Electroencephalogram signal processing system and method and computer equipment

Abstract

The invention discloses an electroencephalogram signal processing system and method and computer equipment, and relates to the field of medical instruments and signal processing.The system comprises a signal acquisition module, a signal processing module and a signal processing module, the feature extraction module is connected with the signal acquisition module and is used for performing feature extraction on the electroencephalogram signals to obtain electroencephalogram features; the electroencephalogram features comprise time domain features, frequency domain features and nonlinear dynamic features; the classification module is connected with the feature extraction module and used for determining the category of the electroencephalogram signals based on a pre-trained electroencephalogram classification model according to the electroencephalogram features; the electroencephalogram classification model comprises a convolutional neural network, a long short-term memory network and a classifier which are connected in sequence; the category comprises a normal state and different disease states; and the display module is connected with the classification module and is used for visualizing the electroencephalogram signals and the categories of the electroencephalogram signals. According to the invention, the processing precision of the electroencephalogram signal is improved.

Description

Technical Field

[0001] The present application relates to the field of medical devices and signal processing, and in particular to an electroencephalogram signal processing system, method and computer equipment for neurology. Background Art

[0002] Electroencephalogram (EEG), as an important tool to reflect the state of brain activity, plays a key role in the diagnosis of neurological diseases. However, traditional EEG signal processing methods have many shortcomings, such as insufficient accuracy and susceptibility to noise interference, which seriously limit their application in clinical high-precision diagnosis. In recent years, deep learning technology has made significant progress in the field of signal processing, providing new possibilities and solutions for EEG signal processing.

[0003] Although there have been some attempts to apply deep learning to EEG signal processing, these methods often only focus on a specific aspect of the EEG signal, such as time domain features, frequency domain features, or spatial features, while ignoring the integrity and dynamics of the EEG signal as a complex system. Therefore, these methods still need to be further improved in terms of improving the accuracy and stability of EEG signal processing. Summary of the invention

[0004] The purpose of this application is to provide an electroencephalogram signal processing system, method and computer equipment, which can improve the accuracy of electroencephalogram signal processing.

[0005] To achieve the above objectives, this application provides the following solutions:

[0006] In a first aspect, the present application provides an electroencephalogram signal processing system, comprising:

[0007] A signal acquisition module, used to collect the patient's electroencephalogram signal;

[0008] A feature extraction module, connected to the signal acquisition module, is used to extract features from the EEG signal to obtain EEG features; the EEG features include time domain features, frequency domain features and nonlinear dynamic features;

[0009] A classification module connected to the feature extraction module, for determining the category of the EEG signal according to the EEG feature based on a pre-trained EEG classification model; the EEG classification model includes a convolutional neural network, a long short-term memory network and a classifier connected in sequence; the categories include a normal state and different disease states;

[0010] A display module is connected to the classification module and is used to visualize the EEG signal and the category of the EEG signal.

[0011] Optionally, the signal acquisition module is a multi-channel lead electrode; the multi-channel lead electrode is placed on the patient's scalp to collect the patient's electroencephalogram signal.

[0012] Optionally, the electroencephalogram signal processing system further includes:

[0013] The preprocessing module is connected to the signal acquisition module and the feature extraction module respectively, and is used to perform denoising, filtering and baseline correction on the EEG signal in sequence, and send the processed EEG signal to the feature extraction module for feature extraction.

[0014] Optionally, the feature extraction module includes:

[0015] A time domain feature extraction unit, used for segmenting the electroencephalogram signal, dividing the electroencephalogram signal into a plurality of continuous time periods in time sequence, and calculating the statistics of the electroencephalogram signal in each time period to obtain a time domain feature;

[0016] A frequency domain feature extraction unit, used for performing a fast Fourier transform on the EEG signal to obtain a spectrum, smoothing the spectrum, and extracting key frequency components in the smoothed spectrum to obtain time domain features;

[0017] The nonlinear dynamic feature extraction unit is used to convert the electroencephalogram signal into a high-dimensional phase space, and calculate the nonlinear dynamic parameters of the electroencephalogram signal in the phase space to obtain nonlinear dynamic features.

