EEG signal analysis methods, devices, electronic equipment and storage media

Abstract

This invention discloses a method, apparatus, electronic device, and storage medium for analyzing electroencephalogram (EEG) signals. The method includes: acquiring an EEG signal to be analyzed and preprocessing the EEG signal; extracting a target EEG signal from the preprocessed EEG signal and generating an EEG signal topographic map sequence based on the target EEG signal; and inputting the EEG signal topographic map sequence into a preset EEG signal analysis model to obtain a classification result of the EEG signal to be analyzed. The preset EEG signal analysis model includes a convolutional neural network module, a long short-term memory network module, and a fully connected network module. The technical solution of this invention solves the problem in existing EEG signal analysis algorithms where data features are not fully mined, leading to low recognition accuracy of target EEG signals. It achieves deep learning of EEG signal features, improving the accuracy of EEG signal analysis results.

Description

Technical Field

[0001] This invention relates to the field of electroencephalogram (EEG) signal processing technology, and more particularly to an EEG signal analysis method, apparatus, electronic device, and storage medium. Background Technology

[0002] Mild cognitive impairment (MCI) and Alzheimer's disease (AD) are neurological disorders at different stages of development. If MCI or AD can be detected and intervened in a timely manner, the progression of the disease can be slowed and the patient's quality of life can be improved.

[0003] Currently, most methods for identifying related diseases based on electroencephalogram (EEG) signals involve clustering analysis of the global energy spectrum of EEG signals to classify and identify microstates within the signals. However, this analysis does not consider the timing of each microstate's occurrence and fails to fully explore more signal features within the EEG signals. Therefore, the accuracy of current EEG-based disease identification algorithms still needs improvement. Summary of the Invention

[0004] This invention provides a method, apparatus, electronic device, and storage medium for analyzing electroencephalogram (EEG) signals. One of the technical problems it aims to solve is that the data features in the EEG signal analysis algorithm are not fully mined, resulting in low recognition accuracy of the target EEG signal. The invention aims to achieve deep learning of EEG signal features and improve the accuracy of EEG signal analysis results.

[0005] In a first aspect, embodiments of the present invention provide a method for analyzing electroencephalogram (EEG) signals, the method comprising:

[0006] Acquire the EEG signal to be analyzed and preprocess the EEG signal to be analyzed;

[0007] The target EEG signal is extracted from the preprocessed EEG signal to be analyzed, and an EEG signal topographic map sequence is generated based on the target EEG signal.

[0008] The EEG signal topographic map sequence is input into a preset EEG signal analysis model to obtain the classification result of the EEG signal to be analyzed. The preset EEG signal analysis model includes a convolutional neural network module, a long short-term memory network module, and a fully connected network module.

[0009] Secondly, embodiments of the present invention also provide an electroencephalogram (EEG) signal analysis device, the device comprising:

[0010] The signal preprocessing module is used to acquire the EEG signal to be analyzed and to preprocess the EEG signal to be analyzed.

[0011] The topographic map sequence generation module is used to extract target EEG signals from the preprocessed EEG signals to be analyzed, and generate an EEG signal topographic map sequence based on the target EEG signals.

[0012] The signal analysis module is used to input the EEG signal topographic map sequence into a preset EEG signal analysis model to obtain the classification result of the EEG signal to be analyzed. The preset EEG signal analysis model includes a convolutional neural network module, a long short-term memory network module, and a fully connected network module.

[0013] Thirdly, embodiments of the present invention also provide an electronic device, the electronic device comprising:

[0014] One or more processors;

[0015] Storage device for storing one or more programs.

[0016] When the one or more programs are executed by the one or more processors, the one or more processors implement the electroencephalogram (EEG) signal analysis method provided in any embodiment of the present invention.

[0017] Fourthly, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the electroencephalogram (EEG) signal analysis method provided in any embodiment of the present invention.

[0018] The technical solution of this invention involves preprocessing the acquired EEG signal to filter out noise data. Then, a target EEG signal is extracted from the preprocessed signal, and a sequence of EEG signal topographic maps is generated based on this target signal. This sequence of multiple EEG signal topographic maps arranged chronologically is input into a preset EEG signal analysis model to obtain the classification result of the EEG signal. The preset EEG signal analysis model includes a convolutional neural network (CNN) module, a long short-term memory (LSM) network module, and a fully connected network module. The CNN module extracts data features from the EEG signal topographic maps, the LSM module extracts and learns time-related features from the sequence, and finally, the fully connected layers analyze the data to obtain the analysis result. The network structure based on the preset EEG signal analysis model allows for a more thorough mining of data features in the EEG signal, resulting in more accurate analysis results. The technical solution of this invention solves the problem that the data features in existing EEG signal analysis algorithms are not fully mined, resulting in low recognition accuracy of target EEG signals. It realizes deep learning of EEG signal features and improves the accuracy of EEG signal analysis results.

