in each frequency band;

[0034] Calculate the corresponding power spectrum density of all decomposition results to obtain the characteristic frequency band power spectrum data P(f) of neural activity, which is expressed as:

[0035]

[0036] Where P(f) represents the power spectrum density at frequency f, T represents the signal time length, and D j (f) represents the representation of the jth sub-signal in the frequency domain.

[0037] As a preferred embodiment of the real-time electronic processing system for neuroelectrophysiological signals during brain surgery described in the present invention, the deep learning model based on the long short-term memory network (LSTM) is used to perform time series modeling and neural activity type classification on the characteristic frequency band power spectrum data of neural activity to obtain the neural function area state recognition result, and the specific steps are as follows:

[0038] The power spectrum data P(f) of the neural activity characteristic frequency band obtained from the adaptive wavelet packet transform is classified into frequency bands using a predefined frequency division method. The entire spectrum is divided into five typical EEG rhythm frequency bands, and the energy mean of each frequency band is extracted as the input feature vector to obtain the time series feature matrix for model training.

[0039] A deep learning model with a single-layer bidirectional LSTM structure was constructed. By introducing forward and backward LSTM units, bidirectional modeling of the temporal dependencies of neural electrical signals was achieved, resulting in a high-dimensional temporal feature representation of neural activity.

[0040] The high-dimensional time series feature representation output by the bidirectional LSTM is represented by a global average pooling operation to reduce the dimensionality, compressing the time series information into a feature vector of fixed length to obtain the compressed feature vector;

[0041] Based on the feature vector, the fully connected layer and the Softmax classifier are connected to define the Softmax function, which is expressed as follows:

[0042]

[0043] Among them, p k represents the probability that the sample belongs to the kth class, z k represents the raw score of the kth class output in the fully connected layer, and K represents the total number of neural activity types;

[0044] The neural activity type is probabilistically judged to obtain the neural function area state recognition result Y.

[0045] As a preferred solution of the real-time electronic processing system of neuroelectrophysiological signals in brain surgery described in the present invention, the edge computing architecture is used to fuse the results of neural functional area status recognition and send them to the surgical operation terminal via a 5G / fiber transmission channel to obtain visual feedback information. The specific steps are as follows:

[0046] The edge computing node is used to preliminarily process the neural function area state recognition result Y to obtain the preprocessed state recognition result;

[0047] The pre-processed state recognition result is fused with the pre-operative image data and the real-time positioning coordinates, and the expression is:

[0048]

[0049] Among them, F(x,y,z) represents the fused three-dimensional space data points, I i (x, y, z) represents the grayscale value of the i-th preoperative image data at position (x, y, z), N represents the number of preoperative images involved in the fusion, w i Represents the weight coefficient of the i-th image data, Y proc (x, y, z) represents the value of the state recognition result at the position (x, y, z) after preprocessing;

[0050] Visualize the fused spatial data F(x, y, z), map different types of neural activity states to different colors or transparencies, and generate intuitive visual feedback information;

[0051] The visual feedback information is sent to the surgical operation terminal using a 5G or optical fiber transmission channel.

[0052] As a preferred embodiment of the real-time electronic processing system for neuroelectrophysiological signals during brain surgery described in the present invention, the system generates neuroprotection path planning suggestions based on visual feedback information, combines preoperative imaging data with real-time positioning coordinates, and superimposes AR auxiliary markers in the surgical microscope field of view to guide the surgeon to avoid critical neurological functional areas and perform precise resection operations. The specific steps are as follows:

[0053] A 3D reconstruction algorithm is used to spatially register and reconstruct the visual feedback information received from the edge computing architecture and the preoperative imaging data to generate a 3D brain tissue model containing neural activity state markers. The expression is:

[0054] M 3D (x,y,z)=α·F(x,y,z)+(1-α)·I(x,y,z);