[0018] Optionally, the EEG feature also includes a spatial feature; and the feature extraction module also includes:

[0019] The spatial feature extraction unit is used to perform spatial distribution analysis on the electroencephalogram signal, and to construct a brain function connection network of the electroencephalogram signal using a brain network analysis method, and to analyze the topological structure and dynamic changes of the brain function connection network to obtain spatial features.

[0020] Optionally, the display module is also used to construct a visual image of the brain functional connection network to demonstrate the connection strength and dynamic changes between different brain regions.

[0021] Optionally, the display module uses a time series graph to display the changing trend of the EEG signal over time, and marks the category of the EEG signal in the time series graph.

[0022] Optionally, the display module is a display screen; the feature extraction module and the classification module are both functional modules in a processor.

[0023] In a second aspect, the present application provides an electroencephalogram signal processing method, comprising:

[0024] Collecting the patient's electroencephalogram signal;

[0025] Extracting features from the electroencephalogram signal to obtain electroencephalogram features; the electroencephalogram features include time domain features, frequency domain features and nonlinear dynamic features;

[0026] According to the EEG features, the category of the EEG signal is determined based on a pre-trained EEG classification model; the EEG classification model includes a convolutional neural network, a long short-term memory network and a classifier connected in sequence; the categories include a normal state and different disease states;

[0027] The EEG signals and categories of the EEG signals are visualized.

[0028] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned electroencephalogram signal processing method.

[0029] According to the specific embodiments provided in this application, this application has the following technical effects:

[0030] The present application provides an EEG signal processing system, method and computer equipment, which comprehensively considers the time domain characteristics, frequency domain characteristics and nonlinear dynamic characteristics, can comprehensively analyze the EEG signal, and improve the information utilization rate and classification accuracy. The EEG classification model constructed by the convolutional neural network and the long-term and short-term extreme network realizes the comprehensive and high-precision analysis of the EEG signal, and visualizes the EEG signal and its category, which can intuitively display the analysis results and help doctors make more accurate diagnoses. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0032] Figure 1 A schematic diagram of the structure of an electroencephalogram signal processing system provided in one embodiment of the present application;

[0033] Figure 2 A schematic diagram of a flow chart of an electroencephalogram signal processing method provided in one embodiment of the present application; DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0035] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0036] In an exemplary embodiment, Figure 1 As shown, an electroencephalogram signal processing system is provided, including: a signal acquisition module 101, a preprocessing module 102, a feature extraction module 103, a classification module 104 and a display module 105.

[0037] In an exemplary embodiment, the signal acquisition module 101 is a multi-channel lead electrode. The multi-channel lead electrode is placed on the patient's scalp to collect the patient's electroencephalogram signal. The preprocessing module 102, the feature extraction module 103 and the classification module 104 are all functional modules in the processor. The display module 105 is a display screen. The display screen can be a display screen provided by the processor, or a separate display screen connected to the processor.

[0038] The signal acquisition module 101 is used to acquire the patient's electroencephalogram signal.

[0039] The preprocessing module 102 is connected to the signal acquisition module 101 and the feature extraction module 103 respectively. The preprocessing module 102 is used to perform denoising, filtering and baseline correction on the EEG signal in sequence, and send the processed EEG signal to the feature extraction module 103 for feature extraction.

[0040] The denoising process uses the Independent Components Analysis (ICA) method and the wavelet transform denoising method, and the filtering process uses the bandpass filtering method.

[0041] ICA is used to separate independent source signals from multi-channel EEG signals. In EEG signal processing, ICA can identify and remove signal sources that are not related to brain activity, such as eye movements, cardiac activity, etc. Through ICA analysis, independent components that are more relevant to brain activity can be extracted, thereby improving the quality of EEG signals.

[0042] Wavelet transform is used to decompose EEG signals into components of different frequencies and scales. Wavelet transform is used to remove noise and interference while retaining the signal's detailed information. By selecting the appropriate wavelet basis and decomposition layer number, the EEG signal can be finely denoised.

[0043] Unnecessary frequency components in the EEG signal are removed by bandpass filtering. By setting an appropriate frequency range, the frequency components related to brain activity are retained, while low-frequency noise such as breathing and heartbeat, and high-frequency noise such as electromyographic interference are removed.

