Artifact removal method and device for electroencephalogram signals

Abstract

The application provides an artifact removing method and device of electroencephalogram. The method records the electroencephalogram of participants by using an electroencephalogram acquisition device and performs pretreatment; an artifact identification model is used to mark artifacts; the influence degree of different artifact types is quantitatively analyzed; and the electroencephalogram is de-artifacted by using a comprehensive PCA-LLE analysis method. By introducing the pretreatment of electroencephalogram, the artifact identification model and the de-artifact method based on PCA-LLE, the application realizes accurate classification and effective removal of artifact signals, simultaneously performs quantitative analysis on artifact interference, maximally retains the integrity and effectiveness of electroencephalogram, and thus solves the problems of inaccurate artifact identification of electroencephalogram, decreased electroencephalogram integrity caused by artifact processing and limited processing capacity of complex artifacts in the prior art.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of electroencephalogram signal processing, and particularly relates to a method and device for removing artifacts from electroencephalogram signals, a computer readable storage medium and an electroencephalogram monitoring system. BACKGROUND

[0002] An electroencephalogram signal is a weak, nonlinear and non-stationary signal, and its characteristics make it extremely susceptible to artifact interference, especially during actual collection, which is mixed with a large amount of artifact components. To effectively apply the electroencephalogram signal to scientific research, it is necessary to remove these artifacts.

[0003] In the early stage of electroencephalogram signal research, researchers usually use experimental control methods to suppress the generation of artifacts. For example, by strictly restricting the behavior of the experimenter, such as trying to avoid eye movement, blinking, swallowing and other actions, to reduce the interference of electrooculogram and electromyogram artifacts. Although this method can reduce the influence of artifacts to a certain extent, it causes additional psychological burden on the experimental object, which may activate unnecessary brain activity, and thus adversely affect the accuracy of the experimental results.

[0004] In addition, there is a method of removing artifacts by removing the brain electrical signal band similar to the artifact waveform characteristics. However, this method usually results in the loss of part of the useful signal, further reducing the integrity of the electroencephalogram signal and the reliability of the analysis.

[0005] During actual collection, due to the interference of other physiological activities of the human body, such as eye blinking, muscle movement, heartbeat, etc., artifact signals are inevitably generated. Eye blinking and eye movement artifacts are the most common sources of artifacts. The amplitude of the electrooculogram signal is usually hundreds or even thousands of times that of the electroencephalogram signal, and the interference with the electroencephalogram signal is extremely serious. Electromyogram signals are prone to occur when the subject is nervous, and their high-frequency characteristics make them difficult to filter directly. In addition, non-physiological artifacts such as power frequency interference, collection electrode and amplifier noise also have a significant impact on the accurate recording of the electroencephalogram signal.

[0006] Although the current method can partially alleviate the above problems, there are still many deficiencies. For example, limiting the behavior of the subject to reduce artifacts is simple, but it may lead to limitations in the experimental environment and affect the natural state of the electroencephalogram signal; the waveform removal method is effective, but it cannot accurately distinguish between artifacts and useful signals, and there is a risk of losing important information. At the same time, the effectiveness of removing complex artifacts and real-time identification of artifact signals still needs to be further improved. These problems significantly limit the reliability and universality of the electroencephalogram signal in practical applications. SUMMARY

[0007] The main purpose of the present application is to provide an electroencephalogram artifact removal method, an electroencephalogram artifact removal device, a computer readable storage medium and an electroencephalogram monitoring system to at least solve the problems of inaccurate artifact identification, reduced electroencephalogram integrity due to artifact processing and limited complex artifact processing capability in the prior art.

[0008] To achieve the above-mentioned purpose, according to one aspect of the present application, an electroencephalogram artifact removal method is provided, comprising: acquiring a first target signal by an electroencephalogram acquisition device, pre-processing the first target signal to obtain a second target signal, the pre-processing at least including filtering, cropping and removing non-physiological artifact signals in the first target signal; inputting the second target signal into an artifact identification model to classify the second target signal and obtain the artifact signals in the second target signal; quantitatively analyzing the artifact signals to determine the first influence degree of each artifact signal on the electroencephalogram; removing the artifact signals in the second target signal based on the first influence degree by a PCA-LLE analysis method to obtain a target electroencephalogram.

[0009] Optionally, the electroencephalogram acquisition device includes a plurality of electrode pieces, and the first target signal is obtained by signal acquisition by the electrode pieces installed at predetermined positions, and the first target signal includes electroencephalogram, electrooculogram, electrocardiogram, electromyogram and artifact signals, the electrooculogram includes blink frequency and eye movement trajectory, the electrocardiogram includes electrocardiogram data and heart rate variability data, the electromyogram includes electromyogram data, electromyographic activity frequency and electromyographic activity amplitude, and the artifact signals include environmental noise data, device interference data and background variables.

[0010] Optionally, the pre-processing of the first target signal to obtain the second target signal includes: band-pass filtering based on the first target signal to remove power frequency interference and low-frequency noise to obtain a third target signal; dividing the third target signal by a preset time window in a sliding window manner to obtain a plurality of fourth target signals; generating a frequency spectrum based on the fourth target signal and calculating a power spectral density based on the frequency spectrum; in the case that the power spectral density is greater than a first threshold, determining the fourth target signal as the artifact signal and screening out; in the case that the power spectral density is less than or equal to the first threshold, determining the fourth target signal as the second target signal.

[0011] Optionally, the second target signal is input into a artifact identification model to classify the second target signal to obtain the artifact signal in the second target signal, comprising: obtaining a reference signal, the reference signal comprising a reference electrooculogram signal, a reference electrocardiogram signal and a reference electromyogram signal, calculating an identification strength based on the reference signal to obtain a target identification strength; calculating a classification score based on the reference signal by a strong classifier to obtain a first classification score, the strong classifier being obtained based on a combination of a KNN classifier and a logistic regression classifier; determining a target discrimination threshold based on the first classification score based on a first preset formula: T=T base + β · (C final - C mean ); wherein T is the target discrimination threshold, T base is a preset classification strength threshold, β is an adjustment coefficient of the first classification score on the target discrimination threshold, C mean is a mean value of the reference signal corresponding to the target discrimination threshold, C final is the target discrimination threshold corresponding to the reference signal; updating the intensity of the second target signal based on the target discrimination threshold to obtain a fifth target signal, and classifying based on the fifth target signal and the target discrimination threshold to obtain the artifact signal in the second target signal.

[0012] Optionally, updating the intensity of the second target signal based on the target discrimination threshold to obtain a fifth target signal, and classifying based on the fifth target signal and the target discrimination threshold to obtain the artifact signal in the second target signal, comprising: determining an influence degree of the artifact signal on the target electroencephalogram signal based on the intensity of the second target signal to obtain a second influence degree; updating the intensity of the second target signal based on the first influence degree and the target discrimination threshold to obtain the fifth target signal: wherein STR' is the intensity of the fifth target signal, I is the first influence degree, δ is an adjustment coefficient of the intensity of the second target signal and the estimated intensity of the artifact signal on the intensity of the fifth target signal, STR total is the intensity of the second target signal, and STR new is the estimated intensity of the artifact signal; in the case that the intensity of the fifth target signal is greater than or equal to a second threshold, determining that the fifth target signal is the electrooculogram signal; in the case that the intensity of the fifth target signal is greater than or equal to a third threshold and less than the second threshold, determining that the fifth target signal is the electrocardiogram signal; in the case that the intensity of the fifth target signal is greater than or equal to a fourth threshold and less than the third threshold, determining that the fifth target signal is the electromyogram signal; and determining the electrooculogram signal, the electrocardiogram signal and the electromyogram signal as the artifact signal.

[0013] Optionally, the artifact signals are quantitatively analyzed to determine a first influence degree of each of the artifact signals on the electroencephalogram signal, comprising: inputting the artifact signals into an artifact influence model to obtain the first influence degree: ψ(W i ,No i ) = μ1·(W i - μ2·No i ); wherein Q is the first influence degree of the artifact signal, n is a kind of the artifact signal, I i is the second influence degree corresponding to the artifact signal in the i-th, ψ(W i ,No i ) is a relationship between the artifact signal W i and a noise component No i introduced when the artifact signal W i is collected, and μ1 and μ2 are preset coefficients.

