Electroencephalogram epilepsy detection method and system based on node adaptive graph neural network

Abstract

The application discloses an electroencephalogram epilepsy detection method and system based on a node adaptive graph neural network, relates to a computer system based on a biological model, and is proposed in view of problems such as fixed graph structure in the prior art. The method comprises the following steps: an electroencephalogram signal acquisition and preprocessing step; mixed EEG graph construction; adaptive residual graph optimization; node-specific diffusion convolution; time sequence feature modeling; and classification output. The system comprises a data acquisition module, a mixed graph construction module, an adaptive graph construction module, a node-specific convolution module, a time sequence modeling module, and a classification output module; when the system is running, the method steps are executed, so that the electroencephalogram epilepsy detection function based on the node adaptive graph neural network is realized. Technical advantages include: (1) graph structure self-learning; (2) brain region personalized modeling, and strengthened regional feature expression; and (3) combined space-time dependence modeling, and complete description of the epilepsy dynamic process.

Description

Technical Field

[0001] This invention relates to computer systems based on biological models, and more particularly to a method and system for detecting epilepsy based on node adaptive graph neural networks. Background Technology

[0002] Epilepsy is a chronic neurological disorder caused by abnormal electrical activity in the brain, affecting approximately 50 million people worldwide. Epileptic seizures manifest in a variety of ways, ranging from brief loss of consciousness to severe tonic-clonic convulsions. Accurate identification of seizures and their types is crucial for guiding individualized antiepileptic drug treatment and surgical options.

[0003] Electroencephalography (EEG), as a non-invasive method for monitoring neural electrical activity, has advantages such as high temporal resolution, real-time performance, and repeatable acquisition, and is a core data source for the detection and classification of epileptic seizures.

[0004] However, traditional EEG signal analysis still mainly relies on the manual visual interpretation of neurophysiologists, which has the following obvious problems:

[0005] (1) High subjectivity and low efficiency: Manual interpretation of EEG requires browsing signals frame by frame for hours or even tens of hours, which is a huge workload and results are inconsistent.

[0006] (2) Significant reliance on experience: Different experts have different criteria for judging epileptiform discharges (such as spikes and sharp waves), which affects the accuracy and repeatability of the detection;

[0007] (3) Difficult to achieve real-time monitoring: In the long-term video EEG monitoring process, the cost of manual review is extremely high, making it difficult to meet the needs of real-time seizure detection and clinical monitoring.

[0008] To improve the efficiency of EEG signal analysis, researchers have attempted to introduce machine learning and deep learning methods into automated EEG analysis. Early methods were mostly based on handcrafted features (such as band energy, sample entropy, Hjorth parameters, etc.) and combined with traditional classifiers such as support vector machines (SVM), decision trees, or random forests to detect epileptic seizures. While these methods improved automation to some extent, their feature design relied on experience, resulting in poor generalization ability and difficulty in maintaining stable performance under different patients and data collection conditions.

[0009] With the development of deep learning, Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs / LSTMs) have been widely used for automatic feature extraction and temporal modeling of EEG signals. CNNs can capture the local spatial and spectral features of EEG segments, while LSTM models can model the long-term dependencies of signals. However, these networks all assume that the input data is distributed on a regular Euclidean grid structure, while the spatial distribution of EEG electrodes on the scalp surface is an irregular topological structure, and the functional connections between different brain regions also exhibit non-Euclidean relationships. Therefore, traditional CNN or RNN models are unable to fully characterize the topological dependencies and functional interactions between EEG channels, limiting their ability to model dynamic activities between brain regions.