[0044] The purpose of baseline correction is to eliminate the baseline offset problem in the EEG signal. Through baseline correction, the baseline of the EEG signal can be restored to zero level, thereby ensuring the accuracy of subsequent analysis.

[0045] It should be noted that the order of filtering and denoising processing can be adjusted, and the order of ICA and wavelet transform can also be adjusted.

[0046] The feature extraction module 103 is connected to the signal acquisition module 101 and is used to extract features from the EEG signal to obtain EEG features. The EEG features include time domain features, frequency domain features, nonlinear dynamic features and spatial features.

[0047] In an exemplary embodiment, the feature extraction module 103 includes: a time domain feature extraction unit, a frequency domain feature extraction unit, a nonlinear dynamic feature extraction unit and a space feature extraction unit.

[0048] The time domain feature extraction unit is used to perform segmentation processing on the electroencephalogram signal, divide the electroencephalogram signal into multiple continuous time periods in time sequence, and calculate the statistics of the electroencephalogram signal in each time period to obtain the time domain feature.

[0049] The time period can also be called a "window". The length of each time period can be set to a few seconds or tens of seconds. Statistics include mean, standard deviation, maximum, minimum, etc., to reflect the statistical characteristics of the EEG signal in different time periods. Time domain features include statistics such as the mean, standard deviation, maximum, minimum, etc. of the EEG signal in each time period.

[0050] The frequency domain feature extraction unit is used to perform Fast Fourier Transform (FFT) on the EEG signal to obtain a spectrum, smooth the spectrum, and extract key frequency components in the smoothed spectrum to obtain time domain features.

[0051] By performing FFT on the EEG signal and converting it from the time domain to the frequency domain, FFT can reveal the energy distribution of the EEG signal at different frequencies, that is, the spectrum. The obtained spectrum is smoothed to reduce the influence of noise and artifacts. Smoothing can be achieved through methods such as moving average and Gaussian filtering. Key frequency components, such as the main frequency and the secondary frequency, are extracted from the smoothed spectrum. These frequency components can reflect the activity of the EEG signal at specific frequencies, such as alpha waves, beta waves and other brain wave components. Frequency domain features include the extracted key frequency components.

[0052] The nonlinear dynamic feature extraction unit is used to convert the electroencephalogram signal into a high-dimensional phase space, and calculate the nonlinear dynamic parameters of the electroencephalogram signal in the phase space to obtain nonlinear dynamic features.

[0053] By converting the one-dimensional EEG signal into a high-dimensional phase space, the dynamic behavior of the EEG signal in the high-dimensional space is revealed. Phase space reconstruction can be achieved by methods such as delayed embedding. The nonlinear dynamic parameters such as attractors and Lyapunov exponents of the EEG signal are calculated in the phase space. These nonlinear dynamic parameters can reflect the nonlinear characteristics and complexity of the EEG signal, such as chaotic behavior and self-similarity. The nonlinear dynamic characteristics include nonlinear dynamic parameters such as attractors and Lyapunov exponents of the EEG signal.

[0054] The spatial feature extraction unit is used to perform spatial distribution analysis on the electroencephalogram signal, and use a brain network analysis method to construct a brain function connection network of the electroencephalogram signal, and analyze the topological structure and dynamic changes of the brain function connection network to obtain spatial features.

[0055] Among them, by performing spatial distribution analysis on the EEG signals collected by multi-channel lead electrodes, the coherence, synchronization and changes in spatial patterns between electrodes are determined. These indicators can reflect the connection and synergy between different brain regions.

[0056] The nodes in the brain functional connection network represent different brain region electrodes, and the edges represent the connection strength between brain regions. Analyze the topological structure and dynamic changes of the brain functional connection network, such as the sparsity, modularity, and node importance of the brain functional connection network. These parameters can reveal the organizational structure and dynamic characteristics of the brain functional connection network, and provide important clues for understanding the functions and mechanisms of the brain.

[0057] Spatial characteristics include spatial distribution analysis results such as coherence, synchronization and changes in spatial patterns between electrodes, as well as the topological structure and dynamic change parameters of the brain functional connection network such as sparsity, modularity, and node importance.

[0058] The classification module 104 is connected to the feature extraction module 103 and is used to determine the category of the EEG signal based on the EEG feature and a pre-trained EEG classification model. The EEG classification model includes a convolutional neural network (CNN), a long short-term memory network (LSTM) and a classifier connected in sequence. The categories include normal states and different disease states.