[0014] Optionally, the artifact signals in the second target signal are removed by a PCA-LLE analysis method based on the first influence degree to obtain a target electroencephalogram signal, comprising: in a case where the first influence degree is greater than a fifth threshold, sequentially removing the artifact signals in the second target signal by a PCA algorithm and an LLE algorithm to obtain the target electroencephalogram signal; in a case where the first influence degree is less than or equal to the fifth threshold, sequentially removing the artifact signals in the second target signal by an LLE algorithm and a PCA algorithm to obtain the target electroencephalogram signal.

[0015] According to another aspect of the present application, there is provided an artifact removal device for an electroencephalogram signal, the device comprising: a first acquisition unit configured to perform signal acquisition on a target object by an electroencephalogram wave acquisition device to obtain a first target signal, and to perform preprocessing on the first target signal to obtain a second target signal, wherein the preprocessing at least includes filtering, cropping, and removing non-physiological artifact signals in the first target signal; a classification unit configured to input the second target signal into an artifact recognition model to classify the second target signal to obtain the artifact signals in the second target signal; a first calculation unit configured to quantitatively analyze the artifact signals to determine a first influence degree of each of the artifact signals on the electroencephalogram signal; and a second calculation unit configured to remove the artifact signals in the second target signal by a PCA-LLE analysis method based on the first influence degree to obtain a target electroencephalogram signal.

[0016] According to still another aspect of the present application, a computer readable storage medium is provided, the computer readable storage medium comprising a stored program, wherein the computer readable storage medium is caused to perform any of the methods described when the program is run.

[0017] According to still another aspect of the present application, an electroencephalogram monitoring system is provided, comprising one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs comprising instructions for performing any of the methods described.

[0018] According to the technical solution of the present application, first, a signal of a target object is collected by an electroencephalogram acquisition device to obtain a first target signal, the first target signal is preprocessed to obtain a second target signal, and the preprocessing at least includes filtering, cropping, and removing non-physiological artifact signals in the first target signal; then the second target signal is input into an artifact identification model to classify the second target signal to obtain the artifact signals in the second target signal; thereafter, the artifact signals are quantitatively analyzed to determine a first influence degree of each artifact signal on the electroencephalogram signal; finally, the artifact signals in the second target signal are removed based on the first influence degree by a PCA-LLE analysis method to obtain a target electroencephalogram signal. By introducing the preprocessing of the electroencephalogram signal, the artifact identification model, and the artifact removal method based on PCA-LLE, the present application realizes accurate classification and effective removal of the artifact signals, simultaneously quantitatively analyzes the artifact interference, maximally retains the integrity and effectiveness of the electroencephalogram signal, and thus solves the problems of inaccurate artifact identification of the electroencephalogram signal, reduced integrity of the electroencephalogram signal due to artifact processing, and limited processing capacity of complex artifacts in the prior art. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 A hardware structure block diagram of a mobile terminal for removing artifacts of an electroencephalogram signal according to an embodiment of the present application is shown;

[0020] Figure 2 A flowchart of a method for removing artifacts of an electroencephalogram signal according to an embodiment of the present application is shown;

[0021] Figure 3 A structure block diagram of an apparatus for removing artifacts of an electroencephalogram signal according to an embodiment of the present application is shown.

[0022] Among the above drawings, the following reference signs are included:

[0023] 102, processor; 104, memory; 106, transmission device; 108, input / output device. Detailed Implementation

[0024] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0025] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0026] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0027] For ease of description, the following explains some of the nouns or terms used in the embodiments of this application:

[0028] PCA-LLE (Principal Component Analysis-Locally Linear Embedding) is a comprehensive processing technique that combines linear and nonlinear dimensionality reduction methods. It aims to leverage the advantages of both PCA and LLE to achieve efficient dimensionality reduction and feature extraction. The PCA component performs initial dimensionality reduction on high-dimensional data, quickly removing noise and redundant features while preserving the main linear structure, thus reducing computational complexity. The LLE component performs nonlinear dimensionality reduction after PCA processing, further capturing local nonlinear relationships and preserving the manifold structure and local geometric properties of the data. Through this staged processing approach, PCA-LLE can reduce computational costs while considering both linear and nonlinear characteristics of the data, making it particularly suitable for the analysis and processing of complex high-dimensional data. This method is commonly used in scenarios requiring accurate feature extraction, such as pattern recognition, signal processing, and image analysis.

[0029] KNN (K-Nearest Neighbors): KNN is a simple and efficient classification and regression algorithm. Its basic idea is: for a sample to be classified, the class of the K nearest neighbors in the sample space is used to determine the classification result of the sample. In the classification task, the majority class among the neighbors is usually used to determine the result; in the regression task, the average value of the neighbors is used as the predicted value. KNN does not require a training process, has low model complexity, and is suitable for solving linear or nonlinear problems, but has low computational efficiency for large-scale data and is greatly affected by data distribution.

[0030] As introduced in the background, the prior art has deficiencies in the accuracy of artifact identification, signal integrity preservation, and complex artifact processing capability, and it is difficult to fully meet the needs of high-precision applications of electroencephalogram signals. To solve the problems of inaccurate artifact identification of electroencephalogram signals, reduced electroencephalogram signal integrity caused by artifact processing, and limited complex artifact processing capability in the prior art, embodiments of the present application provide an artifact removal method for electroencephalogram signals, an artifact removal device for electroencephalogram signals, a computer readable storage medium, and an electroencephalogram monitoring system.

[0031] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0032] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal, or a similar computing device. Taking a mobile terminal as an example, Figure 1 is a hardware structure block diagram of a mobile terminal of an electroencephalogram signal artifact removal method according to an embodiment of the present application. As shown in Figure 1 , the mobile terminal can include one or more (only one is shown in Figure 1 ) processor 102 (the processor 102 can include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the above-mentioned mobile terminal can further include a transmission device 106 for communication function and an input and output device 108. Those skilled in the art can understand that Figure 1 the structure shown is only schematic, and it does not limit the structure of the above-mentioned mobile terminal. For example, the mobile terminal can include more or fewer components than those shown in Figure 1 , or have a different configuration from that shown in Figure 1 .

[0033] The memory 104 can be used to store computer programs, such as software programs of application software and modules, such as a computer program corresponding to the device information display method in the embodiments of the present application. The processor 102 executes various functional applications and data processing, i.e., implements the above method, by running the computer programs stored in the memory 104. The memory 104 can include a high-speed random access memory, and can further include a non-volatile memory, such as one or more magnetic storage devices, flash memories, or other non-volatile solid-state memories. In some examples, the memory 104 can further include memories disposed remotely with respect to the processor 102, which can be connected to the mobile terminal through a network. Examples of the above network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof. The transmission device 106 is used to receive or send data via a network. The specific examples of the above network can include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet in a wireless manner.

[0034] In the embodiments, a method for removing artifacts of electroencephalogram signals running on a mobile terminal, a computer terminal or a similar computing device is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described herein can be executed in an order different from that shown.

[0035] Figure 2 is a flowchart of a method for removing artifacts of electroencephalogram signals according to the embodiments of the present application. As shown in Figure 2 , the method comprises the following steps:

[0036] In step S201, a signal of a target object is collected by an electroencephalogram acquisition device to obtain a first target signal, and the first target signal is preprocessed to obtain a second target signal. The preprocessing at least includes filtering, cropping and removing non-physiological artifact signals in the first target signal.

[0037] Specifically, electrodes are placed according to standard scalp electrode positions, and electroencephalogram signals of participants in a resting state are recorded, while electrooculogram signals, electrocardiogram signals, and electromyogram signals are collected, and non-physiological signals of the equipment are recorded during the collection process, and artifact removal features are extracted. The preprocessing content includes removing power frequency interference and low frequency noise signals by using a band-pass filtering method, segmenting and cutting the signals, and removing obvious non-physiological artifact signals.

[0038] Step S202, inputting the second target signal into an artifact identification model to classify the second target signal to obtain artifact signals in the second target signal;

[0039] Specifically, the electroencephalogram signals obtained after removing the non-physiological artifact signals are input into the trained artifact identification model. The artifact identification model is based on a deep learning or machine learning algorithm and can automatically classify the artifact types in the signals, such as eye movement artifacts, electromyographic artifacts, and electrocardiographic artifacts, and generate artifact labels.

