Electroencephalogram signal detection method, device and equipment and storage medium

Abstract

The application discloses a brain electrical signal detection method, device and equipment and a storage medium, relates to the technical field of brain electrical detection, and comprises the following steps: acquiring a plurality of brain electrical signals to be detected of a target user; wherein the brain electrical signals to be detected are collected by a plurality of electrode groups, and the plurality of electrode groups are arranged in zones; taking an electrode group as a node, taking a phase lag index value between brain electrical signals to be detected corresponding to any two electrode groups as a connection edge between the two nodes, and constructing a to-be-detected region connected graph of all brain electrical signals to be detected; and identifying a brain electrical signal detection result according to the to-be-detected region connected graph. The application can comprehensively reflect the overall brain activity state of the target user, solves the technical problem that the existing brain electrical signal detection method is based on single time-frequency domain features or nonlinear features for brain electrical signal detection, and has low accuracy, and the accuracy of brain electrical signal detection is improved.

Description

Technical Field

[0001] This application relates to the field of electroencephalography (EEG) detection technology, and in particular to a method, apparatus, device, and storage medium for detecting EEG signals. Background Technology

[0002] With the development of society, industry, and economy, the incidence of tinnitus is increasing. Related technologies involve extracting time-frequency domain features or nonlinear features from the electroencephalogram (EEG) signals of tinnitus patients, and then constructing an EEG signal detection model based on machine learning to achieve automatic EEG signal detection and assist in tinnitus diagnosis. However, existing EEG signal detection methods, based on single time-frequency domain features or nonlinear features, have relatively low accuracy. Summary of the Invention

[0003] The main objective of this application is to provide a method, apparatus, device, and storage medium for detecting electroencephalogram (EEG) signals, aiming to solve the technical problem that existing EEG signal detection methods, which rely on single time-frequency domain features or nonlinear features for EEG signal detection, have low accuracy.

[0004] To achieve the above objectives, this application adopts the following technical solution:

[0005] Firstly, this application provides a method for detecting electroencephalogram (EEG) signals, the method comprising:

[0006] Multiple EEG signals to be detected from the target user are acquired; the EEG signals to be detected are acquired by multiple electrode groups, which are arranged in zones.

[0007] Using electrode groups as nodes, and the phase lag index value between any two EEG signals to be detected as the connection edge between the two nodes, a connected graph of the regions to be detected for all EEG signals to be detected is constructed.

[0008] Based on the connectivity graph of the region to be detected, the EEG signal detection results are identified.

[0009] Optionally, an electrode set acquires at least two EEG signals to be detected;

[0010] The steps of constructing a connected graph of all EEG signals to be detected, using electrode sets as nodes and the phase lag exponent between any two electrode sets as the connection edge between the two nodes, include:

[0011] Use the electrode assembly as a node;

[0012] For each electrode group, the weighted average of at least two EEG signals to be detected corresponding to the electrode group is used to obtain the target EEG signal for each electrode group.

[0013] Determine the target phase hysteresis index value between any two target EEG signals;

[0014] By using the target phase lag index value as the edge between the two corresponding nodes, a connected graph of the region to be detected is obtained.

[0015] Optionally, the step of identifying the EEG signal detection result based on the connectivity graph of the region to be detected includes:

[0016] The connected graph of the region to be detected is input into the EEG signal detection model to identify the EEG signal detection results. The EEG signal detection results include the target EEG abnormality degree and the target EEG abnormality location of the target user. The EEG signal detection model is trained using a first sample set, which includes connected graphs of abnormal regions of multiple users with abnormal EEG, and labels of the EEG abnormality degree and EEG abnormality location corresponding to each connected graph of abnormal regions.

[0017] Optionally, before the step of inputting the connectivity graph of the region to be detected into the EEG signal detection model and identifying the EEG signal detection result, the method further includes:

[0018] Multiple abnormal EEG signals were acquired from each user with abnormal EEG; the abnormal EEG signals were collected from multiple electrode groups.

[0019] For each user with abnormal EEG, each abnormal EEG signal is divided into multiple EEG segment signals.

[0020] Based on the segmented phase lag index between the segmented EEG signals corresponding to any two electrode groups, construct a connectivity graph of multiple abnormal regions for each user with abnormal EEG.

[0021] Based on the degree and location of EEG abnormalities of each user with abnormal EEG, the degree and location of abnormalities are labeled in the connectivity graphs of multiple abnormal regions to obtain the first sample set.

[0022] A multi-task detection model was trained using the first sample set to obtain an EEG signal detection model.