[0019] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of exemplary embodiments of the present invention, the accompanying drawings used in describing the embodiments are briefly introduced below. Obviously, the accompanying drawings described are only a portion of the drawings of the embodiments to be described in this invention, and not all of the drawings. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0021] Figure 1 A flowchart of an electroencephalogram (EEG) signal analysis method provided in an embodiment of the present invention;

[0022] Figure 2 This is a schematic diagram of a global energy spectrum peak signal provided in an embodiment of the present invention;

[0023] Figure 3 This is a schematic diagram of an electroencephalogram (EEG) signal topography sequence provided in an embodiment of the present invention;

[0024] Figure 4 A flowchart of an electroencephalogram (EEG) signal analysis method provided in an embodiment of the present invention;

[0025] Figure 5 This is a schematic diagram of a preprocessed electroencephalogram (EEG) signal provided in an embodiment of the present invention;

[0026] Figure 6 This is a schematic diagram of a primary preprocessed electroencephalogram (EEG) signal provided in an embodiment of the present invention.

[0027] Figure 7 A flowchart of an electroencephalogram (EEG) signal analysis method provided in an embodiment of the present invention;

[0028] Figure 8 This is a schematic diagram of an electroencephalogram (EEG) signal analysis model provided in an embodiment of the present invention;

[0029] Figure 9 This is a schematic diagram of the structure of an electroencephalogram (EEG) signal analysis device provided in an embodiment of the present invention;

[0030] Figure 10 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0031] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the invention.

[0032] At the same time, it should be understood that, for ease of description, the dimensions of the various parts shown in the accompanying drawings are not drawn according to actual scale.

[0033] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use. Techniques, methods, and apparatus known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and apparatus should be considered part of the specification.

[0034] It should be noted that similar labels and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be discussed further in subsequent figures.

[0035] Figure 1 This is a flowchart illustrating an electroencephalogram (EEG) signal analysis method provided in an embodiment of the present invention. This embodiment is applicable to scenarios involving the processing and analysis of EEG signals, particularly for identifying some neurological lesions based on EEG signals. The method can be executed by an EEG signal analysis device, which can be implemented through software and / or hardware and can be configured in a terminal and / or server to implement the EEG signal analysis method in this embodiment of the present invention.

[0036] like Figure 1 As shown, the EEG signal analysis method in this embodiment may specifically include:

[0037] S110. Acquire the EEG signal to be analyzed and preprocess the EEG signal to be analyzed.

[0038] The EEG signals to be analyzed are potential changes in the cerebral cortex generated by neuronal activity, collected by placing paired electrodes (such as eight-electrodes or sixteen-electrodes) on the scalp of the target subject. The target subject can be anyone undergoing a neurological function examination, including healthy individuals, individuals with cognitive impairment, individuals with Alzheimer's disease, and others requiring neurological function status assessment. The EEG signals to be analyzed serve as the data basis for analyzing the neurological function status of the target subject.

[0039] However, the acquired EEG signals to be analyzed contain noise signals such as body movement artifacts, electrooculography artifacts, electromyography artifacts, power line noise, and signal crosstalk between electrode channels. Preprocessing of the EEG signals is necessary to remove these interfering noise components. Preprocessed EEG signals can make the analysis results more closely approximate the true state of the EEG signals in subsequent signal analysis.

[0040] S120. Extract the target EEG signal from the preprocessed EEG signal to be analyzed, and generate an EEG signal topographic map sequence based on the target EEG signal.

[0041] Based on existing research in the field of EEG signal analysis, the brain neurological characteristics of different target populations will manifest as EEG signal characteristics concentrated in different frequency bands. During the analysis of the EEG signal to be analyzed, the frequency band signal with a high correlation to the EEG signal characteristics of the target population can be extracted as the target EEG signal.

[0042] For example, the preprocessed EEG signal to be analyzed can be bandpass filtered to obtain signals in the delta band [1-4Hz], theta band [4-8Hz], alpha band [8-14Hz], beta band [14-31Hz], and gamma band [31-49Hz]. When analyzing the microstates of EEG signals, the preprocessed alpha band target EEG signal is commonly used.

[0043] Furthermore, the target EEG signal can be standardized, and then the global energy spectrum of the standardized target EEG signal can be calculated.

[0044] The standardization process for the target EEG signal can be expressed as: EEG[channel, time] standardized value = (EEG[channel, time] - AVG(EEG[channel, time])) / Std(EEG[channel, time]). That is, the standardized value of each electrode signal channel at each sampling time point is defined as the difference between the sampled data of each electrode signal channel at each sampling time point and the mean of the sampled data of all channels at the corresponding sampling time; then, it is divided by the standard deviation of the sampled data of all sampling channels. The calculation of the global field power (GFP) of the standardized target EEG signal can be expressed as: GFP (Global field power) = Std(EEG[channel, time] standardized value), which is the standard deviation of the standardized target EEG signal.

[0045] After obtaining the global energy spectrum of the target EEG signal, peak data of the target energy spectrum can be extracted from the global energy spectrum according to a preset sampling strategy. In one optional implementation, a minimum time interval for selecting peak data can be set, and the local max function can be used to extract the peak data of the target energy spectrum from the global energy spectrum. The time interval between each peak data point of the target energy spectrum is greater than the preset minimum time interval, for example, one percent of the sampling rate. This avoids acquiring peak data of local jitter signals and reduces sampling noise. For example, the extracted peak data of the target energy spectrum can be as follows: Figure 2 The data shown is in Figure 2 In the graph, the horizontal axis represents time, and the vertical axis represents the GFP peak value. Furthermore, an EEG signal topographic map sequence can be generated based on the amplitude of the EEG signal to be analyzed corresponding to the peak energy spectrum data of each target, such as... Figure 3 The few listed brainwave topographic sequences.