[0040] Step S203, quantitatively analyzing the artifact signals to determine the first influence degree of each artifact signal on the electroencephalogram signals;

[0041] Specifically, an artifact influence model is constructed to quantitatively analyze the influence degree of different artifact types, and the artifact signals are quantitatively analyzed, mainly including: counting the average amplitude and standard deviation of the artifact signals; analyzing the energy distribution of the artifact signals in a specific frequency band; and evaluating the trend of the artifact signals over time. Through these analyses, the interference degree of the artifact signals on the electroencephalogram signals is quantified, and the first influence degree of each artifact signal is obtained.

[0042] Step S204, removing the artifact signals in the second target signal based on the first influence degree by a PCA-LLE analysis method to obtain target electroencephalogram signals.

[0043] Specifically, based on the quantitative analysis results, a PCA-LLE (principal component analysis combined with local linear embedding) algorithm is used to perform dimensionality reduction processing on the second target signal to remove the components related to artifacts in the signal. PCA (principal component analysis) extracts the principal components of the signal to separate the artifact components; LLE (local linear embedding) further optimizes the signal embedding to retain important features in the electroencephalogram signals and remove the artifact signals. Finally, high-quality target electroencephalogram signals are obtained.

[0044] Through the embodiment, the application provides a method for removing artifacts of electroencephalogram signals. The electroencephalogram signals of participants are recorded by an electroencephalogram acquisition device and preprocessed, the artifacts are labeled by an artifact identification model, the influence degree of different artifact types is quantitatively analyzed by constructing an artifact influence model, and the electroencephalogram signals are de-artifacted by a comprehensive PCA-LLE analysis method. The application aims to solve the technical problem that electroencephalogram signals are easily disturbed by electrooculogram, electrocardiogram, electromyogram artifacts and environmental noise during the collection process. By constructing an artifact identification model and an artifact influence model, quantitative analysis and effective removal of artifact signals are realized, and the reliability and analysis accuracy of electroencephalogram signal data are improved. The comprehensive PCA-LLE analysis method is adopted to reduce the data dimension and retain the local geometric structure of the signal, so as to improve the de-artifact effect, improve the quality of electroencephalogram signals, and solve the problems of inaccurate artifact identification of electroencephalogram signals, reduced integrity of electroencephalogram signals caused by artifact processing, and limited processing capacity of complex artifacts in the prior art.

[0045] As a possible implementation manner, the electroencephalogram acquisition device includes a plurality of electrode pads. The target object is signal collected by the electroencephalogram acquisition device to obtain a first target signal, which includes:

[0046] The signal is collected by the electrode pads installed at the preset positions respectively, a plurality of signals are obtained and integrated to obtain the first target signal. The first target signal includes electroencephalogram signals, electrooculogram signals, electrocardiogram signals, electromyogram signals and artifact signals. The electrooculogram signals include blink frequency and eye movement trajectory. The electrocardiogram signals include electrocardiogram data and heart rate variability data. The electromyogram signals include electromyogram data, electromyogram activity frequency and electromyogram activity amplitude. The artifact signals include environmental noise data, device interference data and background variables.

[0047] Specifically, the brain wave collection device collects the brain electrical signals of the target object through multiple electrode pads. The brain electrical signals are time series data collected by the EEG device, which can reflect the energy distribution of different frequency bands, and form a multi-channel voltage data matrix by recording the potential difference between different electrodes. These electrode pads are installed on the scalp or other preset positions of the target object, and the electrophysiological signals of each region are collected through the electrode pads respectively, and finally the first target signal is integrated. The blink frequency is the number of blinks of the participant during the experiment, the eye movement trajectory is the angle change of the eye movement monitoring, the electromyography data is the potential change during the muscle contraction and relaxation process, the electromyography activity frequency and amplitude, the device interference data is the device state change or other non-physiological signals that may affect the signal quality during the experiment, and the background variable is the potential influence of variables related to the experimental environment, such as temperature, humidity, etc. on the signal quality. Among them, the brain electrical signals are the main concern, and other signals (electrooculogram, electrocardiogram, electromyogram and artifact signals) may affect the accuracy of the brain electrical signals. Therefore, in the subsequent processing process, these signals need to be removed and corrected, so as to obtain cleaner and more reliable brain electrical signal data.

[0048] As a possible implementation manner, the first target signal is preprocessed to obtain a second target signal, including:

[0049] Step S301, band-pass filtering based on the first target signal to remove power frequency interference and low-frequency noise to obtain a third target signal;

[0050] Specifically, the acquired brain electrical signals are removed by a band-pass filter to remove 50 / 60Hz power frequency interference and low-frequency noise, and the effective frequency band of the brain electrical signals is retained. The third target signal has removed the power frequency interference and low-frequency noise, but still contains possible artifact signals.

[0051] Step S302, the third target signal is divided by a preset time window in a sliding window manner to obtain a plurality of fourth target signals;

[0052] Specifically, the signal is segmented, a 2-second time window is set, and a sliding window manner is adopted. After sliding window division, different brain electrical signal segments S i are obtained. i S

[0053] Step S303, generating a spectrum diagram based on the fourth target signal and calculating the power spectral density based on the spectrum diagram;

[0054] Specifically, the segmented signal data is made into a data spectrum diagram and the power spectral density after segmentation is calculated: Wherein, PSD band (Si ) represents the i-th signal segment S i The power spectrum density calculated in the specified frequency band, [f1, f2] is the physiological signal frequency band, [f3, f4] is the non-physiological artifact signal frequency band, f is the power of the signal at each frequency point, c is a fixed value important to the weighted artifact-related frequency band, X noise (f) is the Fourier transform result of the ambient noise in the signal;

[0055] Step S304, in the case where the power spectrum density is greater than the first threshold value, the fourth target signal is determined as an artifact signal, and is screened out;

[0056] Step S305, in the case where the power spectrum density is less than or equal to the first threshold value, the fourth target signal is determined as a second target signal.

[0057] Specifically, the first artifact signal screening is performed, and the obvious non-physiological artifact signal in the spectrum diagram is screened out:

[0058] P i = w1·PSD band (S i )+ w2·EMG feature (S i )+ w3·EOG feature (S i );

[0059] Wherein, P i represents the comprehensive artifact judgment criterion of the i-th signal segment, PSD represents the power spectrum density, w1, w2, w3 represent the importance weight of each type of feature, EMG feature (S i ) represents the electromyographic activity frequency of the i-th signal segment, which is used to judge whether the current frequency band is disturbed by electromyographic artifact, EOG feature (S i ) represents the blink frequency of the i-th signal segment, which is used to judge whether the current frequency band is disturbed by electrooculographic artifact.

[0060] If P th is the first threshold value, when P i > P th , mark S i as a non-physiological artifact signal for screening out, when P i ≤ P th , otherwise, mark as an effective physiological signal.

[0061] Through the above steps, the first target signal is effectively pre-processed and artifact signals are screened out to obtain a second target signal. The method ensures that the second target signal obtained finally has higher accuracy and reliability through technical means such as band-pass filtering, sliding window segmentation, and spectrum analysis, and provides cleaner data input for subsequent electroencephalogram signal analysis.

[0062] As a possible implementation manner, the second target signal is input into an artifact identification model to classify the second target signal to obtain artifact signals in the second target signal, including:

[0063] In step S401, a reference signal is obtained, the reference signal including a reference electrooculogram signal, a reference electrocardiogram signal, and a reference electromyogram signal, and an identification strength is calculated based on the reference signal to obtain a target identification strength;

[0064] Specifically, the reference signal is obtained, including a reference electrooculogram signal, a reference electrocardiogram signal, and a reference electromyogram signal. These reference signals are usually physiological signals of the participant under normal conditions, and are used for comparison with the target signal. The identification strength can represent the degree of obviousness of the artifact signal, which is calculated by comparing the difference between the target signal and the reference signal.