[0010] In recent years, the emergence of Graph Neural Networks (GNNs) has provided a new technical approach for processing non-Euclidean structural data. Researchers treat EEG channels as graph nodes, defining the spatial distance or signal correlation between channels as graph edges, and using graph convolution operations to propagate information between brain regions. Typical methods include Diffusion Convolutional Recurrent Neural Networks (DCRNNs) and Spatio-Temporal Graph Convolutional Networks (STGCNs). These methods have made some progress in epilepsy detection tasks, but still have the following limitations:

[0011] (1) Graph structure fixation: Existing models usually rely on static graph construction methods (such as generating a fixed adjacency matrix based on electrode Euclidean distance or static Pearson correlation coefficient), which cannot adaptively adjust the graph structure according to the dynamic characteristics of patient differences, state changes or attack stages, thus limiting the model's individualization and time-varying adaptability.

[0012] (2) Node parameter sharing problem: Most GNN models use uniform convolution parameters for all nodes, failing to consider the differences in neural activity patterns and physiological functions of different brain regions, thus weakening the model's performance in capturing local brain region features and heterogeneous seizure patterns.

[0013] (3) Insufficient time-dependent modeling: Some GNN methods only focus on spatial structure information and fail to fully couple with the dynamic changes of time series, making it difficult to fully characterize the propagation pattern of brain regions before and after epileptic seizures.

[0014] (4) Insufficient clinical interpretability: The features output by existing models often lack a correspondence with physiological brain regions or clinical indicators, which limits their application value in auxiliary diagnosis and clinical decision-making.

[0015] In summary, existing epilepsy detection technologies still have significant shortcomings in areas such as adaptive modeling of brain region connectivity structures, learning of node-specific parameters, and spatiotemporal joint feature extraction, which limits detection accuracy, generalization ability, and clinical interpretability.

[0016] Therefore, there is an urgent need for a new intelligent EEG analysis method that can adaptively learn brain region connectivity while maintaining physiological structural constraints, support node-level differential modeling, and achieve spatiotemporally dependent joint representation, so as to improve the accuracy, robustness, and clinical applicability of epileptic seizure detection and type classification. Summary of the Invention

[0017] The purpose of this invention is to provide a method and system for detecting epilepsy based on node adaptive graph neural networks, so as to solve the problems existing in the prior art.

[0018] The EEG epilepsy detection method based on node adaptive graph neural network described in this invention includes the following steps:

[0019] S100: EEG signal acquisition and preprocessing steps: Acquire multi-channel EEG signals from the subject and preprocess the raw signals;

[0020] S200: Hybrid EEG Map Construction Steps: A hybrid map integrating anatomical structure and functional correlation is established between brain electrical channels to characterize the spatial and functional dependencies between brain regions.

[0021] S300: Adaptive Residual Graph Optimization Steps: To address the dynamic changes in brain region connectivity patterns across different patients and time periods, an adaptive residual graph optimization mechanism is introduced, refining the initial graph structure through node embedding learning;

[0022] S400: Node-specific diffusing convolution steps: Utilizing the node-specific diffusing convolution mechanism, different brain regions are equipped with independent convolution kernel parameters;

[0023] S500: Temporal feature modeling steps: Based on node-specific diffusion convolution, a diffusion graph convolution recurrent unit is introduced to capture the dynamic features of brain region activity changing over time, thereby realizing the temporal modeling of the epileptic seizure process; the diffusion graph convolution recurrent unit embeds a diffusion convolution operator in a gated recurrent structure, enabling the model to simultaneously model the spatial diffusion between brain regions and the evolution of signals over time, thereby achieving a joint representation of EEG signal features in the spatiotemporal domain;

[0024] S600: Classification output steps.

[0025] The EEG epilepsy detection system based on node adaptive graph neural network described in this invention includes the following functional modules:

[0026] Data acquisition module: Used to receive multi-channel EEG signals input from the EEG acquisition device and perform preprocessing operations;

[0027] Hybrid graph construction module: used to calculate the spatial and functional similarity between electrodes and generate a hybrid initial adjacency matrix;

[0028] Adaptive mapping module: Performs residual optimization based on node embedding to dynamically adjust the connectivity of brain regions;

[0029] Node-specific convolution module: Performs node-specific diffusing convolution to achieve differential modeling at the brain region level;

[0030] Temporal modeling module: Employs a diffusing convolutional loop structure to capture the temporal evolution of epileptic seizures;

[0031] Classification output module: Detects seizures or identifies seizure types based on global EEG features and outputs prediction results;

[0032] When the EEG epilepsy detection system is running, steps S100 to S600 in the method are executed sequentially to realize the EEG epilepsy detection function based on node adaptive graph neural network.