[0046] In one optional implementation, an EEG topographic map is generated for the amplitude data corresponding to each target energy spectrum peak data. After obtaining the EEG signal topographic map sequence, all EEG topographic maps can be divided into multiple sub-topographic map sequence groups according to time sequence, and each sub-topographic map sequence group generates a corresponding topographic map sequence video. That is, a sequence containing all EEG signal topographic maps is grouped according to time information to obtain multiple sub-topographic map sequence groups, and the time information between each EEG signal topographic map is shown in video form, so that the data analysis object of the preset EEG signal analysis model is more granular and more EEG signal data features are systematically and fully explored.

[0047] S130. Input the EEG signal topographic map sequence into a preset EEG signal analysis model to obtain the classification result of the EEG signal to be analyzed.

[0048] The pre-defined EEG signal analysis model includes a convolutional neural network module, a long short-term memory network module, and a fully connected network module. The convolutional neural network module consists of three convolutional layers, with batch normalization and pooling layers following each convolutional layer.

[0049] The EEG signal topographic map sequence input into the preset EEG signal analysis model can be video data generated from all topographic maps, or video data generated from topographic maps in a sub-topographic map sequence group after data grouping. The EEG signal topographic map sequence first undergoes feature extraction through three convolutional layers in the preset EEG signal analysis model, with the output data from each convolutional layer undergoing batch normalization for optimization. Then, the data features extracted by the convolutional neural network module are input to the Long Short-Term Memory (LSTM) network module for learning, with the input dimension matching the number of topographic maps in the topographic map sequence. The LSM module has an inherent advantage in learning time-series data, capable of remembering feature information for extended periods and fully learning time-based features in the data, thereby uncovering more data features in the EEG signals. Finally, the output of the LSM module is processed through a multi-layer fully connected network to output the classification results of the EEG signals to be analyzed.

[0050] It should be noted that the classification result of the EEG signal to be analyzed depends on the training samples of the preset EEG signal analysis model during the training process. If the training samples are EEG signal topographic sequences obtained by processing EEG signals from a preset patient population, the classification result indicates whether the EEG signal to be analyzed belongs to the preset patient population. For example, if the training samples of the preset EEG signal analysis model are sampled from EEG signals of people with mild cognitive impairment, then the preset EEG signal analysis model, after training, can determine whether an EEG signal sequence belongs to people with mild cognitive impairment.

[0051] Specifically, when a large sequence containing all EEG signal topographic maps is grouped into multiple sub-topographic map sequence groups, multiple EEG signal topographic map sequences will be input into a preset EEG signal analysis model, resulting in multiple preliminary classification results. Statistical analysis can be performed on these preliminary classification results to obtain the final classification result. For example, if there are 1000 EEG topographic maps, and they are divided into sub-topographic map sequences of twenty maps each, in chronological order, 50 sub-topographic map sequences can be obtained. This means 50 EEG topographic map videos can be used as input to the preset EEG signal analysis model, resulting in 50 classification results. The number of classification results indicating EEG signals from the target population can be counted. If this number exceeds a certain value, the EEG signal being analyzed can be determined to be from the target population; otherwise, it is considered a non-target population EEG signal.

[0052] The technical solution of this embodiment involves preprocessing the acquired EEG signal to filter out noise data. Then, a target EEG signal is extracted from the preprocessed signal, and a sequence of EEG signal topographic maps is generated based on this target signal. This sequence of multiple EEG signal topographic maps arranged chronologically is input into a preset EEG signal analysis model to obtain the classification result of the EEG signal. The preset EEG signal analysis model includes a convolutional neural network module, a long short-term memory network module, and a fully connected network module. The convolutional neural network module extracts data features from the EEG signal topographic maps, the long short-term memory network module extracts and learns time-related feature information from the EEG signal topographic map sequence, and finally, the fully connected layer analyzes the data to obtain the analysis result. The network structure based on the preset EEG signal analysis model can more fully mine the data features in the EEG signal, thus obtaining a more accurate analysis result. The technical solution of this invention solves the problem that the data features in existing EEG signal analysis algorithms are not fully mined, resulting in low recognition accuracy of target EEG signals. It realizes deep learning of EEG signal features and improves the accuracy of EEG signal analysis results.

[0053] Figure 4 This is a flowchart illustrating an electroencephalogram (EEG) signal analysis method provided in an embodiment of the present invention. This method can be combined with various optional schemes in the EEG signal analysis methods provided in the above embodiments, further describing the preprocessing process of EEG signals. Through the technical solution of this embodiment, the preprocessing effect of removing motion artifacts can be improved. This method can be executed by an EEG signal analysis device, which can be implemented in software and / or hardware, and can be configured in an electronic device, such as a mobile terminal or server device.

