Electroencephalogram signal data processing method and device
By using independent component analysis and artifact recognition algorithms in EEG signal processing, combining confusion matrix and amplitude threshold to identify and remove artifact signals, the problem of difficulty in accurately removing artifacts in the prior art is solved, and the signal is improved accuracy and reliability.
Patent Information
- Application Number
- CN202411000183.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-24
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-07-24
AI Technical Summary
The prior art is difficult to accurately remove artifacts in EEG signals, resulting in data distortion and energy drift problems, affecting the reliability of EEG signals research and clinical interpretation.
By obtaining multi-channel physiological signal data in the preset time window for independent component analysis, divide the time window, combine the confusion matrix and amplitude threshold to identify global and local artifact signals, use the Fast ICA method to solve the confusion matrix and independent signals, and use the Fast ICA method to solve the confusion matrix and independent signals, and according to The ECG channel signal or Q wave recognition algorithm removes artifact signals.
It effectively removes the artifacts in the EEG signal, improves the accuracy and reliability of the signal, avoids interference with normal EEG components, and enhances the accuracy of EEG signal processing.
Smart Images

Figure CN118924309A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of data processing technology, and in particular to a method and device for processing electroencephalogram (EEG) signal data. Background Art
[0002] Electroencephalogram (EEG) is a non-invasive brain activity observation technology. It has the advantages of little harm to patients and easy operation. It is widely used in the diagnosis of neurological diseases, fatigue and depression detection, and research on brain functional areas. However, the amplitude of the EEG signal generated by brain activity is weak, often in the order of hundreds of microvolts. The collected signal is easily interfered by other physiological signals such as electrocardiogram, electrooculogram, electromyography, and the sliding and poor contact of the acquisition equipment itself, showing difficult-to-distinguish artifact (also known as artifact, artifact, etc.) characteristics. High-precision artifact removal is crucial for the study and clinical interpretation of EEG signals.
[0003] At present, the mainstream EEG artifact removal solutions include BSS (Blind Source Separation) algorithm and deep learning-based denoising methods. Among them, most traditional BSS algorithms rely on ICA (Independent Component Analysis) technology, using manual or semi-manual methods to select artifact components; most automatic BSS algorithms are only for eye movement signals, and algorithms are designed to identify high-amplitude baseline drifts; deep learning-based denoising methods mainly face two major problems: data authenticity is not guaranteed and training set data is difficult to construct, which may lead to data distortion and energy drift after artifact removal. At the same time, when reconstructing EEG signals through inverse ICA, traditional methods usually set the identified artifact components to 0 as a whole, but this may interfere with and suppress normal EEG components, affecting the reliability of the final processing results.
[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention
[0005] The embodiments of the present application provide a method and device for processing EEG signal data, so as to at least solve the technical problem in the related art that it is difficult to accurately remove artifacts in EEG signals.
[0006] According to one aspect of an embodiment of the present application, a method for processing EEG signal data is provided, comprising: obtaining multi-channel physiological signal observation data within a preset time window, performing independent component analysis on the physiological signal observation data, and obtaining a confusion matrix corresponding to the physiological signal observation data and multi-dimensional independent signals; determining an artifact signal from the multi-dimensional independent signals, comprising: dividing the preset time window into multiple sub-time windows, restoring the physiological signal of each channel in each sub-time window according to the confusion matrix and the independent signal of each dimension, determining whether the independent signal of each dimension is a global artifact signal according to the amplitude of each physiological signal and a preset global amplitude threshold, or determining a local artifact signal in the independent signal of each dimension according to the amplitude of each physiological signal and a preset local amplitude threshold; and / or determining whether the independent signal of each dimension is an electrooculographic artifact signal according to a target row element corresponding to the independent signal of each dimension in the confusion matrix; and / or determining the electrocardiogram artifact signal in the independent signal of each dimension according to a preset electrocardiogram channel signal or a Q-wave recognition algorithm; removing the artifact signal in the multi-dimensional independent signal, and determining the target EEG signal according to the confusion matrix and the independent signal after the artifact signal is removed.
[0007] Optionally, multi-channel physiological signal observation data within a preset time window is obtained, and independent component analysis is performed on the physiological signal observation data to obtain a confusion matrix corresponding to the physiological signal observation data and multi-dimensional independent signals, including: obtaining multi-channel physiological signal observation data x(t) within a preset time window, denoted as: x(t) =
[0008] [x1(t),…,x p (t),x p+1 (t),…,x p+q (t)] T ,t=1,…,T, where x i (t) represents the physiological signal observation data of the i-th dimension at time t, i = 1, ..., p + q, p represents the number of dimensions of the EEG signal in the physiological signal observation data, q represents the number of dimensions of the non-EEG signal in the physiological signal observation data, T represents the length of the preset time window, and T is greater than the preset time length; the signal data relationship is determined as: x(t) = As(t), where,
[0009] s(t)=[s1(t),…,s p (t),s p+1 (t),…,s p+q (t)] T , represents the multi-dimensional independent signal to be calculated, A represents the confusion matrix to be calculated; the fast independent component analysis Fast ICA method is used to solve the confusion matrix and the multi-dimensional independent signal.
[0010] Optionally, whether the independent signal of each dimension is a global artifact signal is determined based on the amplitude of each physiological signal and a preset global amplitude threshold, including: for the independent signal of each dimension, determining the first maximum amplitude in the amplitude of the physiological signal of each channel restored based on the independent signal within each sub-time window; if the maximum amplitude is greater than the preset global amplitude threshold, determining that the independent signal is a global artifact signal.