[0033] The EEG epilepsy detection method and system based on node adaptive graph neural network described in this invention has the following technical advantages:

[0034] (1) Graph structure self-learning overcomes static limitations. Through a residual optimization mechanism driven by node embedding, adaptive learning and dynamic adjustment of brain region connectivity are achieved. During training, the adjacency matrix structure can be automatically corrected according to the EEG characteristics of different subjects and time segments, thereby overcoming the lack of adaptability caused by the reliance on fixed graph structures in traditional methods.

[0035] (2) Personalized brain region modeling to enhance regional feature representation. A node-specific parameter learning mechanism is introduced into the diffusion map convolution framework, enabling each EEG channel to independently learn convolution weights and feature transformation functions. This fully reflects the physiological differences and functional heterogeneity between brain regions, helping the model to accurately capture the characteristic patterns and participation levels of different brain regions in epileptic seizures.

[0036] (3) Joint spatiotemporal dependency modeling to fully depict the dynamic process of epilepsy. Based on spatial diffusion convolution, a recurrent neural structure is introduced to construct a spatiotemporal joint modeling framework. It can simulate the diffusion and propagation law of EEG signals in space, and capture the dynamic changes of the time series before and after the seizure, realizing the unified modeling of epileptic activity in both time and space dimensions. Attached Figure Description

[0037] Figure 1 This is a schematic flowchart of the brainwave epilepsy detection method described in this invention.

[0038] Figure 2 This is a flowchart illustrating step S200 of the electroencephalogram (EEG) epilepsy detection method described in this invention.

[0039] Figure 3 This is a flowchart illustrating step S300 of the electroencephalogram (EEG) epilepsy detection method described in this invention.

[0040] Figure 4 This is a flowchart illustrating step S400 of the electroencephalogram (EEG) epilepsy detection method described in this invention.

[0041] Figure 5 This is a flowchart illustrating step S500 of the electroencephalogram (EEG) epilepsy detection method described in this invention.

[0042] Figure 6 This is a schematic diagram of the structure of the EEG epilepsy detection system described in this invention. Detailed Implementation

[0043] like Figures 1 to 5 As shown, the EEG epilepsy detection method based on node adaptive graph neural network described in this invention includes the following steps:

[0044] S100: EEG signal acquisition and preprocessing steps;

[0045] Multichannel EEG signals were collected from the subjects, and the raw signals were preprocessed:

[0046] A 0.5–70 Hz bandpass filter and a 50 / 60 Hz notch filter are used to remove DC drift and power frequency noise from the original signal.

[0047] Independent Component Analysis (ICA) is used to remove artifacts such as electrooculography (EOG) and electromyography (EMG); the independent component analysis is also known as ICA.

[0048] Dividing the data into segments based on fixed time windows to form EEG signal tensors

[0049]

[0050] in Indicates the number of brainwave channels. This indicates the number of sampling points for each segment; Indicates the first The signal sequence of each channel within this time window.

[0051] Z-score normalization is performed on each signal segment. To reduce the difference in signal amplitude among different subjects, among which For channel The mean, For channel The standard deviation.

[0052] S200: Steps for constructing a hybrid EEG graph;

[0053] A hybrid map integrating anatomical structures and functional correlations is constructed between brain electrical pathways to characterize spatial and functional dependencies between brain regions. This includes the following sub-steps:

[0054] S201: Calculate the Euclidean distance between the electrodes to obtain the spatial similarity matrix. ;

[0055] No. The electrodes corresponding to each channel are represented in three-dimensional space coordinates as follows: The spatial distance between any two channels is defined as:

[0056]

[0057] Spatial similarity is defined as:

[0058]

[0059] in Indicates channel Spatial connection strength with channel j; The scaling parameter of the Gaussian kernel is used to control the distance decay rate.