[0055] Among them, M 3D (x, y, z) represents the value of the 3D brain tissue model at position (x, y, z), F(x, y, z) represents the fused spatial data point, I(x, y, z) represents the grayscale value of the preoperative imaging data at position (x, y, z), and α is a weight coefficient;

[0056] According to the three-dimensional brain tissue model M 3D Different types of neural activity areas are marked in the figure, and a path planning algorithm is used to generate neural protection path planning suggestions. The expression is:

[0057] P path =arg m P (∫ P γ(P(s))ds);

[0058] Where P represents a possible surgical path, γ(P(s)) represents the risk assessment function for each point s on the path, ∫ P γ(P(s))ds represents the cumulative risk value along path P;

[0059] The neuroprotection path planning suggestion P path Converted to a format suitable for augmented reality display, the expression is:

[0060]

[0061] Among them, A R (x, y, z) indicates whether a certain point (x, y, z) in three-dimensional space belongs to the recommended surgical path;

[0062] Overlay the path mask A in the field of view of the surgical microscope R , and dynamically update the display content according to the real-time positioning coordinates to obtain a real-time surgical view V with AR auxiliary markingsAR .

[0063] As a preferred solution of the real-time electronic processing system for neurophysiological signals in brain surgery of the present invention, wherein: the real-time surgical view V with AR auxiliary mark AR Provides clear operation guidance to help doctors avoid critical nerve functional areas and perform precise resection operations;

[0064] The real-time surgical view V with AR auxiliary marking AR Feedback is provided to the doctor as the main navigation tool during the operation, while all operation trajectories and related data are recorded.

[0065] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the real-time electronic processing system for neuroelectrophysiological signals during brain surgery as described in the first aspect of the present invention is implemented.

[0066] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the real-time electronic processing system for neuroelectrophysiological signals in brain surgery as described in the first aspect of the present invention is implemented.

[0067] The beneficial effects of the present invention are: by adopting the adaptive wavelet packet transform method to perform multi-band decomposition and energy distribution analysis on the enhanced neural electrical signals, a fine analysis of the characteristics of neural activity is achieved, and the obtained characteristic frequency band power spectrum data provides rich time-frequency domain features for the pattern recognition module, so that the LSTM model can more accurately classify the type of neural activity. The method not only improves the analysis accuracy, but also reduces the misjudgment rate, which helps doctors make more accurate surgical decisions. By adopting a deep learning model based on the long short-term memory network LSTM to perform time series modeling and neural activity type classification on the characteristic frequency band power spectrum data of neural activity, efficient recognition of complex neural activity patterns is achieved. The method can not only distinguish between normal and abnormal neural activities, but also identify the specific functional area status, provide reliable neural functional area status recognition results for the dynamic feedback module, support real-time visual feedback, and greatly improve the safety and success rate of the operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0069] Figure 1 Schematic diagram of the real-time electronic processing system for neuroelectrophysiological signals during brain surgery in Example 1.

[0070] Figure 2 This is a flow chart of the enhanced neural electrical signal in Example 1. DETAILED DESCRIPTION

[0071] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0072] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0073] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0074] Example 1, with reference to Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a real-time electronic processing system for neuroelectrophysiological signals during brain surgery, comprising:

[0075] Electrode sensing module, preamplification module, spectrum analysis module, pattern recognition module, dynamic feedback module and surgical guidance module;

[0076] The electrode sensing module is used to perform high-sensitivity contact acquisition of electrical signals from cerebral cortical neurons using a nano-composite electrode array to obtain raw neuroelectrophysiological signals;

[0077] Furthermore, a flexible polyimide substrate was used to support a graphene-silicon carbide conductive layer. A vertically aligned zinc oxide nanopillar array was constructed on the substrate through micro-nanofabrication technology. Each nanopillar was 200nm high and 50nm in diameter, with a 100nm spacing between adjacent nanopillars. This formed a multi-point contact interface to enhance the charge coupling efficiency with the cerebral cortex tissue, resulting in a composite electrode structure with high specific surface area and low impedance.