[0023] Optionally, the EEG signal detection model includes graph neural networks and long short-term memory networks;

[0024] Before the step of inputting the connectivity graph of the region to be detected into the EEG signal detection model and identifying the EEG signal detection results, the method also includes:

[0025] Extract the brainwave rhythm signals corresponding to multiple brainwave signals to be detected; the brainwave rhythm signals to be detected include brainwave signals of multiple rhythms;

[0026] The steps for inputting the connectivity graph of the region to be detected into the EEG signal detection model and identifying the EEG signal detection results include:

[0027] Input the connectivity graph of the region to be detected and the EEG rhythm signal to be detected into the EEG signal detection model;

[0028] The graph neural network identifies the degree and location of the first EEG abnormality in the target patient based on the connectivity graph of the region to be detected.

[0029] The second degree of EEG abnormality and the second location of EEG abnormality in the target patient are obtained by using a long short-term memory network based on the target EEG rhythm signal. The long short-term memory network is trained using a second sample set, which includes abnormal EEG rhythm signals from multiple users with abnormal EEG, EEG abnormality degree labels and EEG abnormality location labels corresponding to each abnormal EEG rhythm signal, and the abnormal EEG rhythm signals include multiple rhythm EEG signals extracted from each abnormal EEG signal.

[0030] The target EEG abnormality level is obtained based on the first and second EEG abnormality levels, and the target EEG abnormality location is obtained based on the first and second EEG abnormality locations.

[0031] Optionally, multiple electrode groups are arranged in the left frontal region, right frontal region, left ear region, right ear region, central region, left parietal lobe region, and right parietal lobe region of the target user.

[0032] Optionally, the step of acquiring multiple EEG signals to be detected from the target patient includes:

[0033] Acquire multiple initial EEG signals from the target user;

[0034] Multiple initial EEG signals were processed by abnormal fluctuation removal, electrode localization, filtering, baseline drift processing, and independent component analysis to obtain multiple EEG signals to be detected.

[0035] Secondly, this application also provides an electroencephalogram (EEG) signal detection device, the device comprising:

[0036] The acquisition module is used to acquire multiple EEG signals to be detected from the target user; wherein, the EEG signals to be detected are acquired by multiple electrode groups, and the multiple electrode groups are arranged in zones;

[0037] The module is used to construct a connected graph of the regions to be detected for all EEG signals, with electrode groups as nodes and the phase lag index value between the EEG signals to be detected corresponding to any two electrode groups as the connection edge between the two nodes.

[0038] The detection module is used to identify the EEG signal detection results based on the connectivity graph of the region to be detected.

[0039] Thirdly, this application also provides an electroencephalogram (EEG) signal detection device, which includes: a memory, a processor, and an EEG signal detection program stored in the memory and executable on the processor, wherein the EEG signal detection program is configured to implement the steps of any of the above-described EEG signal detection methods.

[0040] Fourthly, this application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the electroencephalogram (EEG) signal detection method as described above.

[0041] The above-mentioned one or more technical solutions provided in this application may have the following advantages or at least achieve the following technical effects:

[0042] This application provides a method, apparatus, device, and storage medium for detecting electroencephalogram (EEG) signals, which acquires multiple EEG signals to be detected from a target user. The EEG signals are acquired by multiple electrode groups arranged in zones. Each electrode group is used as a node, and the phase lag index between any two corresponding EEG signals from different electrode groups is used as a connection edge between the two nodes to construct a connected graph of all EEG signals to be detected. The EEG signal detection result is identified based on this connected graph.

[0043] Therefore, this application constructs a connectivity graph of the detection regions for all EEG signals based on the partitioned arrangement of multiple electrode groups and the phase lag index value between the EEG signals to be detected. The connectivity graph of the detection regions reflects the connectivity status between the partitions of multiple electrode groups. Thus, the EEG signal detection results identified based on the connectivity graph of the detection regions can more comprehensively reflect the overall brain activity state of the target user and have higher accuracy compared to EEG signal detection based on a single time-frequency domain feature or nonlinear feature. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0045] Figure 1 This is a schematic diagram of the structure of the electroencephalogram (EEG) signal detection device in the hardware operating environment involved in the embodiments of this application;

[0046] Figure 2 This is a flowchart illustrating the first embodiment of the electroencephalogram (EEG) signal detection method of this application;

[0047] Figure 3 This is an exemplary partitioned arrangement diagram of multiple electrode groups;

[0048] Figure 4 This is a schematic diagram of the modules of the first embodiment of the EEG signal detection device of this application.