[0060] S202: Calculate the functional similarity matrix between EEG channels :

[0061] Functional similarity, based on the Pearson correlation coefficient, is defined as follows:

[0062]

[0063] in For channel The signal at time step t, For channel The average value of this segment, The range of values ​​is , indicating channel The correlation strength between j signals.

[0064] S203: Combine spatial similarity and functional similarity to obtain a hybrid initial adjacency matrix. :

[0065]

[0066] in, This is a learnable weight matrix used to control the fusion ratio of the spatial graph; It is a learnable weight matrix used to control the fusion ratio of the function graphs; This is the bias matrix; This is the Sigmoid function, used to normalize the output; This is element-wise multiplication.

[0067] S300: Adaptive residual map optimization steps;

[0068] To address the dynamic changes in brain region connectivity patterns across different patients and time periods, this invention designs a residual adaptive graph optimization mechanism that refines the initial graph structure through node embedding learning. Specifically, it includes the following sub-steps:

[0069] S301: Define the node embedding matrix ,in Embedding dimension for nodes, row vector Indicates the first The potential representation of each node;

[0070] S302: Calculate the node similarity matrix :

[0071]

[0072] in It is a linear rectified function used to ensure that the similarity is non-negative.

[0073] S303: Through a learnable mapping matrix Projection similarity, and forming residuals with the initial image:

[0074]

[0075] S304: Update to obtain the adaptive adjacency matrix:

[0076]

[0077] in For adaptive correction terms, This is the final adaptive adjacency matrix.

[0078] Through this residual form, the model only needs to learn the relative correction amount, thereby achieving individualized adjustment of brain region connectivity while ensuring structural stability.

[0079] S400: Node-specific diffraction convolution step;

[0080] To address the issue of node parameter sharing in traditional graph convolution, this invention proposes a node-specific diffusing convolution mechanism, enabling different brain regions to have independent convolution kernel parameters. Specifically, it includes the following sub-steps:

[0081] S401: Using node row vectors Input lightweight multilayer perceptron Generate a set of independent diffusing convolution kernel parameters for each node.

[0082]

[0083] in Represents a node The k-th order diffusion weight matrix, For the input feature dimension, This is the output feature dimension.

[0084] S402: In the adaptive adjacency matrix Perform bidirectional diffusing convolution on top:

[0085]

[0086] in Output the feature matrix for each node. Let the diffusion order be , Forward-diffused nuclei, It is a reverse-diffusion nucleus.

[0087] S500: Steps for modeling time-series features;

[0088] To achieve time-dependent modeling of EEG signals, this invention introduces a diffusion graph convolutional recurrent unit (DCGRU) based on node-specific diffusion convolution. This DCGRU captures the dynamic characteristics of brain region activity over time, enabling temporal modeling of the epileptic seizure process. The DCGRU embeds diffusion convolution operators within a gated recurrent structure, allowing the model to simultaneously model spatial diffusion between brain regions and signal evolution over time, achieving a joint representation of EEG signal features in the spatiotemporal domain. Specifically, it includes the following sub-steps:

[0089] S501: Update gate calculation;

[0090] The update gate is used to control the proportion of the current neural state in a brain region that comes from the historical state.

[0091] At time step The formula for updating the gate is:

[0092]

[0093] in Indicates the current time step. This represents the EEG input feature matrix at the current moment. This represents the hidden state of the previous time step. In the adaptive adjacency matrix Node-specific diffraction convolution operation performed on , To update the input weight matrix of the gate, To update the state weight matrix of the gate, For the Sigmoid function, To update the gate matrix.

[0094] In electroencephalogram (EEG) signal analysis, The degree to which the control model preserves the state of old brain regions: when the EEG is in a stable phase, When the value is small, the model retains its historical state; the stable phase is the non-ictal period. When abnormal discharges occur in a brain region... As the value increases, the model rapidly updates its features, thereby capturing sudden changes in epileptic seizures; the abnormal discharge refers to the seizure period.