[0078] The composite electrode structure is mounted on a surgical head cap or adjustable bracket, and is precisely positioned and bonded to the target cerebral cortex area using preoperative imaging data. A biocompatible gel is used to assist bonding, resulting in a tightly fitted electrode-tissue interface.

[0079] When neurons in the cerebral cortex discharge, the local field potential changes, resulting in dynamic changes in the charge distribution on the surface of the zinc oxide nanopillars. The acquisition algorithm based on spatial weighted integration is used to integrate and calculate the instantaneous current received by multiple nanopillars to obtain the original neuroelectrophysiological signal, which is expressed as:

[0080]

[0081] Among them, V out (t) is the voltage output by the electrode array at time t, R e is the equivalent resistance of the electrode-tissue interface, R i is the internal on-resistance of the electrode, I i (t) represents the instantaneous current intensity detected by the i-th nanopillar at the t-th moment, N is the total number of nanopillars in the electrode array, τ is the system time constant, d i is the spatial distance between the ith nanopillar and the current neural activity center, and D is the maximum radius of the effective coverage of the electrode;

[0082] It should be noted that the nano-scale composite material electrode array structure used has the advantage of significantly improving signal acquisition capabilities. Its zinc oxide nanocolumn array design enhances the contact area between the electrode and the cerebral cortex. Combined with the excellent conductive properties of the graphene-silicon carbide heterojunction, it effectively reduces the interface impedance. At the same time, the application of the spatial weighted integration algorithm enables the system to extract more representative neural activity features from multi-point acquisition data, overcoming the problem that traditional single-point electrodes are susceptible to local noise interference, and providing a high-quality original signal foundation for subsequent processing.

[0083] The preamplifier module is used to pre-process the original neural electrophysiological signal using low-noise differential amplification technology to obtain an enhanced neural electrical signal;

[0084] Furthermore, a low-noise differential amplifier was used to analyze the raw neurophysiological signals V obtained from the nanoscale composite electrode array. out (t) performing preliminary gain adjustment by connecting the input of the differential amplifier to the reference electrode and the working electrode respectively to eliminate common-mode interference and obtain a preliminary amplified neural electrical signal;

[0085] The initially amplified neural electrical signal is subjected to a bandpass filter for frequency band selection, removing components with frequencies below 0.5 Hz and above 500 Hz, retaining the effective signal components related to cerebral cortical neuronal activity, and obtaining a filtered signal;

[0086] The filtered signal is passed through the programmable gain amplifier (PGA) for secondary gain adjustment to adjust the amplification factor and obtain the enhanced neural electrical signal.

[0087] It should be noted that the introduction of low-noise differential amplification technology not only effectively suppresses common-mode interference but also improves the stability and reliability of signal acquisition. The selective retention of the 0.5–500 Hz frequency band by a bandpass filter further focuses on the physiological signal range closely related to cortical neuronal activity. The use of a programmable gain amplifier (PGA) enables adaptive enhancement of signals of varying intensities, ensuring that the signal reaches its optimal dynamic range before entering the next level of analysis, avoiding misjudgments or information loss due to weak or saturated signals.

[0088] The spectrum analysis module is used to perform multi-band decomposition and energy distribution analysis on the enhanced neural electrical signals using the adaptive wavelet packet transform method to obtain the power spectrum data of the characteristic frequency bands of neural activities;

[0089] Furthermore, adaptive wavelet packet transform is used to preliminarily decompose the enhanced neural electrical signals obtained from the low-noise differential amplification technique to obtain sub-signals;

[0090] Calculate the energy distribution of each sub-signal and obtain the energy value of each sub-signal. The expression is:

[0091]

[0092] Among them, E j represents the energy value of the jth sub-signal, S j (t) represents the time series of the jth sub-signal; according to the energy distribution of the sub-signals, the Shannon entropy is used as the evaluation criterion to determine the optimal decomposition layer number,