[0065] The electroencephalogram signal obtained after pre-processing and removing non-physiological artifact signals is input into the trained artifact identification model, and the historical data of the participant is continuously input to calculate the identification strength STR of the artifact:

[0066]

[0067] Where STR is the dynamically calculated artifact signal strength, N is the sample number of the participant, and STR 0,i represents the initial estimate of the i th artifact signal strength, F i is a specific feature parameter of the i th artifact signal, F mean represents the average feature of the historical data of the participant, and a is a regulation factor for controlling the influence of feature deviation, and i is a variable index;

[0068] In step S402, a classification score is calculated based on the reference signal by a strong classifier to obtain a first classification score, and the strong classifier is obtained based on the combination of a KNN classifier and a logistic regression classifier.

[0069] Specifically, an ensemble learning strategy is adopted to combine the KNN classifier and the logistic regression classifier into a strong classifier, and the output of each classifier is defined as C j , and the comprehensive classification score C final is:

[0070] Where C finalrepresents the final classification result after the output of all base classifiers, M is the number of base classifiers, C j is the output result of the judgment of the jth base classifier on the current condition, j is a variable index, STR max represents the maximum intensity in all artifact signals;

[0071] In step S403, based on the first classification score, a target discrimination threshold is determined based on a first preset formula:

[0072] T = T base + β · (C final - C mean );

[0073] Wherein, T is the target discrimination threshold, T base is a preset classification intensity threshold, β is an adjustment coefficient of the first classification score on the target discrimination threshold, C mean is the mean value of the target discrimination threshold corresponding to the reference signal, C final is the target discrimination threshold corresponding to the reference signal;

[0074] Specifically, this formula automatically adjusts the target discrimination threshold according to the change of the classification score, ensuring that the classification process is more flexible and accurate.

[0075] In step S404, the intensity of the second target signal is updated based on the target discrimination threshold to obtain a fifth target signal, and the fifth target signal and the target discrimination threshold are classified to obtain the artifact signal in the second target signal.

[0076] Specifically, the intensity of the second target signal is updated based on the target discrimination threshold to obtain a fifth target signal. By adjusting the intensity of the signal, the distinction between artifacts and effective electroencephalogram signals is enhanced. According to the updated signal intensity and the target discrimination threshold, the fifth target signal is classified to identify the artifact signal. Through this process, the artifact signal in the second target signal can be distinguished from the effective signal.

[0077] As a possible implementation manner, based on the target discrimination threshold, the intensity of the second target signal is updated to obtain a fifth target signal, and the fifth target signal and the target discrimination threshold are classified to obtain the artifact signal in the second target signal, comprising:

[0078] In step S501, based on the intensity of the second target signal, the influence degree of the artifact signal on the target electroencephalogram signal is determined to obtain a second influence degree;

[0079] Specifically, the intensity of the artifact signal and the influence on the electroencephalogram signal are comprehensively considered to calculate the artifact influence I:

[0080] I = ξ · STR total + γ · Jcov ;

[0081] wherein I is the influence degree of the artifact signal on the electroencephalogram signal, and ξ is an adjustment factor of the STR total a weight coefficient of the proportion in the influence degree calculation, and STR total is the intensity of the comprehensive artifact signal, indicating the comprehensive effect of the sum of the artifact signal intensities of all artifacts; and γ is an adjustment factor of the J cov a weight coefficient of the influence degree, and J cov represents the covering influence degree of the artifact signal on the electroencephalogram signal, and characterizes the interference situation caused by the artifact.

[0082] In step S502, the intensity of the second target signal is updated based on the first influence degree and the target discrimination threshold, to obtain a fifth target signal:

[0083]

[0084] wherein STR' is the intensity of the fifth target signal, I is the first influence degree, δ is an adjustment factor of the intensity of the second target signal and the estimated intensity of the artifact signal on the intensity of the fifth target signal, STR total is the intensity of the second target signal, and STR new is the estimated intensity of the artifact signal.

[0085] Specifically, the signal intensity is updated according to the influence degree and the target discrimination threshold, to provide refined intensity data for artifact signal classification.

[0086] In step S503, the fifth target signal is determined as an electrooculogram signal when the intensity of the fifth target signal is greater than or equal to the second threshold value.

[0087] In step S504, the fifth target signal is determined as an electrocardiogram signal when the intensity of the fifth target signal is greater than or equal to the third threshold value and less than the second threshold value.

[0088] In step S505, the fifth target signal is determined as an electromyogram signal when the intensity of the fifth target signal is greater than or equal to the fourth threshold value and less than the third threshold value.

[0089] Specifically, the updated artifact intensity value is used in combination with the discrimination threshold set according to the comprehensive classification score to determine the category of the current artifact signal. By setting multiple threshold values, the signal is classified according to different intensity threshold values, to determine whether the fifth target signal belongs to an electrooculogram signal, an electrocardiogram signal or an electromyogram signal, so as to classify and identify the artifact signal.

[0090] In step S506, the electrooculogram signal, the electrocardiogram signal and the electromyogram signal are determined as artifact signals.

[0091] Specifically, the electro-oculogram signal, the electrocardiogram signal and the electromyogram signal are finally identified as artifact signals, and the artifact signals are removed in the electroencephalogram signal processing.

[0092] The embodiment updates the signal intensity in detail by comprehensively considering the intensity of the artifact signal and the influence on the electroencephalogram signal, and classifies the signals according to different intensity thresholds. Through these steps, the artifact signals can be effectively identified and marked, and clear basis is provided for subsequent artifact removal processing.

[0093] As a possible implementation manner, the artifact signals are subjected to quantitative analysis to determine the first influence degree of each artifact signal on the electroencephalogram signal, including:

[0094] The artifact signal is input into an artifact influence model to obtain the first influence degree:

[0095]

[0096] ψ(W i ,No i )=μ1·(W i -μ2·No i );

[0097] Wherein, Q is the first influence degree of the artifact signal, n is the type of the artifact signal, I i is the second influence degree corresponding to the i artifact signal, ψ(W i ,No i ) is the relationship between the i artifact signal W i and the noise component No i introduced during the collection of the artifact signal W i , and μ1 and μ2 are preset coefficients.

[0098] Specifically, the artifact influence model is constructed, the influence degree of different artifact types is quantitatively analyzed, and the application order of the PCA (principal component analysis) and LLE (locally linear embedding) algorithms is determined according to the influence degree, so as to remove the artifact signals and obtain the target electroencephalogram signal.

[0099] As a possible implementation manner, the application order of the PCA and LLE analysis methods is determined according to the first influence degree of the artifact signal, so as to optimize the artifact removal process. The artifact signal in the second target signal is removed by the PCA-LLE analysis method based on the first influence degree, and the target electroencephalogram signal is obtained, including:

[0100] In step S601, when the first influence degree is greater than the fifth threshold, the artifact signals in the second target signal are removed in sequence by the PCA algorithm and the LLE algorithm, and the target electroencephalogram signal is obtained;

[0101] Specifically, when the influence degree of the artifact signal is large, PCA is used to remove the artifact signal, so as to effectively reduce the global noise, reduce the data dimension, and clear the principal component, so that the LLE can work on the clear structure and reflect the main features of the signal.

[0102] Further, when the PCA is selected to process the signal, the nonlinear transformation for feature reservation is performed:

[0103]

[0104] D = D - E * E T * D PCA is the output data matrix after PCA processing, D is the input data matrix of the multi-channel electroencephalogram signal data, E PCA is the PCA principal component matrix, which is used to project the data to a new feature space; η is an adjustment parameter for controlling the weight of the original signal and the nonlinear mapping result, and the range is [0, 1]; n is the number of signal channels, j is a variable index, D j is the signal data of the jth channel, and σ is an activation function.

[0105] LLE dynamic weighting adjustment is adopted to reserve the local geometric structure: h(D PCA,k ) = R(D PCA ) - p(D PCA,k ).

[0106] D = h(D LLE ) + R(D k ) - p(D PCA,k ). PCA is the output data matrix after LLE processing, K represents the number of neighbors used in LLE reconstruction, ω k represents the importance weight of the kth neighbor in reconstruction, h(D PCA,k ) is a local reconstruction function, which calculates the adjustment amount related to the kth neighbor; R(D PCA ) represents the reconstructed signal from the global feature, and p(D PCA,k ) is a nonlinear function based on the local feature of the PCA output signal.