[0011] Optionally, the local artifact signal in the independent signal of each dimension is determined based on the amplitude of each physiological signal and a preset local amplitude threshold, including: determining the second maximum amplitude of the physiological signal observation data in each channel and each sub-time window; for each sub-time window, determining the target quantile of the second maximum amplitude of multiple channels corresponding to the sub-time window, and taking the maximum value of the target quantile and the preset local amplitude threshold as the dynamic threshold corresponding to the sub-time window; for the independent signal of each dimension, determining the third maximum amplitude corresponding to each sub-time window in the amplitude of the physiological signal of each channel in each sub-time window restored based on the independent signal, for each sub-time window, if the third maximum amplitude is greater than the dynamic threshold corresponding to the sub-time window, determine that the local signal of the independent signal in the sub-time window is a local artifact signal.
[0012] Optionally, determining whether the independent signal of each dimension is an electrooculogram artifact signal is based on the target row elements corresponding to the independent signal of each dimension in the confusion matrix, including: for the independent signal of each dimension, determining the column vector corresponding to the independent signal in the confusion matrix, and determining the first sum of the target row elements of the first preset number of rows in the column vector, determining the second sum of all the row elements in the column vector, and determining the ratio of the first sum to the second sum; sorting the ratios corresponding to the independent signals of each dimension, and determining that the independent signals corresponding to the second preset number of ratios in the front order are electrooculogram artifact signals.
[0013] Optionally, the ECG artifact signal in the independent signal of each dimension is determined according to the preset ECG channel signal or Q wave recognition algorithm, including: for the independent signal of each dimension, determining the correlation coefficient between the independent signal and the preset ECG channel signal, if the correlation coefficient is greater than the preset correlation coefficient threshold, determining the independent signal as an ECG artifact signal, wherein the type of the correlation coefficient includes at least one of the following: Pearson correlation coefficient, nonlinear correlation coefficient; or, for the independent signal of each dimension, using the Q wave recognition algorithm to determine the Q wave occurrence time in the independent signal; for every two adjacent Q wave occurrence times, determining the sampling number of the physiological signal observation data in each sub-time window and the time difference between the adjacent Q wave occurrence times, determining the ratio of the sampling number to the time difference as a group of real-time heart rate data corresponding to the previous Q wave occurrence time in the adjacent Q wave occurrence times; for each sub-time window, determining the ratio of the standard deviation of multiple groups of real-time heart rate data corresponding to multiple Q wave occurrence times in the sub-time window to the average value as the coefficient of variation corresponding to the sub-time window, if the coefficient of variation is less than the preset coefficient of variation threshold, determining the local signal of the independent signal in the sub-time window as an ECG artifact signal.
[0014] Optionally, removing artifact signals from multi-dimensional independent signals, including: for the independent signal of each dimension, if the independent signal is determined to be a global artifact signal, an electrooculogram artifact signal, or an electrocardiogram artifact signal, assigning the independent signal a value of 0; if a local signal corresponding to at least one sub-time window in the independent signal is determined to be a local artifact signal or an electrocardiogram artifact signal, using a preset window function to smooth the local signal in the independent signal, wherein the type of the window function includes at least one of the following: a rectangular window function, a trapezoidal window function, a sinusoidal window function, and a maximum-normalized kernel function.
[0015] According to another aspect of the embodiment of the present application, there is also provided an EEG signal data processing device, including: an acquisition module, used to acquire multi-channel physiological signal observation data within a preset time window, perform independent component analysis on the physiological signal observation data, and obtain a confusion matrix corresponding to the physiological signal observation data and multi-dimensional independent signals; a first determination module, used to determine the artifact signal from the multi-dimensional independent signals, including: dividing the preset time window into multiple sub-time windows, restoring the physiological signal of each channel in each sub-time window according to the confusion matrix and the independent signal of each dimension, and determining the artifact signal according to the amplitude of each physiological signal and the preset global amplitude; An amplitude threshold is used to determine whether the independent signal of each dimension is a global artifact signal, or, the local artifact signal in the independent signal of each dimension is determined according to the amplitude of each physiological signal and a preset local amplitude threshold; and / or, whether the independent signal of each dimension is an electrooculogram artifact signal is determined according to the target row element corresponding to the independent signal of each dimension in the confusion matrix; and / or, the ECG artifact signal in the independent signal of each dimension is determined according to a preset ECG channel signal or a Q-wave recognition algorithm; a second determination module is used to remove artifact signals in multi-dimensional independent signals, and determine the target EEG signal according to the confusion matrix and the independent signal after the artifact signal is removed.
[0016] According to another aspect of an embodiment of the present application, a computer program product is further provided, the computer program product comprising: a computer program, wherein when the computer program is executed by a processor, the above-mentioned EEG signal data processing method is implemented.
[0017] According to another aspect of an embodiment of the present application, an electronic device is further provided, which includes: a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to execute the above-mentioned EEG signal data processing method through the computer program.