[0049] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0051] It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0052] In this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that an apparatus or system comprising a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such an apparatus or system. Without further limitation, an element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the apparatus or system that includes that element.

[0053] If the embodiments of the present invention involve descriptions such as "first" and "second," such descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" and "second" may explicitly or implicitly include at least one of those features.

[0054] Tinnitus is an auditory hallucination that occurs in the absence of acoustic stimulation. Its potential threats include hearing loss, head injury, and depression. Persistent tinnitus can affect a patient's social and communication abilities, leading to psychological problems such as insomnia, anxiety, and depression, thus reducing their quality of life. With the development of society, industry, and the economy, the incidence of tinnitus is increasing. Studies have shown that 10%–15% of the population experiences tinnitus symptoms, and 1%–7% of the population suffers severe quality of life due to tinnitus. Electroencephalogram (EEG), as a common physiological signal, has high temporal resolution and a certain degree of spatial resolution, enabling it to reflect the activity state of neurons with high precision. Furthermore, compared to other neuroscience techniques (such as magnetoencephalography (MEG) and resting-state functional magnetic resonance imaging (rs-fMRI)), EEG-based analysis techniques are less expensive and more portable, thus they are widely used in the study of brain activity in tinnitus patients.

[0055] In related technologies, common EEG research methods involve extracting time-frequency domain features, nonlinear features, and connectivity features from EEG signals. Based on machine learning models, EEG signal detection models are constructed (such as Random Forest (RF), Support Vector Machine (SVM), or common deep learning network models, such as Convolutional Neural Networks (CNN) and Recurrent Recurrent Neural Networks (RNN), to construct tinnitus grading models), achieving automatic EEG signal detection and assisting in tinnitus diagnosis. Among these, time-frequency domain features reflect the changes in EEG signals over time and at different frequency bands (alpha, beta, etc.), while nonlinear features characterize the disorder of EEG signals but cannot represent the correlation of activity in different brain regions. Although connectivity features alone can reflect the connection characteristics between different brain regions, they cannot represent the relationships and states between overall EEG activity regions.

[0056] In summary, existing EEG signal detection methods cannot reflect the overall connectivity of the brain and have low accuracy.

[0057] In view of the low accuracy of existing EEG signal detection methods that rely on single time-frequency domain features or nonlinear features, this application provides an EEG signal detection method, the overall idea of ​​which is as follows:

[0058] The method includes: acquiring multiple EEG signals to be detected from a target user; wherein the EEG signals to be detected are acquired by multiple electrode groups, and the multiple electrode groups are arranged in zones; using the electrode groups as nodes, and using the phase lag index value between the EEG signals to be detected corresponding to any two electrode groups as the connection edge between the two corresponding nodes, constructing a connected graph of the regions to be detected for all EEG signals to be detected; and identifying the EEG signal detection results based on the connected graph of the regions to be detected.

[0059] This application provides a method for detecting electroencephalogram (EEG) signals. Based on the zonal arrangement of multiple electrode groups and the phase lag index value between the EEG signals to be detected, a connectivity graph of the regions to be detected for all EEG signals to be detected is constructed. The connectivity graph of the regions to be detected reflects the connectivity status between the zonal areas of multiple electrode groups. Therefore, the EEG signal detection results identified based on the connectivity graph of the regions to be detected can more comprehensively reflect the overall brain activity state of the target user and have higher accuracy compared to EEG signal detection based on a single time-frequency domain feature or nonlinear feature.

[0060] The following provides a detailed description of the electroencephalogram (EEG) signal detection method, apparatus, equipment, and storage medium used in the technical implementation of this application:

[0061] Reference Figure 1 , Figure 1 This is a schematic diagram of the structure of the electroencephalogram (EEG) signal detection device in the hardware operating environment of the embodiment of this application.

[0062] like Figure 1 As shown, the device may include: a processor 1001, such as a CPU; a user interface 1003; a memory 1005; and a communication bus 1002. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include an EEG acquisition device; optionally, the user interface 1003 may also be a display screen, an input unit such as a keyboard, etc. The memory 1005 may be high-speed RAM or stable non-volatile memory, such as a disk drive. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0063] It is understood that the device may also include a network interface 1004, which may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). Optionally, the device may also include RF (Radio Frequency) circuitry, sensors, audio circuitry, a Wi-Fi module, etc.

[0064] Those skilled in the art will understand that Figure 1 The device structure shown does not constitute a limitation on the device and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0065] The following describes in detail the EEG signal detection method, apparatus, device, and storage medium of this application with reference to the accompanying drawings and specific embodiments.