[0107] The artifact-removed output is:

[0108] D = Q - p(Q) * Q cleaned is the final output signal matrix after artifact removal, p(Q) is an adjustment function of the artifact removal degree of Q, Q represents the artifact component to be removed, e1 is a basic adjustment parameter for setting the minimum artifact removal degree, e2 is an adjustment parameter for the proportion of the artifact removal degree, and l is a decay parameter for adjusting the influence degree of Q on the artifact removal degree.

[0109] In step S602, in a case where the first influence degree is less than or equal to the fifth threshold value, the artifact signal in the second target signal is removed through the LLE algorithm and the PCA algorithm in sequence to obtain a target electroencephalogram signal.

[0110] Specifically, when the influence degree of the artifact signal is small and the local signal structure is complex, LLE is selected for preferential processing, which is suitable for processing signals containing complex local structures, maintains the local geometric characteristics of the data, and can better extract effective information from complex artifacts, but may retain global artifacts. Then, the PCA is used to extract main feature information and remove residual noise. When LLE preferential processing is selected, the order of calculation of D PCA and D LLE is replaced.

[0111] By quantitatively analyzing the influence degree of the artifact signal on the electroencephalogram signal, the application determines the order of use of the PCA and LLE algorithms. In the case of a high influence degree, PCA is preferentially used, and vice versa. This flexible analysis method can optimize the artifact removal process according to the influence of different artifact signals, thereby improving the quality of the electroencephalogram signal.

[0112] The application also provides an artifact removal device for an electroencephalogram signal. It should be noted that the artifact removal device for an electroencephalogram signal provided in the application can be used to execute the artifact removal method for an electroencephalogram signal provided in the application. The device is used to implement the above embodiments and preferred embodiments, and will not be described again. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware, or a combination of software and hardware can also be implemented and conceived.

[0113] The artifact removal device for an electroencephalogram signal provided in the application is described below.

[0114] Figure 3 is a structural block diagram of the artifact removal device for an electroencephalogram signal according to the application. As shown in Figure 3 , the device includes a first acquisition unit 10, a classification unit 20, a first calculation unit 30, and a second calculation unit 40.

[0115] The first acquisition unit 10 is configured to collect a signal of a target object through an electroencephalogram acquisition device to obtain a first target signal, and pre-process the first target signal to obtain a second target signal. The pre-processing at least includes filtering, cropping, and removing non-physiological artifact signals in the first target signal.

[0116] Specifically, electrodes are placed according to standard scalp electrode positions, and electroencephalogram signals of participants in a resting state are recorded, while electrooculogram signals, electrocardiogram signals, and electromyogram signals are collected, and non-physiological signals of the equipment are recorded during the collection process, and artifact removal features are extracted. The preprocessing content includes removing power frequency interference and low frequency noise signals by using a band-pass filtering method, segmenting and cutting the signals, and removing obvious non-physiological artifact signals.

[0117] The classification unit 20 is configured to input the second target signal into an artifact identification model to classify the second target signal to obtain artifact signals in the second target signal.

[0118] Specifically, the electroencephalogram signals obtained after removing the non-physiological artifact signals are input into the trained artifact identification model. The artifact identification model is based on a deep learning or machine learning algorithm and can automatically classify the artifact types in the signals, such as eye movement artifacts, electromyogram artifacts, and electrocardiogram artifacts, and generate artifact labels.

[0119] The first calculation unit 30 is configured to quantitatively analyze the artifact signals to determine the first influence degree of each artifact signal on the electroencephalogram signals.

[0120] Specifically, an artifact influence model is constructed to quantitatively analyze the influence degree of different artifact types and to quantitatively analyze the artifact signals, mainly including: counting the average amplitude and standard deviation of the artifact signals; analyzing the energy distribution of the artifact signals in a specific frequency band; and evaluating the trend of the artifact signals over time. Through these analyses, the interference degree of the artifact signals on the electroencephalogram signals is quantified, and the first influence degree of each artifact signal is obtained.

[0121] The second calculation unit 40 is configured to remove the artifact signals in the second target signal based on the first influence degree by using a PCA-LLE analysis method to obtain target electroencephalogram signals.

[0122] Specifically, based on the quantitative analysis results, a PCA-LLE (principal component analysis combined with local linear embedding) algorithm is used to perform dimensionality reduction processing on the second target signal to remove the components related to artifacts in the signal. PCA (principal component analysis) extracts the principal components of the signal to separate the artifact components, and LLE (local linear embedding) further optimizes the signal embedding to retain important features in the electroencephalogram signals and remove the artifact signals. Finally, high-quality target electroencephalogram signals are obtained.

[0123] The embodiment of the present application provides a device for removing artifacts of an electroencephalogram signal, which comprises a first acquisition unit, a classification unit, a first calculation unit and a second calculation unit. The first acquisition unit is used for signal acquisition on a target object by an electroencephalogram acquisition device to obtain a first target signal, pre-processing of the first target signal to obtain a second target signal, and the pre-processing at least comprises filtering, cropping and removing non-physiological artifact signals in the first target signal; the classification unit is used for inputting the second target signal into an artifact identification model to classify the second target signal to obtain artifact signals in the second target signal; the first calculation unit is used for quantitative analysis of the artifact signals to determine a first influence degree of each artifact signal on the electroencephalogram signal; and the second calculation unit is used for removing the artifact signals in the second target signal based on the first influence degree by a PCA-LLE analysis method to obtain a target electroencephalogram signal. The device records the electroencephalogram signal of a participant by using the electroencephalogram acquisition device and pre-processes the electroencephalogram signal, labels artifacts by using the artifact identification model, quantitatively analyzes the influence degree of different artifact types by constructing an artifact influence model, and removes the artifacts of the electroencephalogram signal by the comprehensive PCA-LLE analysis method. The problems of inaccurate artifact identification of the electroencephalogram signal in the prior art, reduced integrity of the electroencephalogram signal caused by artifact processing and limited processing capacity of complex artifacts are solved.

[0124] As a possible implementation manner, the first acquisition unit comprises an acquisition module.

[0125] The acquisition module is used for signal acquisition by electrodes installed at preset positions to obtain a plurality of signals and integrate the plurality of signals to obtain the first target signal, and the first target signal comprises an electroencephalogram signal, an electrooculogram signal, an electrocardiogram signal, an electromyogram signal and an artifact signal. The electrooculogram signal comprises a blink frequency and an eye movement trajectory. The electrocardiogram signal comprises electrocardiogram data and heart rate variability data. The electromyogram signal comprises electromyogram data, electromyogram activity frequency and electromyogram activity amplitude. The artifact signal comprises environmental noise data, device interference data and background variables.

[0126] Specifically, the electroencephalogram signal is the main object of attention, and other signals (the electrooculogram signal, the electrocardiogram signal, the electromyogram signal and the artifact signal) may affect the accuracy of the electroencephalogram signal. Therefore, in the subsequent processing process, these signals need to be removed and corrected to obtain cleaner and more reliable electroencephalogram signal data.

[0127] As a possible implementation manner, the first acquisition unit further comprises a filtering module, a segmentation module, a generation module, a screening module and a first determination module.

[0128] The filtering module is used for band-pass filtering based on the first target signal to remove power frequency interference and low-frequency noise to obtain a third target signal.

[0129] Specifically, the acquired electroencephalogram signal is removed of 50 / 60 Hz power frequency interference and low frequency noise using a band-pass filter, and the effective frequency band of the electroencephalogram signal is retained. A third target signal is obtained, which has been removed of power frequency interference and low frequency noise, but still contains possible artifact signals.

[0130] The cutting module is configured to cut the third target signal through a preset time window in a sliding window manner to obtain a plurality of fourth target signals.

[0131] Specifically, the signal is segmented, a time window of 2 seconds is set, and a sliding window manner is adopted. After sliding window cutting, different electroencephalogram signal segments S i are obtained. i S band represents the signal data of the i-th time segment.

[0132] The generating module is configured to generate a spectrum graph based on the fourth target signal and calculate a power spectral density based on the spectrum graph. Specifically, the signal data after segmentation is made into a data spectrum graph and the power spectral density after segmentation is calculated. wherein PSD i represents the power spectral density calculated in the specified frequency band. i S noise (f) is the Fourier transform result of the environmental noise in the signal.