[0018] In an embodiment of the present application, large window width data is used to optimize the solution of ICA, and small window width data is used to locally remove artifacts, wherein the large window width ensures the accuracy of the ICA solution and can characterize the global noise at the same time, and the small window width ensures the accuracy of artifact removal and can avoid excessive impact on the EEG signal; further, taking into account the spatiotemporal characteristics and frequency domain information of the physiological signal observation data, the present application classifies, identifies and removes artifact components such as the electrocardiogram (ECG) signal and the electrooculogram (EOG) signal in the physiological signal observation data based on the signal characteristics of the independent signal itself and the EEG signal reconstructed by inverse ICA transformation of the independent signal, effectively solving the technical problem in the related art that it is difficult to accurately remove artifacts in the EEG signal. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:
[0020] Figure 1 is a flowchart of an optional method for processing EEG signal data according to an embodiment of the present application;
[0021] Figure 2 is a schematic diagram of an optional electroencephalogram signal after removing a global artifact signal according to an embodiment of the present application;
[0022] Figure 3 is a schematic diagram of an optional electroencephalogram signal after removing local artifact signals according to an embodiment of the present application;
[0023] Figure 4 is a schematic structural diagram of an optional electroencephalogram signal data processing device according to an embodiment of the present application;
[0024] Figure 5 It is a schematic diagram of the structure of an optional electronic device according to an embodiment of the present application. DETAILED DESCRIPTION
[0025] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work 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 drawings of the present application are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0027] Example 1
[0028] According to an embodiment of the present application, a method for processing EEG signal data 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 can be executed in an order different from that shown here.
[0029] Figure 1 is a flow chart of a method for processing EEG signal data according to an embodiment of the present application, such as Figure 1 As shown, the method comprises the following steps:
[0030] Step S102, acquiring multi-channel physiological signal observation data within a preset time window, performing independent component analysis on the physiological signal observation data, and obtaining a confusion matrix corresponding to the physiological signal observation data and multi-dimensional independent signals;
[0031] Step S104, determining an artifact signal from the multi-dimensional independent signals, including: dividing a preset time window into a plurality of sub-time windows, restoring the physiological signal of each channel in each sub-time window according to the confusion matrix and the independent signal of each dimension, determining whether the independent signal of each dimension is a global artifact signal according to the amplitude of each physiological signal and a preset global amplitude threshold, or determining a local artifact signal in the independent signal of each dimension according to the amplitude of each physiological signal and a preset local amplitude threshold; and / or, determining whether the independent signal of each dimension is an electrooculographic artifact signal according to the target row element corresponding to the independent signal of each dimension in the confusion matrix; and / or, determining the ECG artifact signal in the independent signal of each dimension according to a preset ECG channel signal or a Q-wave recognition algorithm;
[0032] Step S106, removing artifact signals from the multi-dimensional independent signals, and determining the target EEG signal based on the confusion matrix and the independent signals after removing the artifact signals.
[0033] The following describes each step of the EEG signal data processing method in conjunction with a specific implementation process.
[0034] Optionally, firstly obtain multi-channel physiological signal observation data x(t) within a preset time window, which is recorded as:
[0035] x(t)=[x1(t),…,x p (t),x p+1 (t),…,x p+q (t)] T ,t=1,…,T
[0036] In the formula, x i(t) represents the physiological signal observation data of the i-th dimension at time t, i=1,…,p+q, p represents the dimension number of EEG signals in the physiological signal observation data, q represents the dimension number of non-EEG signals in the physiological signal observation data, q>0, corresponding to possible electrooculogram, electrocardiogram, electromyography and other signals, T represents the length of the preset time window, T is greater than the preset time length, the preset time length is usually large, and can be adjusted according to actual conditions, and is not limited here.
[0037] When performing ICA analysis, the signal data relationship can be determined as:
[0038] x(t)=As(t)
[0039] Where s(t)=[s1(t),…,s p (t),s p+1 (t),…,s p+q (t)] T , represents the multi-dimensional independent signal to be calculated, whose spatial dimension is independent but the autocorrelation of the time dimension is allowed to be retained, and A represents the confusion matrix to be calculated. This application follows the general assumption of ICA, that is, there is no model error, and the dimension of the independent signal is equal to the dimension of the observed data.
[0040] Afterwards, the Fast ICA method can be used to solve the confusion matrix and multi-dimensional independent signals.
[0041] Without loss of generality, the physiological signal observation data x(t) can be transformed into spatially independent data z(t) as follows:
[0042]
[0043] In the formula, represents the sample covariance matrix of x(t). In this case, the covariance matrix of z(t) is I p+q , so the space is uncorrelated, and because s(t) is independent in space, it can be seen that H is an orthogonal matrix. If s(t) contains at most one normal component and H is constrained to be standard orthogonal, the ICA analysis problem can be solved by maximizing the negative entropy, maximizing the degree of non-normality expressed by the fourth-order moment, etc. An optional solution process is as follows:
[0044] First, let s(t) = Wz(t), where W = H -1 , is the inverse confusion matrix.
[0045] Determine the probability density function of s(t) as: f(s(t)), then define the differential entropy of s(t) as:
[0046]
[0047] Furthermore, the negative entropy of s(t) is determined based on differential entropy:
[0048]
[0049] In the formula, s gauss (t) is a normally distributed random variable with the same mean and covariance matrix as s(t). Since the negative entropy is invariant to linear transformations, for any non-zero matrix A, there exists J(x(t)) = J(As(t)) = J(s(t)).