[0054] like Figure 4 As shown, the EEG signal analysis method in this embodiment may specifically include:

[0055] S210. Acquire the EEG signal to be analyzed, and remove the body motion artifact signal of the EEG signal to be analyzed based on the optimized EEG signal artifact removal method to obtain the EEG signal to be analyzed after initial preprocessing.

[0056] In the EEG signals to be analyzed, the noise with the largest value is usually motion artifact noise. However, conventional algorithms such as Artifact Subspace Reconstruction (ASR), Independent Component Analysis (ICA), and Common Spatial Pattern (CSP) cannot remove motion artifacts, which will have a certain impact on subsequent signal analysis.

[0057] In this embodiment, the conventional ASR algorithm is optimized to remove motion artifact noise. Specifically, in the process of removing motion artifact noise, firstly, the EEG signal to be analyzed is divided into multiple data segments of a preset data length. In this process, each data segment is arranged in chronological order by default. The length of the data segment can be set according to an empirical value, or determined based on the efficiency and effect analysis of data processing. For example, for the data of each motor channel, the length of the data segment can be set to one-tenth of the sampling rate. Then, data filtering is performed based on the standard deviation of the EEG signal amplitude in each data segment and the standard deviation of the amplitude of all the EEG signals to be analyzed. The filtered data segments are then spliced ​​together in chronological order to obtain the EEG signal to be analyzed after initial preprocessing. That is, data with large noise amplitudes are directly deleted. Furthermore, during the data filtering process, according to the signal acquisition time sequence, for the first data segment, if the standard deviation of the EEG signal amplitude of the first data segment is greater than a first preset multiple of the standard deviation of the amplitudes of all EEG signals to be analyzed, the first data segment is deleted. For each data segment other than the first data segment, if the standard deviation of the EEG signal amplitude of the current data segment is greater than a second preset multiple of the standard deviation of the EEG signal amplitude of the previous data segment, the current data segment is deleted. The first and second preset multiples are usually set to the same value, such as three times; however, corresponding values ​​can also be set according to data processing requirements. EEG signal images before and after motion artifact removal can be found in [reference needed]. Figure 5 and Figure 6 The diagram shown is a signal schematic.

[0058] S220. Perform a second-stage preprocessing on the EEG signal to be analyzed after the initial preprocessing to obtain the preprocessed EEG signal to be analyzed.

[0059] In this step, the EEG signals to be analyzed in the initial preprocessing are further denoised, specifically by removing artifact signals other than body movement artifact signals and noise data.

[0060] For example, the ICA algorithm is used to remove artifacts such as electrooculography (EOG) and electromyography (EMG); then, a bandstop filter is used to remove power frequency interference signals; and a bandpass filter is used for denoising to obtain a signal of 0.5-45Hz (or other EEG signals in the frequency band of interest during data analysis). Finally, the 0.5-45Hz EEG signal is averaged and rereferenced to remove crosstalk between electrode channels, resulting in a preprocessed EEG signal for analysis.

[0061] The average rereference processing involves subtracting the average value of the sampling data from all electrode sampling channels at the corresponding time point from the sampling data of each electrode sampling channel at each time point.

[0062] S230. Extract the target EEG signal from the preprocessed EEG signal to be analyzed, and generate an EEG signal topographic map sequence based on the target EEG signal.

[0063] S240. Input the EEG signal topographic map sequence into a preset EEG signal analysis model to obtain the classification result of the EEG signal to be analyzed.

[0064] The technical solution of this embodiment preprocesses the EEG signal after acquisition. First, it uses an algorithm optimized based on the ASR algorithm to remove motion artifacts and other noise signals, thus increasing the amount of effective information in the EEG signal. Then, it extracts the target EEG signal from the preprocessed signal and generates an EEG signal topography sequence based on the target signal. This sequence of multiple EEG signal topography maps arranged chronologically is input into a pre-defined EEG signal analysis model to obtain the classification result of the EEG signal. The pre-defined EEG signal analysis model includes a convolutional neural network module, a long short-term memory network module, and a fully connected network module. The convolutional neural network module extracts data features from the EEG signal topography map, the long short-term memory network module extracts and learns time-related features from the EEG signal topography map sequence, and finally, the fully connected layer analyzes the data to obtain the analysis result. The network structure based on the pre-defined EEG signal analysis model can more fully mine the data features in the EEG signal, resulting in more accurate analysis results. The technical solution of this invention solves the problem that the data features in existing EEG signal analysis algorithms are not fully mined, resulting in low recognition accuracy of target EEG signals. It realizes deep learning of EEG signal features and improves the accuracy of EEG signal analysis results.

[0065] Figure 7This is a flowchart illustrating an electroencephalogram (EEG) signal analysis method provided in an embodiment of the present invention. This method can be combined with various optional schemes in the EEG signal analysis methods provided in the above embodiments, further describing the training process of a preset EEG signal analysis model. Through this technical solution, a neural network model that more fully mines the data features of EEG signals can be obtained for EEG signal analysis and recognition. This method can be executed by an EEG signal analysis device, which can be implemented in software and / or hardware, and can be configured in an electronic device, such as a mobile terminal or server device.

[0066] like Figure 7 As shown, specific methods for analyzing electroencephalogram (EEG) signals may include:

[0067] S310. Acquire the electroencephalogram (EEG) signals of the preset diseased population and healthy control group, and preprocess the EEG signals.