[0066] Based on, but not limited to, the above hardware structure, refer to Figure 2 and Figure 3 , Figure 2 This is a flowchart illustrating the first embodiment of the EEG signal detection method of this application. Figure 3 This is an example of a partitioned arrangement of multiple electrode groups.

[0067] This embodiment provides a method for detecting electroencephalogram (EEG) signals, such as... Figure 2 As shown, the method may include:

[0068] Step S100: Acquire multiple EEG signals to be detected from the target user.

[0069] The EEG signals to be detected are acquired by multiple electrode groups, which are arranged in zones.

[0070] In this embodiment, the execution entity is the aforementioned EEG signal detection device, which can be a physical server including an independent host, or a virtual server hosted by a host cluster.

[0071] Taking the scenario of using EEG signal detection results to assist in tinnitus diagnosis as an example, the target user can be anyone who needs tinnitus diagnosis. The EEG signal detection device can acquire the target user's EEG signal to be tested, perform detection on the EEG signal, and output the EEG signal detection result to assist in tinnitus diagnosis. Multiple electrode groups can be the electrode groups of the EEG acquisition device, and the EEG signal detection device can acquire multiple EEG signals to be tested from the target user. When the EEG acquisition device acquires the EEG signals to be tested, the multiple electrode groups are respectively set in different EEG acquisition areas of the target user's brain.

[0072] It is understandable that EEG acquisition devices can collect EEG signals from different regions of a target user using different electrode sets. Each electrode set may include at least one electrode, corresponding to the acquisition of at least one EEG signal from a specific EEG acquisition region.

[0073] As one specific implementation, multiple electrode groups are arranged in sections in the target user's left frontal area, right frontal area, left ear area, right ear area, central area, left parietal lobe area, and right parietal lobe area.

[0074] In this embodiment, the EEG acquisition area may include the left prefrontal cortex, right prefrontal cortex, left ear cortex, right ear cortex, central cortex, left parietal lobe cortex, and right parietal lobe cortex. The number of EEG signals to be detected varies depending on the electrode group and the configuration of the EEG acquisition device. Therefore, the partitions corresponding to each electrode in different EEG acquisition devices can be determined according to actual usage requirements.

[0075] like Figure 3As shown, taking a 64-channel EEG acquisition device as an example, the 64-channel EEG acquisition device can include 64 electrodes. The EEG acquisition area corresponding to electrode groups FC1, FC3 and FC5 can be the left frontal region; the EEG acquisition area corresponding to electrode groups FC2, FC4 and FC6 can be the right frontal region; the EEG acquisition area corresponding to electrode groups FT7, T7, C5, C3, TP7, CP5 and CP3 can be the left ear region; the EEG acquisition area corresponding to electrode groups FT8, T8, C6, C4, TP8, CP6 and CP4 can be the right ear region; the EEG acquisition area corresponding to electrode groups P7, P5, P3 and P1 can be the left parietal lobe region; and the EEG acquisition area corresponding to electrode groups P8, P6, P4 and P2 can be the right parietal lobe region.

[0076] Further, step S100 may include: acquiring multiple initial EEG signals of the target user; performing abnormal fluctuation removal processing, electrode localization, filtering processing, baseline drift processing, and independent component analysis on the multiple initial EEG signals to obtain multiple EEG signals to be detected.

[0077] It is understandable that the multiple initial EEG signals acquired by the EEG acquisition device are subject to interference from various factors, which may result in abnormal bands, including electrooculogram (EOG) signals, electrocardiogram (ECG) signals, and other artifact signals. Therefore, after acquiring multiple initial EEG signals, the EEG signal detection device can preprocess these signals to obtain multiple clean EEG signals to be detected.

[0078] In practice, EEG signal detection equipment can use the eeglab toolkit included in Matlab to perform abnormal fluctuation removal, electrode localization, filtering, baseline drift processing, and independent component analysis on multiple initial EEG signals in batches.

[0079] Among them, the abnormal fluctuation removal process can remove the time periodic signals of the current abnormal fluctuations, and at the same time perform interpolation processing on the initial EEG signals of insufficient leads.

[0080] Electrode positioning can be achieved based on the EEG signal acquisition template corresponding to the EEG acquisition device.

[0081] The filtering process may include filtering the power supply frequency at a preset frequency and performing bandpass filtering within a preset frequency range. For example, for the aforementioned 64-channel EEG acquisition device, the preset frequency could be 50Hz, and the preset frequency range could be 0.5 to 90Hz.