[0050] The correlation between random variables can usually be expressed using cross information. For independent random variables, the cross information can be expressed as:
[0051]
[0052] According to the linear transformation invariance of negative entropy, maximizing variable independence is equivalent to minimizing the amount of cross information, which is equivalent to J(s i (t)) are maximized separately. To achieve this goal, the following approximation can be used:
[0053] J(s i (t))≈c(EG(s i (t)) / )-E(G(v))
[0054] In the formula, c is a preset constant, E(u) represents the expectation of the calculated variable u, G(u) = logcoshu, and ν is a random variable with a standard normal distribution. At this point, the ICA analysis problem is transformed into an optimization problem for the following objective function:
[0055]
[0056] In the formula, w i represents the i-th row vector of the inverse confusion matrix W, δ jk is the Kronecker delta function. Only when j = k, δ jk =1, otherwise, δ jk =0.
[0057] Optionally, the Newton method can be used to iteratively solve the above objective function to obtain the target inverse confusion matrix The specific method is: in each iteration process, w is updated by the following target formula i :
[0058]
[0059] In the formula, represents the updated w iIn order to avoid the p+q Converge to the same point, the Gram-Schmidt orthogonalization method can be used to update each Process them to be orthogonal to each other.
[0060] After this iteration is completed, if With w i If the difference is less than the preset threshold, or the number of iterations has reached the preset number, the iteration is stopped and the current Determine the target inverse confusion matrix Otherwise, continue to the next iteration, where w to be updated in the next iteration is i After the previous iterative orthogonalization process is completed
[0061] Finally, the multi-dimensional independent signals are determined as: Determine the confusion matrix as:
[0062] According to the above estimation process, the sample covariance matrix The accurate estimation of the expectation formula in the Newton method requires a high sample size, that is, the accuracy of ICA estimation is affected by the duration of the estimated signal. When the signal is stable, increasing the signal duration can improve the estimated and In practical applications, for physiological signal observation data collected at 500Hz, the physiological signal observation data with a preset time window B = 20 minutes can be selected for ICA analysis to obtain high-precision ICA estimation parameters. and
[0063] When performing artifact signal analysis, the embodiment of the present application comprehensively considers the spatiotemporal characteristics and frequency domain information of different physiological signals, and classifies, identifies and removes various artifact components in the physiological signal observation data based on the signal characteristics of the independent signal itself and the EEG signal reconstructed using the inverse ICA transform of the independent signal.
[0064] Optionally, when analyzing the global artifact signal, it can be done in the following way: for the independent signal of each dimension, determine the first maximum amplitude in the amplitude of the physiological signal of each channel restored based on the independent signal in each sub-time window; if the maximum amplitude is greater than a preset global amplitude threshold, determine that the independent signal is a global artifact signal.
[0065] Since global artifacts usually come from poor electrode contact, wire friction, sweat traces of the subject, or large-scale movement interference, they and electrooculographic artifacts have the characteristics of low frequency, high amplitude, and intensive appearance within a period of time, so they should be globally eliminated by using their spatiotemporal characteristics. Define the vector of the independent signal of the i-th dimension unfolded along time as because has been normalized so that its amplitude can be identified in the multi-channel physiological signal after it is restored by inverse ICA transformation.
[0066] Specific, sure is the multi-channel physiological signal matrix after the restoration of the i-th independent signal, where is a matrix The total observation time, i.e., the preset time window B, is divided into small window widths b, where the kth observation period, i.e., the sub-time window, is Π k =,(k-1)b+1,kb-, then we can define The interval amplitudes in different channels and different observation periods are:
[0067]
[0068] In the formula, represents the maximum amplitude of the multi-channel physiological signal restored by the i-th independent signal in the j-th electrode channel and the k-th sub-time window, represents the maximum amplitude in the kth sub-time window on multiple channels, Indicates its maximum amplitude within the entire preset time window.
[0069] When identifying global artifact signals, a preset global amplitude threshold τ1 can be determined, The independent signal of is the global noise component, which can be assigned to 0 for denoising. As an example, τ1 = 2000 μV can be taken, which has achieved good results in practice, such as Figure 2 As shown, the gray solid line represents the original signal, and the black solid line represents the signal after artifact removal. It can be seen that the high-amplitude oscillation is eliminated after processing, while the original physiological signal is slightly disturbed.
[0070] Optionally, when analyzing electrooculogram artifact signals, it can be done in the following manner: for the independent signal of each dimension, determine the column vector corresponding to the independent signal in the confusion matrix, and determine the first sum of the target row elements of the first preset number of rows in the column vector, determine the second sum of all row elements in the column vector, and determine the ratio of the first sum to the second sum; sort the ratios corresponding to the independent signals of each dimension, and determine that the independent signals corresponding to the second preset number of ratios in the front order are the electrooculogram artifact signals.
[0071] The recognition of electrooculographic artifact signals also utilizes its spatial characteristics. Since the influence range of electrooculographic signals is mainly in the frontal pole area and the influence gradually weakens towards the back, it can be used The prior knowledge of the distribution of coefficients in space is extracted. For example, suppose the electrode channel corresponding to the frontal region is The first row of the vector and the second line Corresponding to the channels corresponding to the Fp1 and Fp2 electrodes in the popular international 10-20 system, it can be calculated at this time
[0072]
[0073] In the formula, α FP,i express The frontal area coefficient of the EEG signal channel can be measured by selecting α FP,i The first two largest independent signals are used as electrooculographic artifact signals, and can be assigned a value of 0 for denoising. If the physiological signal observation data comes from high-density electrodes, other electrode channels corresponding to the frontal region can be selected accordingly, which will not be described here.