[0068] The target population for EEG signal analysis can be individuals with mild cognitive impairment, Alzheimer's disease, or other neurological disorders. During EEG signal acquisition, multiple channels of EEG data are collected using a number of electrodes, with sampling rates typically set to 256Hz or 512Hz. The acquired EEG signals are then preprocessed to remove artifacts and noise.

[0069] For example, based on the optimized EEG signal artifact removal method, body movement artifacts are removed from the EEG signal to obtain a pre-processed EEG signal. Then, the pre-processed EEG signal to be analyzed is further denoised, specifically by removing other artifacts besides body movement artifacts and noise data.

[0070] S320. Extract the target EEG signal from the preprocessed EEG signal and generate an EEG signal topography sequence based on the target EEG signal.

[0071] Based on existing research in the field of EEG signal analysis, the brain neurological characteristics of different target populations will manifest as EEG signal characteristics concentrated in different frequency bands. During the analysis of the EEG signal to be analyzed, the frequency band signal with a high correlation to the EEG signal characteristics of the target population can be extracted as the target EEG signal.

[0072] Furthermore, the target EEG signal can be standardized, and then the global energy spectrum of the standardized target EEG signal can be calculated.

[0073] After obtaining the global energy spectrum of the target EEG signal, peak data of the target energy spectrum can be extracted from the global energy spectrum according to a preset sampling strategy. In one optional implementation, a minimum time interval for selecting peak data can be set, and the local max function can be used to extract the peak data of the target energy spectrum from the global energy spectrum. In another optional implementation, an EEG topographic map is generated for the amplitude data corresponding to each peak data of the target energy spectrum. After obtaining the EEG signal topographic map sequence, all EEG topographic maps can be divided into multiple sub-topographic map sequence groups according to time sequence, and each sub-topographic map sequence group can generate a corresponding topographic map sequence video. That is, a large sequence containing all EEG signal topographic maps is grouped according to time information to obtain multiple sub-topographic map sequence groups, and the time information between each EEG signal topographic map is shown in video form, so that the data analysis object of the preset EEG signal analysis model is more granular and more EEG signal data features are systematically and fully explored.

[0074] S330. The EEG signal topographic map sequence and the corresponding signal sampling object labels are used as model training samples and input into the initial EEG signal analysis model for model training to obtain the target EEG signal analysis model.

[0075] In this step, all collected sample data can be split into a training set, a validation set, and a test set, with a split ratio of 7:1:2.

[0076] During model training, topographic video sequences of multiple sub-topographic map sequences for each sample in the training dataset can be input into the initial EEG signal analysis model. The loss function value of the initial EEG signal output is used to iteratively train the model until the model converges, thus obtaining the target EEG signal analysis model.

[0077] For example, the structure of the target EEG signal analysis model algorithm can be referred to Figure 8The algorithm structure is shown below. The EEG signal topographic map sequence video is input to the first convolutional layer (conv1). The output of the first convolutional layer is then fed into a batch normalization (BN) and pooling layer (pool1). Further, the features processed by batch normalization (BN) and pooling layer (pool1) are fed into the second convolutional layer (conv2), followed by another batch normalization (BN) and pooling layer (pool2), then the third convolutional layer (conv3), and finally another batch normalization (BN) and pooling layer (pool3). Each convolutional layer uses the ReLU activation function to reduce gradient vanishing. After feature extraction by the convolutional neural network module, the features are fed into a Long Short-Term Memory (LSTM) network for further learning to extract more data features from the EEG signals, especially the temporal information in the EEG topographic map sequence. Finally, the output of the Long Short-Term Memory network passes through three fully connected layers (Dense1, 2, 3) to obtain the final signal analysis output.

[0078] S340. When the EEG signal to be analyzed is acquired, the EEG signal to be analyzed is processed according to steps S310 and S320 to obtain the target EEG signal topographic map sequence.

[0079] The target EEG signal topographic map sequence is presented in the form of a topographic map video.

[0080] S350. Input the target EEG signal topographic map sequence into the target EEG analysis model to obtain the classification result of the EEG signal to be analyzed.

[0081] The classification results of the EEG signals to be analyzed can indicate whether the EEG signals to be analyzed belong to the preset disease population.

[0082] The technical solution of this embodiment collects EEG signals from a preset population as training sample data for the model. The sample EEG signals are preprocessed to filter out noise. Then, a target EEG signal is extracted from the preprocessed sample EEG signals, and an EEG signal topography sequence is generated based on the target EEG signal. This involves using multiple EEG signal topography sequences arranged chronologically as the overall analysis object data, inputting them into an initial EEG signal analysis model for training to obtain the target EEG signal analysis model, which is used for screening EEG signals from the preset population. The target EEG signal analysis model includes a convolutional neural network module, a long short-term memory network module, and a fully connected network module. The convolutional neural network module extracts data features from the EEG signal topography map, the long short-term memory network module extracts and learns time-related feature information from the EEG signal topography sequence, and finally, the fully connected layer analyzes the EEG signal to obtain the analysis results. The network structure based on the preset EEG signal analysis model can more fully mine the data features in the EEG signals, thereby obtaining more accurate analysis results. In the application of the target EEG signal analysis model, the EEG signal to be analyzed is processed according to the processing procedure of the sample EEG signal to obtain the model input data, which is then input into the target EEG signal analysis model to obtain the analysis result of the EEG signal to be analyzed. The technical solution of this invention solves the problem that the data features are not fully explored in existing EEG signal analysis algorithms, resulting in low recognition accuracy of the target EEG signal. It realizes deep learning of EEG signal features and improves the accuracy of EEG signal analysis results.