[0093] The expression is:

[0094]

[0095] Among them, H(L) represents the overall entropy value under the decomposition layer number L, N L represents the number of sub-signals under the decomposition layer L, p j Indicates the energy proportion of the j-th sub-signal;

[0096] The enhanced neural electrical signal is subjected to the final wavelet packet transform according to the selected wavelet basis function and the optimal decomposition layer number L to obtain the detailed decomposition results in each frequency band;

[0097] Calculate the corresponding power spectrum density of all decomposition results to obtain the characteristic frequency band power spectrum data P(f) of neural activity, which is expressed as:

[0098]

[0099] Where P(f) represents the power spectrum density at frequency f, T represents the signal time length, and D j (f) represents the representation of the j-th sub-signal in the frequency domain;

[0100] It should be noted that the adaptive wavelet packet transform method used is more suitable for processing non-stationary and nonlinear neural electrical signals than the traditional Fourier transform. Its multi-resolution characteristics can achieve fine decomposition of complex EEG rhythms. Combined with Shannon entropy as an evaluation criterion, it can dynamically select the optimal number of decomposition layers to avoid information redundancy or omission. The final characteristic frequency band power spectrum data provides high-dimensional and physically meaningful input features for the deep learning model, thereby improving the classification accuracy and robustness of the pattern recognition module.

[0101] The pattern recognition module is used to perform time series modeling and neural activity type classification on the characteristic frequency band power spectrum data of neural activity using a deep learning model based on the long short-term memory network (LSTM), thereby obtaining the neural functional area state recognition results;

[0102] Furthermore, a predefined frequency division method is used to classify the power spectrum data P(f) of the neural activity characteristic frequency band obtained from the adaptive wavelet packet transform, dividing the entire spectrum into five typical EEG rhythm frequency bands, and extracting the energy mean of each frequency band as the input feature vector to obtain the time series feature matrix for model training;

[0103] A deep learning model with a single-layer bidirectional LSTM structure was constructed. By introducing forward and backward LSTM units, bidirectional modeling of the temporal dependencies of neural electrical signals was achieved, resulting in a high-dimensional temporal feature representation of neural activity.

[0104] The high-dimensional time series feature representation output by the bidirectional LSTM is represented by a global average pooling operation to reduce the dimensionality, compressing the time series information into a feature vector of fixed length to obtain the compressed feature vector;

[0105] Based on the feature vector, the fully connected layer and the Softmax classifier are connected to define the Softmax function, which is expressed as follows:

[0106]

[0107] Among them, p k represents the probability that the sample belongs to the kth class, z k represents the raw score of the kth class output in the fully connected layer, and K represents the total number of neural activity types;

[0108] Probabilistically discriminate the type of neural activity and obtain the neural function area state recognition result Y;

[0109] It should be noted that the LSTM-based bidirectional time series modeling method fully considers the temporal continuity and contextual dependence of neural activity. Compared with the traditional convolutional neural network (CNN), it is more suitable for processing EEG signals with strong time series characteristics. The global average pooling operation effectively compresses the feature dimension and improves the model inference efficiency. The Softmax classifier realizes the probabilistic judgment of multiple neural activity states, providing a reliable basis for functional area identification and risk warning during surgery, and significantly enhancing the system's intelligent decision-making capabilities.