[0074] Optionally, when analyzing ECG artifact signals, it can be performed in the following two ways.
[0075] In the first method, for the independent signal of each dimension, the correlation coefficient between the independent signal and the preset ECG channel signal is determined. If the correlation coefficient is greater than the preset correlation coefficient threshold, the independent signal is determined to be an ECG artifact signal, wherein the type of correlation coefficient includes at least one of the following: Pearson correlation coefficient, nonlinear correlation coefficient.
[0076] Specifically, if the ECG channels in the data are known, the identification of ECG signal artifacts can be achieved by calculating the global correlation coefficient. For example, the i-th independent signal can be calculated With ECG channel [x j (1),…,x j (T)] TThe Pearson correlation coefficient or nonlinear correlation coefficient between them is used to select independent signals with correlation coefficients greater than the preset correlation coefficient threshold τ2 as ECG artifact signals, which can be subsequently assigned an overall value of 0 for denoising. Among them, τ2 can be set based on experience, such as when it is set to 0.7, it shows good results in practice.
[0077] In the second method, for the independent signal of each dimension, the Q wave identification algorithm is used to determine the Q wave occurrence time in the independent signal; for every two adjacent Q wave occurrence times, the sampling number of the physiological signal observation data in each sub-time window and the time difference between the adjacent Q wave occurrence times are determined, and the ratio of the sampling number to the time difference is determined as a group of real-time heart rate data corresponding to the previous Q wave occurrence time in the adjacent Q wave occurrence times; for each sub-time window, the ratio of the standard deviation of multiple groups of real-time heart rate data corresponding to multiple Q wave occurrence times in the sub-time window to the average value is determined as the coefficient of variation corresponding to the sub-time window. If the coefficient of variation is less than the preset coefficient of variation threshold, the local signal of the independent signal in the sub-time window is determined to be an ECG artifact signal.
[0078] Specifically, if the ECG channel in the data is unknown, the corresponding independent signal can be selected by using the periodic rhythm characteristics of the ECG. For example, the most classic Pan-Tompkins algorithm for identifying Q waves in ECG can be considered. For the i-th independent signal The Pan-Tompkins algorithm can be used to extract the corresponding time of the Q wave in the signal, recorded as *t i1 ,…,t iK +, for any adjacent moments, the real-time heart rate in each sub-time window can be calculated based on their time difference:
[0079]
[0080] In the formula, f s represents the sampling rate per second of the physiological signal observation data, and b is the length of the sub-time window of the small window width.
[0081] Based on real-time heart rate data*f i1 ,…,f i(K-1) +, the coefficient of variation in each sub-time window, that is, the ratio of the standard deviation to the mean, can be calculated, and the independent signal of the sub-time window with a coefficient of variation less than a preset coefficient of variation threshold can be selected as the ECG artifact signal and subjected to noise reduction processing. The preset coefficient of variation threshold can be set based on experience, such as 0.2.
[0082] Optionally, when analyzing local artifact signals, it can be done in the following manner: determine the second maximum amplitude of the physiological signal observation data in each channel and each sub-time window; for each sub-time window, determine the target quantile of the second maximum amplitude of multiple channels corresponding to the sub-time window, and use the maximum value of the target quantile and the preset local amplitude threshold as the dynamic threshold corresponding to the sub-time window; for the independent signal of each dimension, determine the third maximum amplitude corresponding to each sub-time window in the amplitude of the physiological signal of each channel in each sub-time window restored based on the independent signal; for each sub-time window, if the third maximum amplitude is greater than the dynamic threshold corresponding to the sub-time window, determine that the local signal of the independent signal in the sub-time window is a local artifact signal.
[0083] Since the above-mentioned artifact selection methods are all aimed at specific electrooculogram, electrocardiogram or large-scale baseline drift situations, more sophisticated local artifact removal algorithms are required for more common local artifacts or electromyography artifacts. The embodiment of the present application takes into account the variability of EEG signals. The use of a simple fixed threshold method may affect intense brain activity, such as incorrectly removing EEG signals during epileptic seizures. For this reason, a dynamic threshold method is proposed to identify artifacts using their local features.
[0084] For example, the maximum amplitude of the i-th independent signal in the k-th time window obtained by the above calculation can be considered as And use dynamic threshold Select the local artifact period. You can define is the original physiological signal observation sequence, and similarly, its maximum amplitude in the kth sub-time window and the jth channel is calculated as:
[0085]
[0086] For each sub-time window, we can consider The upper 90% quantile on multiple channels
[0087] Setting dynamic threshold in time domain By detecting the independent signal segments according to the dynamic threshold, the EEG signal can be retained during intense brain activity, avoiding the impact of direct overall removal on the EEG during active periods. Among them, τ3 is the preset local amplitude threshold, which can be set according to experience. For example, when it is set to 800μV, it shows good results in practice.
[0088] As an optional implementation, when removing artifact signals in multi-dimensional independent signals, it can be done in the following way: for the independent signal of each dimension, if the independent signal is determined to be a global artifact signal, an electrooculographic artifact signal, or an electrocardiographic artifact signal, the independent signal is assigned a value of 0; if the local signal corresponding to at least one sub-time window in the independent signal is determined to be a local artifact signal or an electrocardiographic artifact signal, the local signal in the independent signal is smoothed using a preset window function, wherein the type of the window function includes at least one of the following: a rectangular window function, a trapezoidal window function, a sinusoidal window function, and a maximum-normalized kernel function.