[0083] Figure 9 This is a schematic diagram of an electroencephalogram (EEG) signal analysis device provided in an embodiment of the present invention. The EEG signal analysis device provided in this embodiment is suitable for processing and analyzing acquired EEG signals. This device can be implemented in software and / or hardware, and can be configured in an electronic device, such as a mobile terminal or server device.

[0084] like Figure 9 As shown, the EEG signal analysis device includes: a signal preprocessing module 410, a topographic map sequence generation module 420, and a signal analysis module 430.

[0085] The signal preprocessing module 410 is used to acquire the EEG signal to be analyzed and preprocess the EEG signal to be analyzed; the topographic map sequence generation module 420 is used to extract the target EEG signal from the preprocessed EEG signal to be analyzed and generate an EEG signal topographic map sequence based on the target EEG signal; the signal analysis module 440 is used to input the EEG signal topographic map sequence into a preset EEG signal analysis model to obtain the classification result of the EEG signal to be analyzed, wherein the preset EEG signal analysis model includes a convolutional neural network module, a long short-term memory network module, and a fully connected network module.

[0086] The technical solution of this invention involves preprocessing the acquired EEG signal to filter out noise data. Then, a target EEG signal is extracted from the preprocessed signal, and a sequence of EEG signal topographic maps is generated based on this target signal. This sequence of multiple EEG signal topographic maps arranged chronologically is input into a preset EEG signal analysis model to obtain the classification result of the EEG signal. The preset EEG signal analysis model includes a convolutional neural network (CNN) module, a long short-term memory (LSM) network module, and a fully connected network module. The CNN module extracts data features from the EEG signal topographic maps, the LSM module extracts and learns time-related features from the sequence, and finally, the fully connected layers analyze the data to obtain the analysis result. The network structure based on the preset EEG signal analysis model allows for a more thorough mining of data features in the EEG signal, resulting in more accurate analysis results. The technical solution of this invention solves the problem that the data features in existing EEG signal analysis algorithms are not fully mined, resulting in low recognition accuracy of target EEG signals. It realizes deep learning of EEG signal features and improves the accuracy of EEG signal analysis results.

[0087] In one optional implementation, the signal preprocessing module 410 is specifically used for:

[0088] The EEG signal to be analyzed is divided into multiple data segments of a preset data length;

[0089] Data filtering is performed based on the standard deviation of the EEG signal amplitude in each data segment and the standard deviation of all the EEG signal amplitudes to be analyzed.

[0090] The filtered data segments are then spliced ​​together in chronological order.

[0091] In an optional implementation, the signal preprocessing module 410 is further configured to:

[0092] Arrange and combine the data segments according to the order of signal acquisition time;

[0093] For the first data segment, if the standard deviation of the EEG signal amplitude of the first data segment is greater than a first preset multiple of the standard deviation of the amplitude of all the EEG signals to be analyzed, the first data segment is deleted.

[0094] For each data segment other than the first data segment, if the standard deviation of the EEG signal amplitude of the current data segment is greater than a second preset multiple of the standard deviation of the EEG signal amplitude of the previous data segment, the current data segment is deleted.

[0095] In one optional implementation, the topographic map sequence generation module 420 includes a target EEG signal extraction submodule, a global energy spectrum calculation submodule, and a topographic map sequence generation submodule;

[0096] Specifically, the target EEG signal extraction submodule is used to: extract the EEG signal of the target frequency band from the preprocessed EEG signal to be analyzed, and use it as the target EEG signal;

[0097] The global energy spectrum calculation submodule is specifically used for: standardizing the target EEG signal and calculating the global energy spectrum of the standardized target EEG signal;

[0098] The topographic map sequence generation submodule is specifically used to: extract target energy spectrum peak data from the global energy spectrum according to a preset sampling strategy, and generate an EEG signal topographic map sequence based on the amplitude of the EEG signal to be analyzed corresponding to the target energy spectrum peak data.

[0099] In one optional implementation, the topographic map sequence generation submodule can be used to:

[0100] Local energy peaks with a time interval greater than a preset time interval are selected from the global energy spectrum to obtain the target energy spectrum peak data.

[0101] In one optional implementation, the topographic map sequence generation submodule can be used to:

[0102] Generate an EEG topographic map for the amplitude data corresponding to each of the target energy spectrum peak data;

[0103] The EEG topographic map is divided into multiple sub-topographic map sequence groups according to time sequence, and each sub-topographic map sequence group generates a corresponding topographic map sequence video.

[0104] In one optional implementation, the signal analysis module 430 is specifically used for:

[0105] Each of the aforementioned topographic map video sequences is input into a preset EEG signal analysis model to obtain multiple preliminary classification results.