[0082] Step S200: Using the electrode group as nodes, and the phase lag index value between any two electrode groups corresponding to the EEG signals to be detected as the connection edge between the two corresponding nodes, construct a connected graph of the regions to be detected for all EEG signals to be detected.

[0083] In this embodiment, the connected graph of the region to be detected may include nodes and edge values. Nodes may be different electrode groups corresponding to different EEG acquisition regions, and edge values ​​may be the phase-lag index (PLI) values ​​between the EEG signals to be detected corresponding to different electrode groups.

[0084] It can be understood that when one electrode group corresponds to one EEG signal to be detected, the edge values ​​of the connected graph of the region to be detected are the PLI values ​​between the EEG signals to be detected corresponding to each electrode group. When one electrode group corresponds to multiple EEG signals to be detected, the multiple EEG signals to be detected corresponding to each electrode group can be integrated to obtain the target EEG signal corresponding to each electrode group. The edge values ​​of the connected graph of the region to be detected are the PLI values ​​between the target EEG signals to be detected corresponding to each electrode group.

[0085] As one specific implementation, an electrode array acquires at least two EEG signals to be detected.

[0086] Step S200 may include: using electrode groups as nodes; for each electrode group, performing a weighted average of at least two EEG signals to be detected corresponding to the electrode group to obtain the target EEG signal corresponding to each electrode group; determining the target phase lag index value between any two target EEG signals; and using the target phase lag index value as the edge between the corresponding two nodes to obtain a connected graph of the region to be detected.

[0087] In this embodiment, when one electrode group corresponds to multiple EEG signals to be detected, the multiple EEG signals corresponding to each electrode group can be weighted and averaged to obtain the target EEG signal corresponding to each electrode group. The specific weighted averaging algorithm is set according to the actual usage.

[0088] Specifically, the target phase lag index between target EEG signals can be determined using Formula 1. Formula 1 is:

[0089]

[0090] Where i represents the i-th channel, k represents the k-th channel, t represents the time point, T represents the total time, sgn represents the sign (-1 represents a negative value, +1 represents a positive value, and 0 represents a zero value), and Im(Z i (t)Z k (t)*) represents the phase difference between the i-th channel and the k-th channel.

[0091] Step S300: Identify the EEG signal detection results based on the connectivity graph of the region to be detected.

[0092] In this embodiment, the connected graph of the region to be detected can be input into a machine learning model, and the machine learning model can perform EEG signal detection based on the connected graph of the region to be detected to obtain the EEG signal detection result.

[0093] As one specific implementation, step S300 may include:

[0094] Step S310: Input the connectivity graph of the region to be detected into the EEG signal detection model and identify the EEG signal detection results.

[0095] The EEG signal detection results include the target EEG abnormality level and the target EEG abnormality location of the target user. The EEG signal detection model is trained using a first sample set, which includes abnormal region connectivity graphs of multiple EEG abnormal users, EEG abnormality level labels and EEG abnormality location labels corresponding to each abnormal region connectivity graph.

[0096] In this embodiment, the EEG signal detection model may include a multi-task learning model y = f(x, w, b), where f is a nonlinear model implemented using deep learning methods, w and b are the model's weight and bias parameters, obtained through a loss function and gradient descent algorithm, and y is the annotation information of the connected region graph. Preferably, the multi-task learning model is a graph convolutional network (GCN). The target EEG abnormality level can be any one of no abnormality, mild abnormality, moderate abnormality, and severe abnormality, and the target EEG abnormality location can be any one of no abnormality, left-sided abnormality, right-sided abnormality, and bilateral abnormality.

[0097] For scenarios where EEG signal detection results are used to assist in tinnitus diagnosis, users with abnormal EEG signals can be identified as tinnitus patients. An abnormal region connectivity map is constructed based on the abnormal EEG signals of each tinnitus patient using the aforementioned method for constructing a connectivity map of the region to be detected. After determining the abnormal region connectivity map corresponding to each tinnitus patient, the degree and location of the EEG abnormality can be labeled according to the severity and location of tinnitus. Tinnitus severity can include no tinnitus, mild abnormality, moderate abnormality, and severe abnormality; tinnitus location can include no tinnitus, left-sided tinnitus, right-sided tinnitus, and bilateral tinnitus.

[0098] In practical use, the degree of the target EEG abnormality can correspond to the severity of tinnitus, and the location of the target EEG abnormality can correspond to the location where tinnitus occurs. After obtaining the EEG signal detection results, the severity of the target user's tinnitus and the location of tinnitus can be determined based on the degree and location of the target EEG abnormality, respectively.