[0095] S502: Reset door calculation;

[0096] The reset gate is used to adjust the influence of the previous state on the current candidate state, and its calculation formula is as follows:

[0097]

[0098] in To reset the input weight matrix of the gate, This is the state weight matrix for resetting the gate; other symbols are defined as above.

[0099] In EEG signal analysis, the value of the reset gate Reflecting the historical inhibitory relationship between brain regions: when the brain region is unrelated to the current seizure, the corresponding Values ​​approaching 0 suppress the influence of irrelevant pathways; when brain regions are involved in seizure propagation, Increasing the value enhances the feature update capability of the region.

[0100] S503: Calculation of candidate hidden states;

[0101] The candidate hidden state is generated jointly by the current input feature and the previous state after being adjusted by the reset gate. Its calculation formula is as follows:

[0102]

[0103] in, This is the spread convolution weight matrix corresponding to the input signal, used to weight the input feature X at the current time step. t Projected into the hidden feature space; The spread convolution weight matrix corresponds to the historical hidden states and is used in the adaptive adjacency matrix. The temporal propagation and updating of execution state characteristics; This represents element-wise multiplication; It is the hyperbolic tangent function; Represents node-specific diffraction convolution operation .

[0104] In EEG modeling, This represents the potential neural activity patterns in brain regions at the current moment. This is achieved by introducing node-specific diffraction convolution operations. Each brain region, when updating its own state, not only relies on its historical signals, but also combines the dynamic connection weights with other brain regions to achieve feature diffusion and collaborative updating between brain regions, thereby characterizing the propagation process of epilepsy signals in the brain region network.

[0105] S504: Hidden state update;

[0106] The hidden state update formula output by the final convolutional recurrent unit of the diffusion graph is:

[0107]

[0108] in Indicates time The hidden state feature matrix, with other symbols defined as above.

[0109] In the semantics of EEG signals, the two parts in the formula have the following meanings: The representative model represents the memory and retention of historical brain region states. The representative model responds to and updates to current sudden neural activity.

[0110] By dynamically adjusting the gating ratio, the model can flexibly balance historical memory and new activation according to the EEG characteristics at different time stages, effectively depicting the occurrence, spread and recovery process of epileptic seizures.

[0111] S600: Classification output steps; specifically including the following sub-steps:

[0112] S601: After the time series feature modeling is completed, perform classification output.

[0113] The hidden state characteristic matrix at time T Perform global average pooling to obtain fragment-level feature vectors:

[0114]

[0115] S602: Global features are input into the fully connected layer and classified using the Softmax function, outputting results for epileptic seizure detection or epilepsy type determination.

[0116]

[0117] in This is the output probability vector; For the number of categories, in the detection task In classification tasks ; These are the output layer weights; This is the bias parameter.

[0118] The present invention also provides an EEG epilepsy detection system that implements the above method, such as... Figure 6 As shown, it includes the following functional modules:

[0119] Data acquisition module: used to receive multi-channel EEG signals input from the EEG acquisition device and perform preprocessing operations such as filtering, segmentation, and standardization;

[0120] Hybrid graph construction module: used to calculate the spatial and functional similarity between electrodes and generate a hybrid initial adjacency matrix;

[0121] Adaptive mapping module: Performs residual optimization based on node embedding to dynamically adjust the connectivity of brain regions;

[0122] Node-specific convolution module: Performs node-specific diffusing convolution to achieve differential modeling at the brain region level;

[0123] Temporal modeling module: Employs a diffusing convolutional loop structure to capture the temporal evolution of epileptic seizures;

[0124] Classification output module: Detects seizures or identifies seizure types based on global EEG features and outputs prediction results.

[0125] Each module can be a software module or a hardware circuit module. When the EEG epilepsy detection system is running, steps S100 to S600 are executed sequentially to realize the EEG epilepsy detection function based on a node adaptive graph neural network.