[0106] Statistical analysis was performed on the multiple preliminary classification results to obtain the final classification results.

[0107] In one alternative implementation, the convolutional neural network module includes three convolutional layers, and a batch normalization layer and a pooling layer are provided after each convolutional layer.

[0108] In one optional implementation, the training samples of the preset EEG signal analysis model during the training process are EEG signal topographic map sequences obtained by processing EEG signals of a preset patient population. Correspondingly, the classification result indicates whether the EEG signal to be analyzed is the EEG signal of the preset patient population.

[0109] The EEG signal analysis device provided in the embodiments of the present invention can execute the EEG signal analysis method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of executing the method.

[0110] Figure 10 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Figure 10 A block diagram is shown of an exemplary electronic device 12 suitable for implementing embodiments of the present invention. Figure 10 The electronic device 12 shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention. The electronic device 12 can be any terminal device with computing capabilities, such as personal computer systems, server computer systems, thin clients, thick clients, handheld or laptop devices, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputer systems, mainframe computer systems, and distributed cloud computing environments including any of the aforementioned systems, etc.

[0111] like Figure 10 As shown, the electronic device 12 is represented in the form of a general-purpose computing device. The components of the electronic device 12 may include, but are not limited to: one or more processors or processing units 16, system memory 28, and bus 18 connecting different system components (including system memory 28 and processing unit 16).

[0112] Bus 18 represents one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus architectures. For example, these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0113] Electronic device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by electronic device 12, including volatile and non-volatile media, removable and non-removable media.

[0114] System memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 40 and / or cache memory 42. Electronic device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 44 may be used to read and write non-removable, non-volatile magnetic media (… Figure 10 Not shown; usually referred to as a "hard drive"). Although Figure 10 Not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk") and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 18 via one or more data media interfaces. System memory 28 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of the present invention.

[0115] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in system memory 28. Such program modules 42 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. Program modules 42 typically perform the functions and / or methods described in the embodiments of the present invention.

[0116] Electronic device 12 can also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), and with one or more devices that enable a user to interact with electronic device 12, and / or with any device that enables electronic device 12 to communicate with one or more other computing devices (e.g., network card, modem, etc.). This communication can be performed via input / output (I / O) interface 22. Furthermore, electronic device 12 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 20. As shown, network adapter 20 communicates with other modules of electronic device 12 via bus 18. It should be understood that, although... Figure 10 As not shown, other hardware and / or software modules may be used in conjunction with electronic device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0117] Processing unit 16 executes various functional applications and data processing by running programs stored in system memory 28, such as implementing the electroencephalogram (EEG) signal analysis method provided in this embodiment, which includes:

[0118] Acquire the EEG signal to be analyzed and preprocess the EEG signal to be analyzed;

[0119] The target EEG signal is extracted from the preprocessed EEG signal to be analyzed, and an EEG signal topographic map sequence is generated based on the target EEG signal.

[0120] The EEG signal topographic map sequence is input into a preset EEG signal analysis model to obtain the classification result of the EEG signal to be analyzed. The preset EEG signal analysis model includes a convolutional neural network module, a long short-term memory network module, and a fully connected network module.

[0121] This invention also provides a computer storage medium storing a computer program that, when executed by a processor, implements the electroencephalogram (EEG) signal analysis method provided in the above embodiments.

[0122] It should be noted that the computer-readable medium described above in this invention can be a computer-readable signal medium, a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), or flash memory (FLASH), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0123] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0124] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0125] The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to:

[0126] Acquire the EEG signal to be analyzed and preprocess the EEG signal to be analyzed;

[0127] The target EEG signal is extracted from the preprocessed EEG signal to be analyzed, and an EEG signal topographic map sequence is generated based on the target EEG signal.

[0128] The EEG signal topographic map sequence is input into a preset EEG signal analysis model to obtain the classification result of the EEG signal to be analyzed. The preset EEG signal analysis model includes a convolutional neural network module, a long short-term memory network module, and a fully connected network module.

[0129] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof, including but not limited to object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0130] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0131] The units described in the embodiments of the present invention can be implemented in software or in hardware. The names of the units and modules do not necessarily limit the specific unit or module; for example, a data generation module can also be described as a "video data generation module".

[0132] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Array (FPGA), Application Specific Integrated Circuit (ASIC), Application Specific Standard Parts (ASSP), System on Chip (SOC), Complex Programmable Logic Device (CPLD), and so on.

[0133] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0134] It should be further noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to mutually. As for the apparatus embodiments, since they largely correspond to the method embodiments, the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0135] The methods and apparatus of the present invention may be implemented in many ways. For example, they may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above-described order of steps for the methods is for illustrative purposes only, and the steps of the methods of the present invention are not limited to the order specifically described above unless otherwise specifically stated. Furthermore, in some embodiments, the present invention may also be implemented as a program recorded on a recording medium, the program comprising machine-readable instructions for implementing the methods according to the present invention. Thus, the present invention also covers recording media storing programs for performing the methods according to the present invention.

[0136] The description of this invention is given for illustrative and descriptive purposes only and is not intended to be exhaustive or to limit the invention to the forms disclosed. Many modifications and variations will be apparent to those skilled in the art. The embodiments were chosen and described in order to better illustrate the principles and practical application of the invention and to enable those skilled in the art to understand the invention and to design various embodiments with various modifications suitable for a particular purpose.