[0059] Firstly, CNN is used to learn the spatial features of EEG features to capture the local patterns and structural information in the EEG signals. Then, LSTM is used to learn the temporal features to capture the time series dependencies in the EEG signals. Finally, the classifier is used to output the categories according to the output results of LSTM.

[0060] This application uses CNN and LSTM to build a deep learning model, train and learn the extracted EEG features, and use a classifier to automatically classify the EEG signals.

[0061] The convolutional layer of CNN is:

[0062] Z l =σ(W l ·X l-1 +b l );

[0063] Among them, Z l is the convolution output of the lth layer, X l-1 is the input of the l-1th layer. For the first layer, X 0 is the EEG feature, W l is the convolution kernel weight of the lth layer, b l is the convolution kernel bias of the lth layer, * represents the convolution calculation, and σ represents the activation function.

[0064] When processing EEG signals, LSTM divides the EEG signals into multiple time steps, or sequences, and applies a hidden state update mechanism at each time step. In this way, LSTM is able to capture the time series dependencies in the EEG signals.

[0065] The hidden state update mechanism of LSTM is:

[0066] h t =LSTM(h t-1 ,x t );

[0067] Among them, h t is the hidden state at time t, x tis the input feature at the tth moment, and the input feature x1 at the 1st moment is the output of the last convolution layer.

[0068] The output of the classifier is represented as:

[0069] y=softmax(W out ·h T +b out );

[0070] Among them, y represents the classification result, usually a probability distribution or category label, and h T represents the hidden state at the last moment, W out is the weight of the classifier, b out is the bias of the classifier.

[0071] During the training process of the EEG classification model, the EEG classification model receives a large number of labeled EEG features as input and tries to learn the labels of these EEG features, such as the mapping relationship between disease states or normal states. The parameters of the EEG classification model, including the convolution kernel weights and biases of CNN, the weights and biases of LSTM, and the weights and biases of the classifier, are continuously adjusted through the back-propagation algorithm and optimizer to minimize the cross-entropy loss function.

[0072] The display module 105 is connected to the classification module 104 and is used to visualize the EEG signal and the category of the EEG signal.

[0073] In an exemplary embodiment, the display module 105 uses a time series graph to display the change trend of the EEG signal over time, and the category of the EEG signal is marked in the time series graph.

[0074] The display module 105 is also used to construct a visual image of the brain functional connection network to show the connection strength and dynamic changes between different brain regions.

[0075] Furthermore, a detailed classification report is generated based on the classification results, showing the contribution of different EEG features to the classification results based on a radar chart or a bar chart.

[0076] Based on the same inventive concept, the present application also provides a method for processing the electroencephalogram signal, such as Figure 2 As shown, the following steps 201 to 205 are included.

[0077] Step 201, collecting the patient's electroencephalogram signal.

[0078] Step 202: De-noising, filtering and baseline correction are performed on the electroencephalogram signal in sequence.

[0079] Step 203, extracting features from the electroencephalogram signal to obtain electroencephalogram features. The electroencephalogram features include time domain features, frequency domain features, nonlinear dynamic features, and spatial features.

[0080] Step 204, according to the EEG features, determine the category of the EEG signal based on a pre-trained EEG classification model. The EEG classification model includes a convolutional neural network, a long short-term memory network and a classifier connected in sequence. The categories include normal state and different disease states.

[0081] Step 205 : Visualize the EEG signal and the category of the EEG signal.

[0082] The above-mentioned step 201 is implemented by the signal acquisition module 101 in the electroencephalogram signal processing system, step 202 is implemented by the preprocessing module 102 in the electroencephalogram signal processing system, step 203 is implemented by the feature extraction module 103 in the electroencephalogram signal processing system, step 204 is implemented by the classification module 104 in the electroencephalogram signal processing system, and step 205 is implemented by the display module 105 in the electroencephalogram signal processing system.

[0083] This application comprehensively considers time domain features, frequency domain features, nonlinear dynamic features and spatial features to comprehensively analyze EEG signals, which has higher information utilization and classification accuracy than methods that only consider a single or a small number of features. The EEG classification model constructed by CNN and LSTM realizes comprehensive and high-precision analysis of EEG signals. By generating detailed classification reports and visual images, intuitive and easy-to-understand analysis results help doctors make more accurate diagnoses.