[0126] For those skilled in the art, various other corresponding changes and modifications can be made based on the technical solutions and concepts described above, and all such changes and modifications should fall within the protection scope of the claims of this invention.

Claims

1. A method for detecting epilepsy via electroencephalography based on node adaptive graph neural networks, characterized in that, Includes the following steps: S100: EEG signal acquisition and preprocessing steps: Acquire multi-channel EEG signals from the subject and preprocess the raw signals; S200: Hybrid EEG Map Construction Steps: A hybrid map integrating anatomical structures and functional correlations is established between brain electrical channels to characterize the spatial and functional dependencies between brain regions. S300: Adaptive Residual Graph Optimization Steps: To address the dynamic changes in brain region connectivity patterns across different patients and time periods, an adaptive residual graph optimization mechanism is introduced, refining the initial graph structure through node embedding learning; S400: Node-specific diffusing convolution steps: Utilizing the node-specific diffusing convolution mechanism, different brain regions are equipped with independent convolution kernel parameters; S500: Temporal feature modeling steps: Based on node-specific diffusion convolution, a diffusion graph convolution recurrent unit is introduced to capture the dynamic features of brain region activity changing over time, thereby realizing the temporal modeling of the epileptic seizure process; the diffusion graph convolution recurrent unit embeds a diffusion convolution operator in a gated recurrent structure, enabling the model to simultaneously model the spatial diffusion between brain regions and the evolution of signals over time, thereby achieving a joint representation of EEG signal features in the spatiotemporal domain; S600: Classification output steps; Step S200 includes the following sub-steps: S201: Calculate the Euclidean distance between the electrodes to obtain the spatial similarity matrix. ; No. The electrodes corresponding to each channel are represented in three-dimensional space coordinates as follows: The spatial distance between any two channels is defined as: Spatial similarity is defined as: in Indicates channel Spatial connection strength with channel j; The scale parameter of the Gaussian kernel is used to control the distance decay rate; S202: Calculate the functional similarity matrix between EEG channels : Functional similarity, based on the Pearson correlation coefficient, is defined as follows: in For channel The signal at time step t, For channel At the average value of the current segment, The range of values ​​is , indicating channel The correlation strength between j signals; S203: Combine spatial similarity and functional similarity to obtain a hybrid initial adjacency matrix. : in, This is a learnable weight matrix used to control the fusion ratio of the spatial graph; It is a learnable weight matrix used to control the fusion ratio of the function graphs; This is the bias matrix; This is the Sigmoid function, used to normalize the output; This is element-wise multiplication; Step S300 includes the following sub-steps: S301: Define the node embedding matrix ,in Embedding dimension for nodes, row vector Indicates the first The potential representation of each node; S302: Calculate the node similarity matrix : in It is a linear rectified function used to ensure that the similarity is non-negative; S303: Through a learnable mapping matrix Projection similarity, and forming residuals with the initial image: S304: Update to obtain the adaptive adjacency matrix: in For adaptive correction terms, This is the final adaptive adjacency matrix; Step S400 includes the following sub-steps: S401: Using node row vectors Input lightweight multilayer perceptron Generate a set of independent diffusing convolution kernel parameters for each node. in Represents a node The k-th order diffusion weight matrix, For the input feature dimension, Output feature dimension; S402: In the adaptive adjacency matrix Perform bidirectional diffusing convolution on top: in Output the feature matrix for each node. Let be the diffusion order. Forward-diffused nuclei, It is a reverse-diffusion nucleus.

2. The method for detecting epilepsy based on node adaptive graph neural networks according to claim 1, characterized in that, Step S100 specifically involves: using a 0.5–70 Hz bandpass filter and a 50 / 60 Hz notch filter to remove DC drift and power frequency noise from the original signal; using independent component analysis to remove artifacts; and dividing the signal into segments according to a fixed time window to form an EEG signal tensor. ; in Indicates the number of brainwave channels. This indicates the number of sampling points for each segment; Indicates the first The signal sequence of each channel within this time window; Z-score normalization is performed on each signal segment. To reduce the difference in signal amplitude among different subjects, among which For channel The mean, For channel The standard deviation.