Claims

1. A method for analyzing electroencephalogram (EEG) signals, characterized in that, include: Acquire the EEG signal to be analyzed, and divide the EEG signal to be analyzed into multiple data segments of a preset data length; arrange and combine the data segments according to the order of signal acquisition time; For the first data segment, if the standard deviation of the EEG signal amplitude of the first data segment is greater than a first preset multiple of the standard deviation of the amplitude of all the EEG signals to be analyzed, the first data segment is deleted. For each data segment other than the first data segment, if the standard deviation of the EEG signal amplitude of the current data segment is greater than a second preset multiple of the standard deviation of the EEG signal amplitude of the previous data segment, the current data segment is deleted, thus achieving data filtering. The filtered data segments are then spliced ​​together in chronological order to complete preprocessing. The target EEG signal is extracted from the preprocessed EEG signal to be analyzed, and an EEG signal topography map is generated based on the target EEG signal. All EEG topographic maps were divided into multiple sub-topographic map sequence groups in chronological order, and each sub-topographic map sequence group was used to generate a corresponding topographic map sequence video. The topographic map video sequence is input into a preset EEG signal analysis model to obtain the classification result of the EEG signal to be analyzed. The preset EEG signal analysis model includes a convolutional neural network module, a long short-term memory network module, and a fully connected network module.

2. The method according to claim 1, characterized in that, The step of extracting the target EEG signal from the preprocessed EEG signal to be analyzed, and generating an EEG signal topography sequence based on the target EEG signal, includes: The target frequency band of the EEG signal is extracted from the preprocessed EEG signal to be analyzed and is used as the target EEG signal. The target EEG signal is standardized, and the global energy spectrum of the standardized target EEG signal is calculated. According to the preset sampling strategy, target energy spectrum peak data is extracted from the global energy spectrum, and an EEG signal topographic map sequence is generated based on the amplitude of the EEG signal to be analyzed corresponding to the target energy spectrum peak data.

3. The method according to claim 2, characterized in that, The step of extracting target energy spectrum peak data from the global energy spectrum according to a preset sampling strategy includes: Local energy peaks with a time interval greater than a preset time interval are selected from the global energy spectrum to obtain the target energy spectrum peak data.

4. The method according to claim 2, characterized in that, The process of generating an EEG signal topography sequence based on the amplitude of the EEG signal to be analyzed corresponding to the target energy spectrum peak data includes: Generate an EEG topographic map for the amplitude data corresponding to each of the target energy spectrum peak data; The EEG topographic map is divided into multiple sub-topographic map sequence groups according to time sequence, and each sub-topographic map sequence group generates a corresponding topographic map sequence video.

5. The method according to claim 4, characterized in that, The step of inputting the EEG signal topographic map sequence into a preset EEG signal analysis model to obtain the classification result of the EEG signal to be analyzed includes: Each of the aforementioned topographic map video sequences is input into a preset EEG signal analysis model to obtain multiple preliminary classification results. Statistical analysis was performed on the multiple preliminary classification results to obtain the final classification results.

6. The method according to claim 1, characterized in that, The convolutional neural network module includes three convolutional layers, and each convolutional layer is followed by a batch normalization layer and a pooling layer.

7. The method according to any one of claims 1-6, characterized in that, The training samples of the preset EEG signal analysis model during the training process are EEG signal topographic map sequences obtained by processing EEG signals of a preset patient population. Correspondingly, the classification result indicates whether the EEG signal to be analyzed is the EEG signal of the preset patient population.

8. A brainwave signal analysis device, characterized in that, include: The signal preprocessing module is used to acquire the EEG signal to be analyzed, and to divide the EEG signal to be analyzed into multiple data segments of a preset data length; and to arrange and combine the data segments according to the order of signal acquisition time. For the first data segment, if the standard deviation of the EEG signal amplitude of the first data segment is greater than a first preset multiple of the standard deviation of the amplitude of all the EEG signals to be analyzed, the first data segment is deleted. For each data segment other than the first data segment, if the standard deviation of the EEG signal amplitude of the current data segment is greater than a second preset multiple of the standard deviation of the EEG signal amplitude of the previous data segment, the current data segment is deleted, thus achieving data filtering. The filtered data segments are then spliced ​​together in chronological order to complete preprocessing. The topographic map sequence generation module is used to extract target EEG signals from the preprocessed EEG signals to be analyzed, and to generate an EEG signal topographic map based on the target EEG signals. All EEG topographic maps were divided into multiple sub-topographic map sequence groups in chronological order, and each sub-topographic map sequence group was used to generate a corresponding topographic map sequence video. The signal analysis module is used to input the topographic map video sequence into a preset EEG signal analysis model to obtain the classification result of the EEG signal to be analyzed. The preset EEG signal analysis model includes a convolutional neural network module, a long short-term memory network module, and a fully connected network module.

9. An electronic device, characterized in that, The electronic device includes: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the electroencephalogram (EEG) signal analysis method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the electroencephalogram (EEG) signal analysis method as described in any one of claims 1-7.

Images