[0110] The dynamic feedback module is used to integrate the results of neural functional area status recognition using an edge computing architecture and send them to the surgical operation terminal via a 5G / fiber transmission channel to obtain visual feedback information;

[0111] Furthermore, the edge computing node is used to perform preliminary processing on the neural function area state recognition result Y to obtain the preprocessed state recognition result;

[0112] The pre-processed state recognition results are fused with the pre-operative image data and real-time positioning coordinates. The expression is:

[0113]

[0114] Among them, F(x,y,z) represents the fused three-dimensional space data points, I i (x, y, z) represents the grayscale value of the i-th preoperative image data at position (x, y, z), N represents the number of preoperative images involved in the fusion, w i Represents the weight coefficient of the i-th image data, Y proc (x, y, z) represents the value of the state recognition result at the position (x, y, z) after preprocessing;

[0115] Visualize the fused spatial data F(x, y, z), map different types of neural activity states to different colors or transparencies, and generate intuitive visual feedback information;

[0116] Use 5G or optical fiber transmission channels to send visual feedback information to the surgical operation terminal;

[0117] It should be noted that the introduction of edge computing architecture has greatly reduced the delay problem caused by cloud transmission, enabling the rapid fusion and processing of neurological functional area status recognition results locally, and combining preoperative images with real-time positioning coordinates to generate accurate spatial mapping data. Visual feedback information is transmitted to the surgical terminal at high speed through 5G / fiber optic channels, ensuring that doctors can obtain timely and intuitive operation guidance, greatly improving the system's response speed and clinical practicality, and meeting the strict real-time requirements of brain surgery.

[0118] The surgical guidance module generates neuroprotective path planning suggestions based on visual feedback information, combined with preoperative imaging data and real-time positioning coordinates. It also overlays AR auxiliary markers in the surgical microscope field of view to guide the surgeon to avoid critical neurological functional areas and perform precise resection operations.

[0119] Furthermore, a 3D reconstruction algorithm is used to spatially register and reconstruct the visual feedback information received from the edge computing architecture and the preoperative imaging data to generate a 3D brain tissue model containing neural activity state markers, expressed as:

[0120] M 3D (x,y,z)=α·F(x,y,z)+(1-α)·I(x,y,z);

[0121] Among them, M 3D (x, y, z) represents the value of the 3D brain tissue model at position (x, y, z), F(x, y, z) represents the fused spatial data point, I(x, y, z) represents the grayscale value of the preoperative imaging data at position (x, y, z), and α is a weight coefficient;

[0122] According to the three-dimensional brain tissue model M 3D Different types of neural activity areas are marked in the figure, and a path planning algorithm is used to generate neural protection path planning suggestions. The expression is:

[0123] P path =arg m P (∫ P γ(P(s))ds);

[0124] Where P represents a possible surgical path, γ(P(s)) represents the risk assessment function for each point s on the path, ∫ P γ(P(s))ds represents the cumulative risk value along path P;

[0125] The neuroprotective pathway planning proposal P path Converted to a format suitable for augmented reality display, the expression is:

[0126]

[0127] Among them, A R (x, y, z) indicates whether a certain point (x, y, z) in three-dimensional space belongs to the recommended surgical path;

[0128] Overlay path mask A on the surgical microscope field of view R , and dynamically update the display content according to the real-time positioning coordinates to obtain a real-time surgical view V with AR auxiliary markings AR ;

[0129] Real-time surgical view with AR-assisted markings AR Provides clear operation guidance to help doctors avoid critical nerve functional areas and perform precise resection operations;

[0130] The real-time surgical view V AR Feedback is provided to the doctor as the main navigation tool during the operation, while all operation trajectories and related data are recorded;

[0131] It should be noted that the combination of 3D reconstruction and AR-assisted identification technology transcends the limitations of traditional 2D navigation, enabling surgeons to visually visualize neuroprotective pathway recommendations and potential risk areas within the microscope's field of view. The path planning algorithm minimizes the cumulative risk function to provide optimal resection path selection, significantly reducing the risk of damaging critical functional areas. This module not only improves surgical precision and safety but also provides a viable path and technical paradigm for the future development of intelligent surgical navigation systems.

[0132] This embodiment also provides a computer device suitable for the real-time electronic processing system of neuroelectrophysiological signals during brain surgery, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions, thereby realizing the real-time electronic processing system of neuroelectrophysiological signals during brain surgery as proposed in the above embodiment.