3. The method for detecting epilepsy based on node adaptive graph neural networks according to claim 2, characterized in that, Step S500 includes the following sub-steps: S501: Update gate calculation; The update gate is used to control the proportion of the current neural state in a brain region that comes from a historical state; At time step The formula for updating the gate is: in Indicates the current time step. This represents the EEG input feature matrix at the current moment. This represents the hidden state of the previous time step. In the adaptive adjacency matrix Node-specific diffraction convolution operation performed on , To update the input weight matrix of the gate, To update the state weight matrix of the gate, For the Sigmoid function, To update the gate matrix; In electroencephalogram (EEG) signal analysis, The degree to which the control model retains the state of previous brain regions: when the EEG is in a stable phase, the model retains the historical state; when abnormal discharges occur in the brain region, As the value increases, the model updates its features, thereby capturing sudden changes in epileptic seizures; S502: Reset door calculation; The reset gate is used to adjust the influence of the previous state on the current candidate state. The calculation formula is as follows: in To reset the input weight matrix of the gate, To reset the state weight matrix of the gate; In EEG signal analysis, the value of the reset gate Reflecting the historical inhibitory relationship between brain regions: when the brain region is unrelated to the current seizure, the corresponding The value approaches 0 to suppress the influence of irrelevant channels; When brain regions are involved in seizure propagation Increasing the value enhances the feature update strength of the region involved in the outbreak spread; S503: Calculation of candidate hidden states; The candidate hidden state is generated jointly by the current input feature and the previous state after being adjusted by the reset gate. Its calculation formula is as follows: in, This is the spread convolution weight matrix corresponding to the input signal, used to weight the input feature X at the current time step. t Projected into the hidden feature space; The spread convolution weight matrix corresponds to the historical hidden states and is used in the adaptive adjacency matrix. The temporal propagation and updating of execution state characteristics; This represents element-wise multiplication; It is the hyperbolic tangent function; Represents node-specific diffraction convolution operation ; S504: Hidden state update; The hidden state update formula output by the final convolutional recurrent unit of the diffusion graph is: in Indicates time The latent state feature matrix.

4. The method for detecting epilepsy based on node adaptive graph neural networks according to claim 3, characterized in that, Includes the following sub-steps: S601: After the temporal feature modeling is completed, perform classification output; The feature matrix of the hidden state at time T Perform global average pooling to obtain fragment-level feature vectors: S602: Global features are input into the fully connected layer and classified using the Softmax function, outputting results for epileptic seizure detection or epilepsy type determination. in This is the output probability vector; For the number of categories, in the detection task In classification tasks ; These are the output layer weights; This is the bias parameter.

5. A brainwave epilepsy detection system based on node adaptive graph neural network, characterized in that, Includes the following functional modules: Data acquisition module: Used to receive multi-channel EEG signals input from the EEG acquisition device and perform preprocessing operations; Hybrid graph construction module: used to calculate the spatial and functional similarity between electrodes and generate a hybrid initial adjacency matrix; Adaptive mapping module: Performs residual optimization based on node embedding to dynamically adjust the connectivity of brain regions; Node-specific convolution module: Performs node-specific diffusing convolution to achieve differential modeling at the brain region level; Temporal modeling module: Employs a diffusing convolutional loop structure to capture the temporal evolution of epileptic seizures; Classification output module: Detects seizures or identifies seizure types based on global EEG features and outputs prediction results; When the EEG epilepsy detection system is running, steps S100 to S600 of the method as described in any one of claims 1 to 4 are executed sequentially to realize the EEG epilepsy detection function based on node adaptive graph neural network.

6. The EEG epilepsy detection system based on node adaptive graph neural network according to claim 5, characterized in that, Each functional module can be either a software module or a hardware circuit module.

Images