[0099] Further, before step S310, the method may also include: acquiring multiple abnormal EEG signals from each user with abnormal EEG; the abnormal EEG signals are acquired by multiple electrode groups; for each user with abnormal EEG, each abnormal EEG signal is divided into multiple EEG segment signals; based on the segmented phase lag index value between the EEG segment signals corresponding to any two electrode groups, a connected graph of multiple abnormal regions for each user with abnormal EEG is constructed; based on the degree and location of the EEG abnormality for each user with abnormal EEG, the connected graph of multiple abnormal regions is labeled with the degree of abnormality and the location of the abnormality, respectively, to obtain a first sample set; and a multi-task detection model is trained using the first sample set to obtain an EEG signal detection model.

[0100] In this embodiment, multiple abnormal EEG signals of each user with abnormal EEG can also be acquired by an EEG acquisition device and obtained after preprocessing using the above-mentioned initial EEG signal preprocessing method.

[0101] When constructing the first sample set, abnormal EEG signals can be divided into multiple EEG segment signals according to a preset time interval, thereby increasing the sample size in the first sample set. Furthermore, a connectivity map of abnormal regions can be constructed using the EEG segment signals from each electrode group corresponding to the same preset time interval. The preset time interval can be set according to time-related needs; preferably, in scenarios where the EEG signal detection results are used to assist in tinnitus diagnosis, the preset time interval can be 30 seconds.

[0102] Furthermore, the EEG signal detection model includes graph neural networks and long short-term memory networks.

[0103] Before step S310, the method may further include: extracting the brain rhythm signals to be detected corresponding to multiple brain signals to be detected; the brain rhythm signals to be detected include multiple brain rhythm signals.

[0104] Step S310 may include: inputting the connected graph of the region to be detected and the EEG rhythm signal to be detected into the EEG signal detection model; identifying the first degree of EEG abnormality and the first location of EEG abnormality of the target patient based on the connected graph of the region to be detected using a graph neural network; obtaining the second degree of EEG abnormality and the second location of EEG abnormality of the target patient based on the target EEG rhythm signal using a long short-term memory network; the long short-term memory network is trained using a second sample set, which includes abnormal EEG rhythm signals of multiple EEG abnormal users, EEG abnormality degree labels and EEG abnormality location labels corresponding to each abnormal EEG rhythm signal, and the abnormal EEG rhythm signals include multiple rhythm EEG signals extracted from each abnormal EEG signal; obtaining the target EEG abnormality degree based on the first degree of EEG abnormality and the second degree of EEG abnormality, and obtaining the target EEG abnormality location based on the first location of EEG abnormality and the second location of EEG abnormality.

[0105] In this embodiment, multiple EEG signals of the target user can also be detected using a Long Short-Term Memory (LSTM) network. The detection results from the graph neural network and the LTM network are then combined to obtain the EEG signal detection result. Specifically, the EEG signal detection result y is:

[0106] y = αy a +βy b ;

[0107] y a For the detection results of the graph neural network, y b The results are from the Long Short-Term Memory network. α and β represent the weights of the two results, respectively. The values ​​of α and β can be determined based on the test results during model training.

[0108] It is understandable that the EEG signal detection result y can include the degree of EEG abnormality y1 and the location of EEG abnormality y2, where,

[0109] y1=α1y a1 +β1y b1 ;

[0110] y2=α2y a2 +β2y b2 .

[0111] It should be noted that the Long Short-Term Memory (LSTM) network is trained using a second sample set constructed from multiple abnormal EEG rhythm signals from various users with abnormal EEG patterns. Specifically, an abnormal EEG rhythm signal is extracted from the abnormal EEG signals corresponding to a single electrode group. When an electrode group corresponds to multiple abnormal EEG signals, the initial EEG rhythm signal corresponding to each abnormal EEG signal is extracted separately. Then, the EEG signals of the same rhythm among the multiple initial EEG rhythm signals are weighted and averaged to obtain the abnormal EEG rhythm signal corresponding to that electrode group.

[0112] In practical application, the rhythm can include delta waves (0.5-3.5Hz), theta waves (4-7.5Hz), alpha waves (8-12Hz), beta waves (13-30Hz), and gamma waves (30.5-44Hz). This embodiment uses wavelet packet functions to perform wavelet transform on the preprocessed abnormal EEG signals, achieving equal-length frequency band segmentation of the abnormal EEG signals. Then, wavelet reconstruction is performed on the EEG signals of each rhythm, thus enabling the extraction of different rhythmic waves from the abnormal EEG signals. Correspondingly, the acquisition method for the EEG rhythm signals to be detected is the same as the acquisition method for abnormal EEG rhythm signals.