[0133] The screening module is configured to determine the fourth target signal as an artifact signal and screen it out when the power spectral density is greater than a first threshold.

[0134] The first determining module is configured to determine the fourth target signal as a second target signal when the power spectral density is less than or equal to the first threshold.

[0135] Specifically, the first artifact signal screening is performed, and the obvious non-physiological artifact signals in the spectrum graph are screened out.

[0136] P i = w1·PSD band (S i ) + w2·EMG feature (S i ) + w3·EOG feature (S i ).

[0137] wherein P iThe comprehensive artifact judgment criterion of the i-th signal segment is a power spectral density, w1, w2, and w3 represent the importance weights of each type of feature, EMG feature (S i The electromyographic activity frequency of the i-th signal segment is used to judge whether the current frequency band is disturbed by electromyographic artifacts, EOG feature (S i The blink frequency of the i-th signal segment is used to judge whether the current frequency band is disturbed by ocular artifacts.

[0138] If P th is the first threshold value, when P i >P th , then mark S i as a non-physiological artifact signal for screening out, when P i ≤P th , otherwise, mark as an effective physiological signal.

[0139] As can be seen, the first target signal is effectively preprocessed and artifact signals are screened out to obtain the second target signal. The method ensures that the final obtained second target signal has higher accuracy and reliability through band-pass filtering, sliding window segmentation, spectral analysis and other technical means, and provides cleaner data input for subsequent electroencephalogram signal analysis.

[0140] As a possible implementation manner, the classification unit comprises: an intensity calculation module, a score calculation module, a threshold determination module, and an updating module.

[0141] The intensity calculation module is configured to obtain a reference signal, the reference signal comprising a reference electrooculogram signal, a reference electrocardiogram signal, and a reference electromyogram signal, calculate an identification intensity based on the reference signal, and obtain a target identification intensity.

[0142] Specifically, the reference signal is obtained, including the reference electrooculogram signal, the reference electrocardiogram signal, and the reference electromyogram signal. These reference signals are usually physiological signals of the participant under normal conditions, and are used for comparison with the target signal. The identification intensity can represent the degree of obviousness of the artifact signal, which is calculated by comparing the difference between the target signal and the reference signal.

[0143] The electroencephalogram signal obtained after preprocessing and removing non-physiological artifact signals is input into the trained artifact identification model, and the historical data of the participant is continuously input to calculate the identification intensity STR of the artifact:

[0144]

[0145] Wherein, STR is the dynamically calculated artifact signal intensity, N is the sample quantity of the participant, STR 0,i represents the initial estimate value of the i-th artifact signal intensity, F iF is the specific feature parameter of the i th artifact signal, F is the average feature of the participant historical data, a is an adjustment factor for controlling the influence of feature deviation, and i is a variable index. mean F is the average feature of the participant historical data, a is an adjustment factor for controlling the influence of feature deviation, and i is a variable index.

[0146] The score calculation module is configured to calculate a classification score based on the reference signal by using a strong classifier to obtain a first classification score, wherein the strong classifier is obtained by combining a KNN classifier and a logistic regression classifier.

[0147] Specifically, an integrated learning strategy is used to combine the KNN classifier and the logistic regression classifier into a strong classifier, and the output of each classifier is defined as C j , and the comprehensive classification score C final is:

[0148]

[0149] wherein C final represents the final classification result after the output of all base classifiers, M is the number of base classifiers, C j is the output result of the judgment of the current condition by the j th base classifier, j is a variable index, and STR max represents the maximum intensity in all artifact signals.

[0150] The threshold determination module is configured to determine a target discrimination threshold based on the first classification score based on a first preset formula.

[0151] T = T base + β · (C final - C mean ).

[0152] wherein T is the target discrimination threshold, T base is a preset classification intensity threshold, β is an adjustment coefficient of the first classification score on the target discrimination threshold, C mean is the mean value of the target discrimination threshold corresponding to the reference signal, and C final is the target discrimination threshold corresponding to the reference signal.

[0153] Specifically, this formula automatically adjusts the target discrimination threshold according to the change of the classification score, ensuring that the classification process is more flexible and accurate.

[0154] The update module is configured to update the intensity of the second target signal based on the target discrimination threshold to obtain a fifth target signal, and to classify the artifact signal in the second target signal based on the fifth target signal and the target discrimination threshold.

[0155] Specifically, the second target signal is intensity updated based on the target discrimination threshold to obtain a fifth target signal. By adjusting the intensity of the signal, the distinction between the artifact and the effective electroencephalogram signal is enhanced. According to the updated signal intensity and the target discrimination threshold, the fifth target signal is classified to identify the artifact signal. Through this process, the artifact signal in the second target signal can be distinguished from the effective signal.

[0156] As a possible implementation manner, the updating module comprises an influence degree submodule, an intensity updating submodule, a second determination submodule, a third determination submodule, a fourth determination submodule and a determination submodule.

[0157] The influence degree submodule is configured to determine, based on the intensity of the second target signal, an influence degree of the artifact signal on the target electroencephalogram signal to obtain a second influence degree.

[0158] Specifically, the intensity of the artifact signal and the influence on the electroencephalogram signal are integrated to calculate the artifact influence degree I:

[0159] I = ξ · STR total + γ · J cov ;

[0160] Wherein, I is the influence degree of the artifact signal on the electroencephalogram signal, ξ is the weight coefficient of STR total in the influence degree calculation, STR total is the intensity of the artifact signal, representing the integrated effect of the sum of the intensities of all artifact signals; γ is the weight coefficient of J cov influencing the influence degree, J cov represents the covering influence degree of the artifact signal on the electroencephalogram signal, and represents the interference caused by the artifact.

[0161] The intensity updating submodule is configured to intensity update the second target signal based on the first influence degree and the target discrimination threshold to obtain a fifth target signal:

[0162]

[0163] Wherein, STR' is the intensity of the fifth target signal, I is the first influence degree, δ is the adjustment coefficient of the intensity of the second target signal and the estimated intensity of the artifact signal on the intensity of the fifth target signal, STR total is the intensity of the second target signal, and STR new is the estimated intensity of the artifact signal.

[0164] Specifically, according to the influence degree and the target discrimination threshold, the signal intensity is updated to provide refined intensity data for artifact signal classification.

[0165] The second determining sub-module is configured to determine that the fifth target signal is an electro-oculogram signal when the intensity of the fifth target signal is greater than or equal to a second threshold value.

[0166] The third determining sub-module is configured to determine that the fifth target signal is an electrocardiogram signal when the intensity of the fifth target signal is greater than or equal to a third threshold value and less than the second threshold value.

[0167] The fourth determining sub-module is configured to determine that the fifth target signal is an electromyogram signal when the intensity of the fifth target signal is greater than or equal to a fourth threshold value and less than the third threshold value.

[0168] Specifically, the updated artifact intensity value is used to determine the category of the current artifact signal in combination with setting a discrimination threshold according to the comprehensive classification score. By setting multiple threshold values, the signal is classified according to different intensity threshold values, and it is determined whether the fifth target signal belongs to an electro-oculogram signal, an electrocardiogram signal or an electromyogram signal, so as to classify and identify the artifact signal.

[0169] The determining sub-module is configured to determine the electro-oculogram signal, the electrocardiogram signal and the electromyogram signal as artifact signals.

[0170] Specifically, the electro-oculogram signal, the electrocardiogram signal and the electromyogram signal are finally identified as artifact signals, and these artifact signals are removed in the electroencephalogram signal processing.

[0171] The embodiment updates the signal intensity in detail by comprehensively considering the intensity of the artifact signal and the influence on the electroencephalogram signal, and classifies the signal according to different intensity threshold values. Through these steps, the artifact signal can be effectively identified and marked, and a clear basis is provided for subsequent artifact removal processing.

[0172] As a possible implementation manner, the first calculation unit comprises an artifact influence module.