[0133] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0134] This embodiment also provides a storage medium having a computer program stored thereon. When the program is executed by a processor, the real-time electronic processing system for neuroelectrophysiological signals during brain surgery as proposed in the above embodiment is implemented. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0135] In summary, the present invention uses an adaptive wavelet packet transform method to perform multi-band decomposition and energy distribution analysis on the enhanced neural electrical signals, thereby achieving a fine analysis of the characteristics of neural activity. The obtained characteristic frequency band power spectrum data provides rich time-frequency domain features for the pattern recognition module, so that the LSTM model can more accurately classify the type of neural activity. The method not only improves the analysis accuracy, but also reduces the misjudgment rate, which helps doctors make more accurate surgical decisions. By using a deep learning model based on the long short-term memory network LSTM, the characteristic frequency band power spectrum data of neural activity is subjected to time series modeling and neural activity type classification, thereby achieving efficient recognition of complex neural activity patterns. The method can not only distinguish between normal and abnormal neural activities, but also identify the specific functional area status, providing a reliable neural functional area status recognition result for the dynamic feedback module, supporting real-time visual feedback, and greatly improving the safety and success rate of the operation.

[0136] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A real-time electronic processing system for neuroelectrophysiological signals during brain surgery, characterized by: include: Electrode sensing module, preamplification module, spectrum analysis module, pattern recognition module, dynamic feedback module and surgical guidance module; The electrode sensing module is used to perform high-sensitivity contact acquisition of cerebral cortical neuron electrical signals using a nano-scale composite material electrode array to obtain original neuroelectrophysiological signals; The preamplifier module is used to pre-process the original neural electrophysiological signal using low-noise differential amplification technology to obtain an enhanced neural electrical signal; The spectrum analysis module is used to perform multi-band decomposition and energy distribution analysis on the enhanced neural electrical signal using an adaptive wavelet packet transform method to obtain characteristic frequency band power spectrum data of neural activity; The pattern recognition module is used to perform time series modeling and neural activity type classification on the characteristic frequency band power spectrum data of neural activity using a deep learning model based on a long short-term memory network (LSTM) to obtain neural function area state recognition results; The dynamic feedback module is used to integrate the results of neurological functional area status recognition using an edge computing architecture and send them to the surgical operation terminal via a 5G / fiber optic transmission channel to obtain visual feedback information; The surgical guidance module is used to generate nerve protection path planning suggestions based on visual feedback information, combined with preoperative imaging data and real-time positioning coordinates, and superimpose AR auxiliary markers in the field of view of the surgical microscope to guide doctors to avoid key nerve functional areas and perform precise resection operations.

2. The real-time electronic processing system for neurophysiological signals during brain surgery according to claim 1, characterized in that: The method uses a nano-scale composite material electrode array to perform high-sensitivity contact acquisition of cerebral cortical neuronal electrical signals to obtain original neuroelectrophysiological signals, and the specific steps are as follows: A flexible polyimide substrate is used to support a graphene-silicon carbide conductive layer. A vertically aligned zinc oxide nanopillar array is constructed on the substrate through micro-nanofabrication technology. Each nanopillar is 200nm high and 50nm in diameter, with a 100nm spacing between adjacent nanopillars. This creates a multi-point contact interface that enhances charge coupling efficiency with cerebral cortical tissue, resulting in a composite electrode structure with high specific surface area and low impedance. The composite electrode structure is mounted on a surgical head cap or adjustable bracket, and is precisely positioned and bonded to the target cerebral cortex area using preoperative imaging data. A biocompatible gel is used to assist bonding, resulting in a tightly fitted electrode-tissue interface. When neurons in the cerebral cortex discharge, the local field potential changes, resulting in dynamic changes in the charge distribution on the surface of the zinc oxide nanopillars. The acquisition algorithm based on spatial weighted integration is used to integrate and calculate the instantaneous current received by multiple nanopillars to obtain the original neuroelectrophysiological signal, which is expressed as: Among them, V out (t) is the voltage output by the electrode array at time t, R e is the equivalent resistance of the electrode-tissue interface, R i is the internal on-resistance of the electrode, I i (t) represents the instantaneous current intensity detected by the i-th nanopillar at the t-th moment, N is the total number of nanopillars in the electrode array, τ is the system time constant, d i is the spatial distance between the ith nanopillar and the current neural activity center, and D is the maximum radius of the effective coverage range of the electrode.