[0089] It is understandable that for global artifact signals, the overall noise reduction can be directly assigned to 0, but for local artifact signals, it is necessary to consider how to maintain the continuity of the restored signal after the artifact signal is removed. To achieve this, a window function in the time domain can be used to suppress the identified local artifact signal and smooth the attenuation intensity of the time domain dimension.
[0090] For example, if the i-th independent signal is determined to be an artifact signal in the k-th sub-time window, the window function can be used to smoothly attenuate the selected independent signal. The specific formula is:
[0091]
[0092] Among them, some commonly used examples of window functions are shown in Table 1. Various kernel functions with maximum value normalization can also be selected as smoothing windows to process independent signals. Figure 3 This is a schematic diagram of an optional EEG signal after removing local artifact signals.
[0093] Table 1
[0094]
[0095] In an embodiment of the present application, large window width data is used to optimize the solution of ICA, and small window width data is used to locally remove artifacts, wherein the large window width ensures the accuracy of the ICA solution and can characterize the global noise at the same time, and the small window width ensures the accuracy of artifact removal and can avoid excessive impact on the EEG signal; further, taking into account the spatiotemporal characteristics and frequency domain information of the physiological signal observation data, the present application classifies, identifies and removes artifact components such as the electrocardiogram (ECG) signal and the electrooculogram (EOG) signal in the physiological signal observation data based on the signal characteristics of the independent signal itself and the EEG signal reconstructed by inverse ICA transformation of the independent signal, effectively solving the technical problem in the related art that it is difficult to accurately remove artifacts in the EEG signal.
[0096] Example 2
[0097] According to an embodiment of the present application, there is also provided an electroencephalogram signal data processing device for implementing the electroencephalogram signal data processing method in embodiment 1, such as Figure 4 As shown, the EEG signal data processing device at least includes: an acquisition module 41, a first determination module 42 and a second determination module 43, wherein:
[0098] An acquisition module 41 is used to acquire multi-channel physiological signal observation data within a preset time window, perform independent component analysis on the physiological signal observation data, and obtain a confusion matrix corresponding to the physiological signal observation data and multi-dimensional independent signals;
[0099] The first determination module 42 is used to determine the artifact signal from the independent signals of multiple dimensions, including: dividing the preset time window into multiple sub-time windows, restoring the physiological signal of each channel in each sub-time window according to the confusion matrix and the independent signal of each dimension, determining whether the independent signal of each dimension is a global artifact signal according to the amplitude of each physiological signal and the preset global amplitude threshold, or determining the local artifact signal in the independent signal of each dimension according to the amplitude of each physiological signal and the preset local amplitude threshold; and / or, determining whether the independent signal of each dimension is an electrooculographic artifact signal according to the target row element corresponding to the independent signal of each dimension in the confusion matrix; and / or, determining the ECG artifact signal in the independent signal of each dimension according to the preset ECG channel signal or Q wave recognition algorithm;
[0100] The second determination module 43 is used to remove artifact signals from the multi-dimensional independent signals, and determine the target EEG signal according to the confusion matrix and the independent signals after the artifact signals are removed.
[0101] The functions of each module of the EEG signal data processing device are described below in conjunction with a specific implementation process.
[0102] Optionally, the acquisition module can acquire multi-channel physiological signal observation data x(t) within a preset time window, which is expressed as: x(t) = [x1(t),…,x p (t),x p+1 (t),…,x p+q (t)] T ,t=1,…,T, where x i (t) represents the physiological signal observation data of the i-th dimension at time t, i = 1,…,p+q, p represents the number of dimensions of the EEG signal in the physiological signal observation data, q represents the number of dimensions of the non-EEG signal in the physiological signal observation data, T represents the length of the preset time window, and T is greater than the preset time length; the signal data relationship is determined as: x(t) = As(t), where s(t) = [s1(t),…,s p (t),s p+1 (t),…,s p+q(t)] T , represents the multi-dimensional independent signal to be calculated, A represents the confusion matrix to be calculated; the fast independent component analysis Fast ICA method is used to solve the confusion matrix and the multi-dimensional independent signal.
[0103] Optionally, when analyzing the global artifact signal, the first determination module can proceed as follows: for the independent signal of each dimension, determine the first maximum amplitude among the amplitudes of the physiological signals of each channel restored based on the independent signal in each sub-time window; if the maximum amplitude is greater than a preset global amplitude threshold, determine that the independent signal is a global artifact signal.
[0104] Optionally, when analyzing the local artifact signal, the first determination module can proceed in the following manner: determine the second maximum amplitude of the physiological signal observation data in each channel and each sub-time window; for each sub-time window, determine the target quantile of the second maximum amplitude of multiple channels corresponding to the sub-time window, and use the maximum value of the target quantile and the preset local amplitude threshold as the dynamic threshold corresponding to the sub-time window; for the independent signal of each dimension, determine the third maximum amplitude corresponding to each sub-time window in the amplitude of the physiological signal of each channel in each sub-time window restored based on the independent signal; for each sub-time window, if the third maximum amplitude is greater than the dynamic threshold corresponding to the sub-time window, determine that the local signal of the independent signal in the sub-time window is a local artifact signal.