[0113] Firstly, this embodiment provides a method for detecting electroencephalogram (EEG) signals. Based on the partitioned arrangement of multiple electrode groups and combined with the phase lag index value between the EEG signals to be detected, a connectivity graph of the regions to be detected for all EEG signals to be detected is constructed. The connectivity graph of the regions to be detected reflects the connectivity status between the partitions of multiple electrode groups. Therefore, the EEG signal detection results identified based on the connectivity graph of the regions to be detected can more comprehensively reflect the overall brain activity state of the target user and have higher accuracy compared to EEG signal detection based on a single time-frequency domain feature or nonlinear feature.

[0114] Secondly, this embodiment also employs a multi-task learning model to analyze the connected graph of the region to be detected. This model can simultaneously output the degree and location of EEG abnormalities, allowing users to quickly determine the corresponding detection results and improving the user experience. Furthermore, it combines the detection results from graph neural networks and long short-term memory networks to obtain the final EEG signal detection result, taking into account the influence of various factors on the EEG signal and improving the accuracy of the EEG signal detection results.

[0115] Based on the same inventive concept, referring to Figure 4 , Figure 4 This is a schematic diagram of the modules of the first embodiment of the electroencephalogram (EEG) signal detection device of this application. This embodiment provides an EEG signal detection device, which may include:

[0116] The acquisition module 10 is used to acquire multiple EEG signals to be detected from the target user; wherein the EEG signals to be detected are acquired by multiple electrode groups, and the multiple electrode groups are arranged in zones.

[0117] Module 20 is used to construct a connected graph of the regions to be detected for all EEG signals, with electrode groups as nodes and the phase lag index value between the EEG signals to be detected corresponding to any two electrode groups as the connection edge between the two nodes.

[0118] The detection module 30 is used to identify the EEG signal detection results based on the connectivity graph of the region to be detected.

[0119] For more details on the specific implementation of the above-mentioned EEG signal detection device, please refer to the description of the specific implementation of the EEG signal detection method in Embodiment 1 above. For the sake of brevity, these details will not be repeated here.

[0120] Furthermore, embodiments of this application also propose a computer storage medium storing a computer program, which, when executed by a processor, implements the steps of the EEG signal detection method described above. Therefore, further details will not be repeated here. Additionally, the beneficial effects of using the same method will not be repeated. For technical details not disclosed in the computer-readable storage medium embodiments of this application, please refer to the description of the method embodiments of this application. As an example, program instructions can be deployed to execute on a single computing device, or on multiple computing devices located at one location, or on multiple computing devices distributed across multiple locations and interconnected via a communication network.

[0121] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for detecting electroencephalogram (EEG) signals, characterized in that, An electrode set acquires at least two EEG signals to be detected, the method comprising: Multiple EEG signals to be detected from a target user are acquired; wherein the EEG signals to be detected are acquired by multiple electrode groups, and the multiple electrode groups are arranged in zones; Using the electrode group as nodes, and the phase lag index value between any two EEG signals to be detected corresponding to the electrode group as the connection edge between the two nodes, a connected graph of the regions to be detected for all the EEG signals to be detected is constructed. Based on the connectivity graph of the region to be detected, the EEG signal detection results are identified; For each of the electrode groups, the weighted average of at least two EEG signals to be detected corresponding to the electrode group is performed to obtain the target EEG signal corresponding to each electrode group. Determine the target phase hysteresis index value between any two target EEG signals; Using the target phase lag index value as the edge between the two corresponding nodes, the connected graph of the region to be detected is obtained; The EEG signal detection model includes a graph neural network and a long short-term memory network, and extracts multiple EEG rhythm signals corresponding to the EEG signals to be detected; the EEG rhythm signals to be detected include multiple EEG rhythms. The connectivity graph of the region to be detected and the EEG rhythm signal to be detected are input into the EEG signal detection model; The graph neural network identifies the first degree of EEG abnormality and the first location of EEG abnormality of the target user based on the connectivity graph of the region to be detected. The Long Short-Term Memory (LSTM) network obtains the second degree of EEG abnormality and the second location of EEG abnormality for the target user based on the target EEG rhythm signal. The LTM network is trained using a second sample set, which includes abnormal EEG rhythm signals from multiple users with abnormal EEG signals, EEG abnormality degree labels and EEG abnormality location labels corresponding to each abnormal EEG rhythm signal, and the abnormal EEG rhythm signals include multiple EEG signals of the rhythms extracted from each abnormal EEG signal. The target EEG abnormality level is obtained based on the first EEG abnormality level and the second EEG abnormality level, and the target EEG abnormality location is obtained based on the first EEG abnormality location and the second EEG abnormality location.