3. The real-time electronic processing system for neurophysiological signals during brain surgery according to claim 2, characterized in that: The low-noise differential amplification technology is used to pre-process the original neural electrophysiological signal to obtain the enhanced neural electrical signal. The specific steps are: The raw neural electrophysiological signals V obtained from the nanoscale composite electrode array were analyzed using a low-noise differential amplifier. out (t) performing preliminary gain adjustment by connecting the input of the differential amplifier to the reference electrode and the working electrode respectively to eliminate common-mode interference and obtain a preliminary amplified neural electrical signal; The preliminarily amplified neural electrical signal is subjected to frequency band selection by a bandpass filter, components with frequencies below 0.5 Hz and above 500 Hz are removed, and effective signal components related to cerebral cortical neuronal activity are retained to obtain a filtered signal; The filtered signal is again passed through the programmable gain amplifier (PGA) for secondary gain adjustment to adjust the amplification factor and obtain the enhanced neural electrical signal.

4. The real-time electronic processing system for neuroelectrophysiological signals during brain surgery according to claim 3, characterized in that: The adaptive wavelet packet transform method is used to perform multi-band decomposition and energy distribution analysis on the enhanced neural electrical signal to obtain the characteristic frequency band power spectrum data of the neural activity. The specific steps are: Adaptive wavelet packet transform is used to preliminarily decompose the enhanced neural electrical signal obtained by low-noise differential amplification technology to obtain sub-signals; Calculate the energy distribution of each sub-signal and obtain the energy value of each sub-signal. The expression is: Among them, E j represents the energy value of the jth sub-signal, S j (t) represents the time series of the jth sub-signal; according to the energy distribution of the sub-signals, the Shannon entropy is used as the evaluation criterion to determine the optimal number of decomposition layers, which is expressed as: Among them, H(L) represents the overall entropy value under the decomposition layer number L, N L represents the number of sub-signals under the decomposition layer L, p j Indicates the energy proportion of the j-th sub-signal; The enhanced neural electrical signal is subjected to the final wavelet packet transform according to the selected wavelet basis function and the optimal decomposition layer number L to obtain the detailed decomposition results in each frequency band; Calculate the corresponding power spectrum density of all decomposition results to obtain the characteristic frequency band power spectrum data P(f) of neural activity, which is expressed as: Where P(f) represents the power spectrum density at frequency f, T represents the signal time length, and D j (f) represents the representation of the jth sub-signal in the frequency domain.

5. The real-time electronic processing system for neurophysiological signals during brain surgery according to claim 4, characterized in that: The deep learning model based on the long short-term memory network (LSTM) is used to perform time series modeling and neural activity type classification on the characteristic frequency band power spectrum data of neural activity to obtain the neural function area state recognition result. The specific steps are as follows: The power spectrum data P(f) of the neural activity characteristic frequency band obtained from the adaptive wavelet packet transform is classified into frequency bands using a predefined frequency division method. The entire spectrum is divided into five typical EEG rhythm frequency bands, and the energy mean of each frequency band is extracted as the input feature vector to obtain the time series feature matrix for model training. A deep learning model with a single-layer bidirectional LSTM structure was constructed. By introducing forward and backward LSTM units, bidirectional modeling of the temporal dependencies of neural electrical signals was achieved, resulting in a high-dimensional temporal feature representation of neural activity. The high-dimensional time series feature representation output by the bidirectional LSTM is represented by a global average pooling operation to reduce the dimensionality, compressing the time series information into a feature vector of fixed length to obtain the compressed feature vector; Based on the feature vector, the fully connected layer and the Softmax classifier are connected to define the Softmax function, which is expressed as follows: Among them, p k represents the probability that the sample belongs to the kth class, z k represents the raw score of the kth class output in the fully connected layer, and K represents the total number of neural activity types; The neural activity type is probabilistically judged to obtain the neural function area state recognition result Y.