[0105] Optionally, when analyzing the electrooculogram artifact signal, the first determination module can proceed as follows: for the independent signal of each dimension, determine the column vector corresponding to the independent signal in the confusion matrix, and determine the first sum of the target row elements of the first preset number of rows in the column vector, determine the second sum of all the row elements in the column vector, and determine the ratio of the first sum to the second sum; sort the ratios corresponding to the independent signals of each dimension, and determine that the independent signals corresponding to the second preset number of ratios in the front ranking are electrooculogram artifact signals.
[0106] Optionally, when analyzing the ECG artifact signal, the first determination module can proceed in the following manner: for the independent signal of each dimension, determine the correlation coefficient between the independent signal and the preset ECG channel signal, if the correlation coefficient is greater than the preset correlation coefficient threshold, determine the independent signal to be an ECG artifact signal, wherein the type of the correlation coefficient includes at least one of the following: Pearson correlation coefficient, nonlinear correlation coefficient; or, for the independent signal of each dimension, use the Q wave recognition algorithm to determine the Q wave occurrence time in the independent signal; for every two adjacent Q wave occurrence times, determine the sampling number of the physiological signal observation data in each sub-time window and the time difference between the adjacent Q wave occurrence times, and determine that the ratio of the sampling number to the time difference is a group of real-time heart rate data corresponding to the previous Q wave occurrence time in the adjacent Q wave occurrence times; for each sub-time window, determine that the ratio of the standard deviation of multiple groups of real-time heart rate data corresponding to multiple Q wave occurrence times in the sub-time window to the average value is the coefficient of variation corresponding to the sub-time window, if the coefficient of variation is less than the preset coefficient of variation threshold, determine that the local signal of the independent signal in the sub-time window is an ECG artifact signal.
[0107] Optionally, when the second determination module removes artifact signals from multi-dimensional independent signals, it can be done in the following manner: for the independent signal of each dimension, if the independent signal is determined to be a global artifact signal, an electrooculographic artifact signal, or an electrocardiographic artifact signal, the independent signal is assigned a value of 0; if the local signal corresponding to at least one sub-time window in the independent signal is determined to be a local artifact signal or an electrocardiographic artifact signal, the local signal in the independent signal is smoothed using a preset window function, wherein the type of the window function includes at least one of the following: a rectangular window function, a trapezoidal window function, a sinusoidal window function, and a maximum-normalized kernel function.
[0108] It should be noted that each module in the EEG signal data processing device in the embodiment of the present application corresponds one by one to each implementation step of the EEG signal data processing method in Example 1. Since a detailed description has been given in Example 1, some details not reflected in this embodiment can be referred to Example 1 and will not be repeated here.
[0109] Example 3
[0110] According to an embodiment of the present application, a computer program product is also provided. The computer program product includes a computer program, wherein when the computer program is executed by a processor, the method for processing EEG signal data in Embodiment 1 is implemented.
[0111] According to an embodiment of the present application, a non-volatile storage medium is also provided, which includes a stored computer program, wherein the device where the non-volatile storage medium is located executes the EEG signal data processing method in Example 1 by running the computer program.
[0112] According to an embodiment of the present application, a processor is also provided, which is used to run a computer program, wherein the EEG signal data processing method in Example 1 is executed when the computer program is running.
[0113] According to an embodiment of the present application, an electronic device is also provided, which includes: a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to execute the EEG signal data processing method in Example 1 through the computer program.
[0114] Specifically, the computer program executes the following steps when it is running: obtaining multi-channel physiological signal observation data within a preset time window, performing independent component analysis on the physiological signal observation data, and obtaining a confusion matrix and multi-dimensional independent signals corresponding to the physiological signal observation data; determining artifact signals from the multi-dimensional independent signals, including: dividing the preset time window into multiple sub-time windows, restoring the physiological signal of each channel in each sub-time window according to the confusion matrix and the independent signal of each dimension, determining whether the independent signal of each dimension is a global artifact signal according to the amplitude of each physiological signal and a preset global amplitude threshold, or determining the local artifact signal in the independent signal of each dimension according to the amplitude of each physiological signal and a preset local amplitude threshold; and / or determining whether the independent signal of each dimension is an electrooculographic artifact signal according to the target row element corresponding to the independent signal of each dimension in the confusion matrix; and / or determining the electrocardiogram artifact signal in the independent signal of each dimension according to a preset electrocardiogram channel signal or a Q-wave recognition algorithm; removing the artifact signals in the multi-dimensional independent signals, and determining the target electroencephalogram signal according to the confusion matrix and the independent signal after the artifact signals are removed.
[0115] As an optional implementation, the electronic device may be in the form of a mobile terminal, a computer terminal or a similar computing device. Figure 5 FIG. 1 shows a hardware structure block diagram of an electronic device for implementing an EEG signal data processing method. Figure 5 As shown, the electronic device 50 may include one or more (502a, 502b, ..., 502n are used to illustrate) processors 502 (the processor 502 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 504 for storing data, and a transmission device 506 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 5 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 5 More or fewer components as shown, or with Figure 5 Different configurations are shown.
[0116] It should be noted that the one or more processors 502 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the electronic device 50. As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).
[0117] The memory 504 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the EEG signal data processing method in the embodiment of the present application. The processor 502 executes various functional applications and data processing by running the software programs and modules stored in the memory 504, that is, realizing the vulnerability detection method of the above-mentioned application. The memory 504 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 504 may further include a memory remotely arranged relative to the processor 502, and these remote memories may be connected to the electronic device 50 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0118] The transmission device 506 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the electronic device 50. In one example, the transmission device 506 includes a network adapter (Network Interface Controller, 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 506 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.