2. The method as described in claim 1, characterized in that, The step of identifying the EEG signal detection result based on the connectivity graph of the region to be detected includes: The connected graph of the region to be detected is input into the EEG signal detection model to identify the EEG signal detection result; wherein, the EEG signal detection result includes the target EEG abnormality degree and the target EEG abnormality location of the target user, and the EEG signal detection model is trained using a first sample set, the first sample set including the abnormal region connected graphs of multiple EEG abnormal users, the EEG abnormality degree label and the EEG abnormality location label corresponding to each abnormal region connected graph.

3. The method as described in claim 2, characterized in that, Before the step of inputting the connectivity graph of the region to be detected into the EEG signal detection model and identifying the EEG signal detection result, the method further includes: Multiple abnormal EEG signals are acquired from each of the aforementioned users with abnormal EEG signals; the abnormal EEG signals are collected by multiple electrode groups. For each of the aforementioned users with abnormal EEG signals, the abnormal EEG signals are divided into multiple EEG segment signals. Based on the segmented phase lag index value between the segmented EEG signals corresponding to any two of the electrode groups, construct a connectivity graph of multiple abnormal regions for each of the users with abnormal EEG. Based on the degree and location of the EEG abnormality of each of the users with abnormal EEG, the abnormality degree and location of the multiple abnormal regions are labeled in the connectivity graphs of the abnormal regions to obtain the first sample set. The EEG signal detection model is obtained by training a multi-task detection model using the first sample set.

4. The method as described in claim 1, characterized in that, Multiple electrode groups are arranged in the left frontal region, right frontal region, left ear region, right ear region, central region, left parietal lobe region, and right parietal lobe region of the target user.

5. The method according to any one of claims 1 to 4, characterized in that, The step of acquiring multiple EEG signals to be detected from the target user includes: Acquire multiple initial EEG signals from the target user; Multiple initial EEG signals are subjected to abnormal fluctuation removal processing, electrode localization, filtering processing, baseline drift processing, and independent component analysis to obtain multiple EEG signals to be detected.

6. A brainwave signal detection device, characterized in that, The device includes: The acquisition module is used to acquire multiple EEG signals to be detected from a target user; wherein the EEG signals to be detected are acquired by multiple electrode groups, and the multiple electrode groups are arranged in zones. A construction module is used to construct a connected graph of the detection region for all the EEG signals, using the electrode groups as nodes and the phase lag index value between any two EEG signals corresponding to the electrode groups as the connection edge between the two nodes. For each electrode group, a weighted average is performed on at least two EEG signals corresponding to the electrode group to obtain the target EEG signal for each electrode group. A target phase lag index value is determined between any two target EEG signals. The target phase lag index value is used as the edge between the two nodes to obtain the connected graph of the detection region. The detection module is used to identify the EEG signal detection results based on the connectivity graph of the region to be detected. The EEG signal detection model includes a graph neural network and a long short-term memory network to extract multiple EEG rhythm signals corresponding to the EEG signals to be detected. The EEG rhythm signals to be detected include multiple EEG rhythms. The connectivity graph of the region to be detected and the EEG rhythm signals to be detected are input into the EEG signal detection model. The graph neural network identifies the first degree of EEG abnormality and the first location of EEG abnormality of the target user based on the connectivity graph of the region to be detected. The long short-term memory network obtains the target EEG rhythm signal based on the target user's target EEG rhythm signal. The user's second EEG abnormality level and second EEG abnormality location; wherein, the long short-term memory network is trained using a second sample set, the second sample set including abnormal EEG rhythm signals of multiple users with EEG abnormalities, EEG abnormality level labels and EEG abnormality location labels corresponding to each abnormal EEG rhythm signal, the abnormal EEG rhythm signals including multiple rhythm EEG signals extracted from each abnormal EEG signal; the target EEG abnormality level is obtained based on the first EEG abnormality level and the second EEG abnormality level, and the target EEG abnormality location is obtained based on the first EEG abnormality location and the second EEG abnormality location.

7. A brainwave signal detection device, characterized in that, The device includes: a memory, a processor, and an EEG signal detection program stored in the memory and executable on the processor, the EEG signal detection program being configured to implement the steps of the EEG signal detection method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps of the electroencephalogram (EEG) signal detection method as described in any one of claims 1 to 5.

Images