[0173] The artifact influence module is configured to input the artifact signal into an artifact influence model to obtain a first influence degree Q:

[0174]

[0175] ψ(W i ,No i )=μ1·(W i -μ2·No i );

[0176] wherein Q is the first influence degree of the artifact signal, n is the category of the artifact signal, I i is the second influence degree corresponding to the i artifact signal, ψ(W i ,No i ) is the second influence degree of the i artifact signal W i and the artifact signal W iNoise component No introduced at the time of acquisition i The relationship between μ1 and μ2 is a preset coefficient.

[0177] Specifically, a artifact influence model is constructed, the influence degree of different artifact types is quantitatively analyzed, and the application order of PCA (principal component analysis) and LLE (local linear embedding) algorithms is determined according to the influence degree, so as to remove the artifact signal and obtain the target electroencephalogram signal.

[0178] As a possible implementation manner, the second calculation unit comprises a first removal module and a second removal module.

[0179] The first removal module is configured to, when the first influence degree is greater than a fifth threshold value, sequentially remove the artifact signal in the second target signal by using the PCA algorithm and the LLE algorithm to obtain the target electroencephalogram signal.

[0180] Specifically, when the influence degree of the artifact signal is large, the PCA is used to remove the artifact signal preferentially, so as to effectively reduce the global noise, reduce the data dimension, clear the principal component, and enable the LLE to work on the clear structure and reflect the main features of the signal.

[0181] The second removal module is configured to, when the first influence degree is less than or equal to the fifth threshold value, sequentially remove the artifact signal in the second target signal by using the LLE algorithm and the PCA algorithm to obtain the target electroencephalogram signal.

[0182] Specifically, when the influence degree of the artifact signal is small and the local signal structure is complex, the LLE is selected to be preferentially processed, which is suitable for processing signals containing complex local structures, maintains the local geometric characteristics of the data, and can better extract effective information from the complex artifact, but can retain the global artifact, and then the PCA is used to extract the main feature information and remove the residual noise.

[0183] The artifact removal device for the electroencephalogram signal comprises a processor and a memory, the first acquisition unit, the classification unit, the first calculation unit and the second calculation unit are stored in the memory as program units, and the corresponding functions are realized by executing the program units stored in the memory by the processor. The modules are located in the same processor; or the modules are located in different processors in any combination.

[0184] The processor comprises a core, and the core retrieves the corresponding program unit from the memory. One or more cores can be set by adjusting the core parameters to improve the efficiency of communication.

[0185] The memory can include non-persistent memory in a computer readable medium, random access memory (RAM), and / or non-volatile memory such as read-only memory (ROM) or flash memory, including at least one memory chip.

[0186] The embodiment of the present application provides a computer readable storage medium, the computer readable storage medium comprises a stored program, wherein the program controls a device where the computer readable storage medium is located to perform the artifact removing method of the electroencephalogram signal when the program is running.

[0187] The embodiment of the present application provides an electroencephalogram monitoring system, comprising one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs comprise a program code for executing the artifact removing method of the electroencephalogram signal.

[0188] Obviously, those skilled in the art should understand that the modules or steps of the present application described above can be realized by general computing devices, which can be concentrated on a single computing device or distributed on a network composed of multiple computing devices, and they can be realized by program codes executable by the computing devices, so that they can be stored in storage devices and executed by the computing devices, and in some cases, the steps shown or described can be executed in different sequences, or they can be manufactured into individual integrated circuit modules respectively, or multiple modules or steps among them can be manufactured into a single integrated circuit module. Therefore, the present application is not limited to any particular combination of hardware and software.

[0189] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt a computer program product in the form of one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program codes.

[0190] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks in the flowcharts and / or combination thereof. Figure 1 one or more flowcharts and / or blocks in the flowcharts and / or combination thereof.

[0191] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks in the flowcharts and / or combination thereof. Figure 1 one or more flowcharts and / or blocks in the flowcharts and / or combination thereof.

[0192] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks in the flowcharts and / or combination thereof. Figure 1 one or more flowcharts and / or blocks in the flowcharts and / or combination thereof.

[0193] In one typical configuration, the computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0194] The memory can include non-persistent memory and / or volatile memory, such as a random access memory (RAM) including a cache area for the temporary storage of data. The memory can also include non-volatile memory, such as read only memory (ROM) for storing structural information and / or instruction code to boot an operating system. The memory can also include solid state non-volatile memory (e.g., flash memory), disk drives, disk arrays, optical storage devices, tape storage devices, etc.

[0195] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible to a computing device. According to the definition herein, computer-readable media does not include transitory media such as modulated data signals and carriers.

[0196] It should also be noted that the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that processes, methods, articles or devices including a series of elements not only include those elements, but also include other elements not explicitly listed or inherent to such processes, methods, articles or devices. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article or device including the element.

[0197] From the above description, it can be seen that the above-mentioned embodiments of the present application achieve the following technical effects:

[0198] 1) The artifact removal method of electroencephalogram signal of the present application records the electroencephalogram signal of the participant using the electroencephalogram acquisition device and performs pretreatment, uses the artifact identification model to mark the artifact, constructs the artifact influence model to quantitatively analyze the influence degree of different artifact types, and removes the electroencephalogram signal artifact through the comprehensive PCA-LLE analysis method. The present application aims to solve the technical problem that the electroencephalogram signal is easily disturbed by electrooculogram, electrocardiogram, electromyogram artifact and environmental noise during the collection process. Through the construction of the artifact identification model and the artifact influence model, the quantitative analysis and effective removal of the artifact signal are realized, and the reliability and analysis accuracy of the electroencephalogram signal data are improved. The method adopts the comprehensive PCA-LLE analysis method, which not only reduces the data dimension, but also retains the local geometric structure of the signal, so as to improve the artifact removal effect, improve the quality of the electroencephalogram signal, and solve the problems of inaccurate artifact recognition of the electroencephalogram signal in the prior art, reduced integrity of the electroencephalogram signal caused by artifact processing, and limited processing capacity of complex artifacts.

[0199] 2) The artifact removal device for electroencephalogram signals of the present application, the device comprises: a first acquisition unit, a classification unit, a first calculation unit and a second calculation unit. The first acquisition unit is used to collect signals of a target object through an electroencephalogram acquisition device to obtain a first target signal, pre-process the first target signal to obtain a second target signal, and the pre-processing at least includes filtering, cropping and removing non-physiological artifact signals in the first target signal; the classification unit is used to input the second target signal into an artifact recognition model to classify the second target signal to obtain artifact signals in the second target signal; the first calculation unit is used to quantitatively analyze the artifact signals to determine a first influence degree of each artifact signal on the electroencephalogram signals; and the second calculation unit is used to remove the artifact signals in the second target signal based on the first influence degree through a PCA-LLE analysis method to obtain a target electroencephalogram signal. The device uses the electroencephalogram acquisition device to record the electroencephalogram signals of the participants and pre-processes them, uses the artifact recognition model to mark artifacts, constructs an artifact influence model to quantitatively analyze the influence degree of different artifact types, and removes artifacts from the electroencephalogram signals through the comprehensive PCA-LLE analysis method. The problems of inaccurate artifact recognition of the electroencephalogram signals in the prior art, reduced integrity of the electroencephalogram signals caused by artifact processing, and limited processing capacity of complex artifacts are solved.