6. The real-time electronic processing system for neurophysiological signals during brain surgery according to claim 5, characterized in that: The edge computing architecture is used to fuse the results of neural functional area status recognition and send them to the surgical operation terminal through the 5G / fiber transmission channel to obtain visual feedback information. The specific steps are as follows: The edge computing node is used to preliminarily process the neural function area state recognition result Y to obtain the preprocessed state recognition result; The pre-processed state recognition result is fused with the pre-operative image data and the real-time positioning coordinates, and the expression is: Among them, F(x,y,z) represents the fused three-dimensional space data point, I i (x, y, z) represents the grayscale value of the i-th preoperative image data at position (x, y, z), N represents the number of preoperative images involved in the fusion, and w i Represents the weight coefficient of the i-th image data, Y proc (x, y, z) represents the value of the state recognition result at the position (x, y, z) after preprocessing; Visualize the fused spatial data F(x, y, z), map different types of neural activity states to different colors or transparencies, and generate intuitive visual feedback information; The visual feedback information is sent to the surgical operation terminal using a 5G or optical fiber transmission channel.

7. The real-time electronic processing system for neuroelectrophysiological signals during brain surgery according to claim 6, characterized in that: Based on the visual feedback information, combined with preoperative imaging data and real-time positioning coordinates, a nerve protection path planning suggestion is generated, and AR auxiliary markers are superimposed in the surgical microscope field of view to guide the doctor to avoid key nerve functional areas and perform precise resection operations. The specific steps are as follows: A 3D reconstruction algorithm is used to spatially register and reconstruct the visual feedback information received from the edge computing architecture and the preoperative imaging data to generate a 3D brain tissue model containing neural activity state markers. The expression is: M 3D (x,y,z)=α·F(x,y,z)+(1-α)·I(x,y,z)? Among them, M 3D (x, y, z) represents the value of the 3D brain tissue model at position (x, y, z), D(x, y, z) represents the fused spatial data point, I(x, y, z) represents the grayscale value of the preoperative imaging data at position (x, y, z), and α is a weight coefficient; According to the three-dimensional brain tissue model M 3D Different types of neural activity areas are marked in the figure, and a path planning algorithm is used to generate neural protection path planning suggestions. The expression is: P path =arg m P (∫ P γ(P(s))ds); Where P represents a possible surgical path, γ(P(s)) represents the risk assessment function for each point s on the path, ∫ P γ(P(s))ds represents the cumulative risk value along path P; The neuroprotection path planning suggestion P path Converted to a format suitable for augmented reality display, the expression is: Among them, A R (x, y, z) indicates whether a certain point (x, y, z) in three-dimensional space belongs to the recommended surgical path; Overlay the path mask A in the field of view of the surgical microscope R , and dynamically update the display content according to the real-time positioning coordinates to obtain a real-time surgical view V with AR auxiliary markings AR .

8. The real-time electronic processing system for neurophysiological signals during brain surgery according to claim 7, characterized in that: The real-time surgical view V with AR auxiliary marking AR Provides clear operation guidance to help doctors avoid critical nerve functional areas and perform precise resection operations; The real-time surgical view V with AR auxiliary marking AR Feedback is provided to the doctor as the main navigation tool during the operation, while all operation trajectories and related data are recorded.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the real-time electronic processing system for neuroelectrophysiological signals in brain surgery according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the real-time electronic processing system for neuroelectrophysiological signals in brain surgery according to any one of claims 1 to 8 are implemented.

Images