[0084] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiment when executing the computer program.

[0085] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiment are implemented.

[0086] In an exemplary embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiment when executed by a processor.

[0087] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0088] In this application, all actions to obtain signals, information or data are carried out in compliance with the relevant data protection laws and policies of the country where they are located and with the authorization given by the owner of the corresponding device.

[0089] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0090] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0091] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0092] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. An electroencephalogram signal processing system, characterized in that: The electroencephalogram signal processing system comprises: A signal acquisition module, used to collect the patient's electroencephalogram signal; A feature extraction module, connected to the signal acquisition module, is used to extract features from the EEG signal to obtain EEG features; the EEG features include time domain features, frequency domain features and nonlinear dynamic features; A classification module connected to the feature extraction module, for determining the category of the EEG signal according to the EEG feature based on a pre-trained EEG classification model; the EEG classification model includes a convolutional neural network, a long short-term memory network and a classifier connected in sequence; the categories include a normal state and different disease states; A display module is connected to the classification module and is used to visualize the EEG signal and the category of the EEG signal.

2. The electroencephalogram signal processing system according to claim 1, characterized in that: The signal acquisition module is a multi-channel lead electrode; the multi-channel lead electrode is placed on the patient's scalp to collect the patient's electroencephalogram signal.

3. The electroencephalogram signal processing system according to claim 1, characterized in that: The electroencephalogram signal processing system also includes: The preprocessing module is connected to the signal acquisition module and the feature extraction module respectively, and is used to perform denoising, filtering and baseline correction on the EEG signal in sequence, and send the processed EEG signal to the feature extraction module for feature extraction.

4. The electroencephalogram signal processing system according to claim 1, characterized in that: The feature extraction module comprises: A time domain feature extraction unit, used for segmenting the electroencephalogram signal, dividing the electroencephalogram signal into a plurality of continuous time periods in time sequence, and calculating the statistics of the electroencephalogram signal in each time period to obtain a time domain feature; A frequency domain feature extraction unit, used for performing a fast Fourier transform on the electroencephalogram signal to obtain a spectrum, smoothing the spectrum, and extracting key frequency components in the smoothed spectrum to obtain time domain features; The nonlinear dynamic feature extraction unit is used to convert the electroencephalogram signal into a high-dimensional phase space, and calculate the nonlinear dynamic parameters of the electroencephalogram signal in the phase space to obtain nonlinear dynamic features.

5. The electroencephalogram signal processing system according to claim 4, characterized in that: The EEG feature also includes a spatial feature; the feature extraction module also includes: The spatial feature extraction unit is used to perform spatial distribution analysis on the electroencephalogram signal, and to construct a brain function connection network of the electroencephalogram signal using a brain network analysis method, and to analyze the topological structure and dynamic changes of the brain function connection network to obtain spatial features.

6. The electroencephalogram signal processing system according to claim 5, characterized in that: The display module is also used to construct a visual image of the brain functional connection network to show the connection strength and dynamic changes between different brain regions.

7. The electroencephalogram signal processing system according to claim 1, characterized in that: The display module uses a time series graph to display the changing trend of the EEG signal over time, and marks the category of the EEG signal in the time series graph.

8. The electroencephalogram signal processing system according to claim 1, characterized in that: The display module is a display screen; the feature extraction module and the classification module are both functional modules in a processor.

9. An electroencephalogram signal processing method, using the electroencephalogram signal processing system according to any one of claims 1 to 8, characterized in that: The electroencephalogram signal processing method comprises: Collecting the patient's electroencephalogram signal; Extracting features from the electroencephalogram signal to obtain electroencephalogram features; the electroencephalogram features include time domain features, frequency domain features and nonlinear dynamic features; According to the EEG features, the category of the EEG signal is determined based on a pre-trained EEG classification model; the EEG classification model includes a convolutional neural network, a long short-term memory network and a classifier connected in sequence; the categories include a normal state and different disease states; The EEG signals and categories of the EEG signals are visualized.

10. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the electroencephalogram signal processing method according to claim 9.

Images