[0200] The above is only the preferred embodiment of the present application and is not used to limit the present application. The present application can have various changes and variations for those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for artifact removal from electroencephalogram (EEG) signals, characterized in that, include: Signals are collected from the target object using an electroencephalogram (EEG) acquisition device to obtain a first target signal. The first target signal is then preprocessed to obtain a second target signal. The preprocessing includes at least filtering, cropping, and removing non-physiological artifact signals from the first target signal. The second target signal is input into the artifact recognition model to classify the second target signal and obtain the artifact signal in the second target signal; Quantitative analysis of the artifact signals is performed to determine the degree of primary influence of each artifact signal on the electroencephalogram (EEG) signal. Based on the first degree of influence, the artifact signals in the second target signal are removed by PCA-LLE analysis to obtain the target EEG signal; The electroencephalogram (EEG) acquisition device includes multiple electrode pads. Signals are acquired from the target object using the EEG acquisition device to obtain a first target signal, including: Signals are acquired by the electrode pads installed at preset positions, and multiple signals are obtained and integrated to obtain the first target signal. The first target signal includes electroencephalogram (EEG) signals, electrooculogram (EOG) signals, electrocardiogram (ECG) signals, electromyogram (EMG) signals, and artifact signals. The EEG signals include blink frequency and eye movement trajectory. The ECG signals include ECG data and heart rate variability data. The EMG signals include EMG data, EMG activity frequency, and EMG activity amplitude. The artifact signals include environmental noise data, device interference data, and background variables. The second target signal is input into the artifact detection model to classify the second target signal and obtain the artifact signal in the second target signal, including: A reference signal is acquired, which includes a reference electrooculogram (EOG) signal, a reference electrocardiogram (ECG) signal, and a reference electromyogram (EMG) signal. The recognition intensity is calculated based on the reference signal to obtain the target recognition intensity. Based on the benchmark signal, a classification score is calculated using a strong classifier to obtain a first classification score. The strong classifier is based on a combination of a KNN classifier and a logistic regression classifier. Based on the first classification score and a first preset formula, the target discrimination threshold is determined as follows: T=T base +β·(C final -C mean ); Where T is the target discrimination threshold, T base Let β be the preset classification strength threshold, and let C be the adjustment coefficient of the first classification score on the target discrimination threshold. mean C is the mean of the reference signal corresponding to the target discrimination threshold. final The target discrimination threshold corresponding to the reference signal; The intensity of the second target signal is updated based on the target discrimination threshold to obtain the fifth target signal. The fifth target signal and the target discrimination threshold are then classified to obtain the artifact signal in the second target signal. The intensity of the second target signal is updated based on the target discrimination threshold to obtain a fifth target signal. Classification is then performed based on the fifth target signal and the target discrimination threshold to obtain the artifact signal within the second target signal, including: Based on the intensity of the second target signal, the degree of influence of the artifact signal on the target EEG signal is determined, and the second degree of influence is obtained; The intensity of the second target signal is updated based on the first degree of influence and the target discrimination threshold to obtain the fifth target signal: Wherein, STR′ is the intensity of the fifth target signal, I is the first degree of influence, and δ is the adjustment coefficient of the intensity of the second target signal and the estimated intensity of the artifact signal on the intensity of the fifth target signal. total STR represents the strength of the second target signal. new The estimated strength of the artifact signal; If the intensity of the fifth target signal is greater than or equal to the second threshold, the fifth target signal is determined to be the electrooculogram signal. If the intensity of the fifth target signal is greater than or equal to the third threshold and less than the second threshold, the fifth target signal is determined to be an electrocardiogram signal. If the intensity of the fifth target signal is greater than or equal to the fourth threshold and less than the third threshold, the fifth target signal is determined to be an electromyographic signal. The electrooculogram (EOG), electrocardiogram (ECG), and electromyogram (EMG) signals are identified as artifact signals.

2. The method according to claim 1, characterized in that, The first target signal is preprocessed to obtain the second target signal, including: The first target signal is bandpass filtered to remove power frequency interference and low frequency noise, resulting in the third target signal. The third target signal is divided into multiple fourth target signals by using a sliding window method through a preset time window; A spectrum diagram is generated based on the fourth target signal, and the power spectral density is calculated based on the spectrum diagram; If the power spectral density is greater than the first threshold, the fourth target signal is identified as the artifact signal and is filtered out. If the power spectral density is less than or equal to the first threshold, the fourth target signal is determined to be the second target signal.

3. The method according to claim 1, characterized in that, Quantitative analysis of the artifact signals is performed to determine the degree of primary influence of each artifact signal on the EEG signal, including: The artifact signal is input into the artifact influence model to obtain the first influence level: ψ(W i ,No i )=μ1·(W i -μ w ·No i ); Where Q represents the first degree of influence of the artifact signal, n represents the type of artifact signal, and I i The degree of influence corresponding to the artifact signal described in the i-th instance is ψ(W). i No i ) represents the i-th artifact signal W i With the artifact signal W i Noise components introduced during data acquisition (No) i The relationship between them, μ1 and μ2 are preset coefficients.

4. The method according to claim 3, characterized in that, Based on the first degree of influence, the artifact signals in the second target signal are removed using PCA-LLE analysis to obtain the target EEG signal, including: If the first level of influence is greater than the fifth threshold, the artifact signals in the second target signal are removed sequentially by PCA algorithm and LLE algorithm to obtain the target EEG signal; If the first level of influence is less than or equal to the fifth threshold, the artifact signals in the second target signal are removed sequentially using the LLE algorithm and the PCA algorithm to obtain the target EEG signal.

5. A device for removing artifacts from electroencephalogram (EEG) signals, characterized in that, The device includes: The first acquisition unit is used to acquire signals from the target object through an electroencephalogram (EEG) acquisition device to obtain a first target signal, and to preprocess the first target signal to obtain a second target signal. The preprocessing includes at least filtering, cropping, and removing non-physiological artifact signals from the first target signal. A classification unit is used to input the second target signal into the artifact recognition model to classify the second target signal and obtain the artifact signal in the second target signal; The first computing unit is used to perform quantitative analysis on the artifact signals to determine the degree of influence of each artifact signal on the electroencephalogram signal. The second calculation unit is used to remove the artifact signals in the second target signal based on the first degree of influence using the PCA-LLE analysis method to obtain the target EEG signal; The first acquisition unit includes: The acquisition module is used to acquire signals through electrode pads installed at preset positions, obtain multiple signals, and integrate them to obtain the first target signal. The first target signal includes electroencephalogram (EEG) signals, electrooculogram (EOG) signals, electrocardiogram (ECG) signals, electromyogram (EMG) signals, and artifact signals. The EEG signals include blink frequency and eye movement trajectory. The ECG signals include ECG data and heart rate variability data. The EMG signals include EMG data, EMG activity frequency, and EMG activity amplitude. The artifact signals include environmental noise data, device interference data, and background variables. The classification unit includes: The intensity calculation module is used to acquire a reference signal, which includes a reference electrooculogram (EOG) signal, a reference electrocardiogram (ECG) signal, and a reference electromyogram (EMG) signal. Based on the reference signal, the recognition intensity is calculated to obtain the target recognition intensity. The score calculation module is used to calculate a classification score based on the reference signal using a strong classifier to obtain a first classification score. The strong classifier is based on a combination of a KNN classifier and a logistic regression classifier. The threshold determination module is used to determine the target discrimination threshold based on the first classification score and a first preset formula. T=T base +β·(C final -C mean ); Where T is the target discrimination threshold, T base Let β be the preset classification strength threshold, and let C be the adjustment coefficient of the first classification score on the target discrimination threshold. mean C is the mean of the reference signal corresponding to the target discrimination threshold. final The target discrimination threshold corresponding to the reference signal; The update module is used to update the intensity of the second target signal based on the target discrimination threshold to obtain a fifth target signal, and to classify the artifact signal in the second target signal based on the fifth target signal and the target discrimination threshold. The update module includes: The influence degree submodule is used to determine the degree of influence of the artifact signal on the target EEG signal based on the intensity of the second target signal, and obtain the second influence degree; The intensity update submodule is used to update the intensity of the second target signal based on the first influence level and the target discrimination threshold to obtain the fifth target signal. Wherein, STR′ is the intensity of the fifth target signal, I is the first degree of influence, and δ is the adjustment coefficient of the intensity of the second target signal and the estimated intensity of the artifact signal on the intensity of the fifth target signal. total STR represents the strength of the second target signal. new The estimated strength of the artifact signal; The second determining submodule is used to determine the fifth target signal as the electrooculogram signal when the intensity of the fifth target signal is greater than or equal to the second threshold. The third determining submodule is used to determine that the fifth target signal is an electrocardiogram signal when the intensity of the fifth target signal is greater than or equal to the third threshold and less than the second threshold. The fourth determining submodule is used to determine that the fifth target signal is an electromyographic signal when the intensity of the fifth target signal is greater than or equal to the fourth threshold and less than the third threshold. A determination submodule is used to identify the electrooculogram (EOG) signal, the electrocardiogram (ECG) signal, and the electromyogram (EMG) signal as the artifact signal.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the method according to any one of claims 1 to 4.

7. A brainwave monitoring system, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs comprising methods for performing any one of claims 1 to 4.

Images