[0119] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the electronic device 50 .
[0120] The serial numbers of the above embodiments are only for description and do not represent the advantages or disadvantages of the embodiments.
[0121] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.
[0122] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.
[0123] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed over multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.
[0124] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0125] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or all or part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of each embodiment method of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc. Various media that can store program codes.
[0126] The above are only preferred implementations of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.
Claims
1. A method for processing electroencephalogram signal data, characterized in that: include: Acquire multi-channel physiological signal observation data within a preset time window, perform independent component analysis on the physiological signal observation data, and obtain a confusion matrix corresponding to the physiological signal observation data and multi-dimensional independent signals; Determining artifact signals from the multi-dimensional independent signals, including: dividing the preset time window into multiple sub-time windows, restoring the physiological signal of each channel in each sub-time window according to the confusion matrix and the independent signal of each dimension, determining whether the independent signal of each dimension is a global artifact signal according to the amplitude of each physiological signal and a preset global amplitude threshold, or determining the local artifact signal in the independent signal of each dimension according to the amplitude of each physiological signal and a preset local amplitude threshold; and / or, determining whether the independent signal of each dimension is an electrooculographic artifact signal according to the target row element corresponding to the independent signal of each dimension in the confusion matrix; and / or, determining the electrocardiographic artifact signal in the independent signal of each dimension according to a preset electrocardiographic channel signal or a Q wave recognition algorithm; Artifact signals are removed from the multi-dimensional independent signals, and a target EEG signal is determined based on the confusion matrix and the independent signals after the artifact signals are removed.
2. The method according to claim 1, characterized in that: Acquire multi-channel physiological signal observation data within a preset time window, perform independent component analysis on the physiological signal observation data, and obtain a confusion matrix corresponding to the physiological signal observation data and multi-dimensional independent signals, including: Obtain the multi-channel physiological signal observation data x(t) within the preset time window, which is recorded as: x(t)=[x1(t),…,x p (t),x p+1 (t),…,x p+q (t)] T ,t=1,…,T In the formula, x i (t) represents the physiological signal observation data of the i-th dimension at time t, i=1,…,p+q, p represents the number of dimensions of the EEG signal in the physiological signal observation data, q represents the number of dimensions of the non-EEG signal in the physiological signal observation data, T represents the length of the preset time window, and T is greater than the preset time length; Determine the signal data relationship as: x(t)=As(t) Where s(t)=[s1(t),…,s p (t),s p+1 (t),…,s p+q (t)] T , represents the multi-dimensional independent signal to be calculated, and A represents the confusion matrix to be calculated; The Fast ICA method is used to solve the confusion matrix and multi-dimensional independent signals.
3. The method according to claim 1, characterized in that Determining whether the independent signal in each dimension is a global artifact signal according to the amplitude of each of the physiological signals and a preset global amplitude threshold comprises: For the independent signal of each dimension, determining the first maximum amplitude among the amplitudes of the physiological signals of each channel restored according to the independent signal within each sub-time window; If the maximum amplitude is greater than the preset global amplitude threshold, it is determined that the independent signal is a global artifact signal.
4. The method according to claim 1, characterized in that: Determining a local artifact signal in an independent signal of each dimension according to the amplitude of each of the physiological signals and a preset local amplitude threshold comprises: Determine the second maximum amplitude of the physiological signal observation data in each channel and each sub-time window; For each sub-time window, determine a target quantile of the second maximum amplitude of multiple channels corresponding to the sub-time window, and use the maximum value between the target quantile and the preset local amplitude threshold as the dynamic threshold corresponding to the sub-time window; For the independent signal of each dimension, determine the third maximum amplitude corresponding to each sub-time window in the amplitude of the physiological signal of each channel in each sub-time window restored based on the independent signal; for each sub-time window, if the third maximum amplitude is greater than the dynamic threshold corresponding to the sub-time window, determine that the local signal of the independent signal in the sub-time window is a local artifact signal.
5. The method according to claim 1, characterized in that Determining whether the independent signal of each dimension is an electrooculographic artifact signal according to the target row element corresponding to the independent signal of each dimension in the confusion matrix includes: For the independent signal of each dimension, determine the column vector corresponding to the independent signal in the confusion matrix, determine the first sum of the target row elements in the first preset number of rows in the column vector, determine the second sum of all row elements in the column vector, and determine the ratio of the first sum to the second sum; The ratios corresponding to the independent signals of each dimension are sorted, and the independent signals corresponding to the second preset number of ratios that are ranked top are determined to be electrooculogram artifact signals.
6. The method according to claim 1, characterized in that Determine the ECG artifact signal in the independent signal of each dimension according to the preset ECG channel signal or Q wave recognition algorithm, including: For the independent signal of each dimension, determine the correlation coefficient between the independent signal and the preset ECG channel signal, and if the correlation coefficient is greater than a preset correlation coefficient threshold, determine that the independent signal is an ECG artifact signal, wherein the type of the correlation coefficient includes at least one of the following: Pearson correlation coefficient, nonlinear correlation coefficient; or, For the independent signal of each dimension, the Q wave identification algorithm is used to determine the Q wave occurrence time in the independent signal; for every two adjacent Q wave occurrence times, the sampling number of the physiological signal observation data in each sub-time window and the time difference between the adjacent Q wave occurrence times are determined, and the ratio of the sampling number to the time difference is determined as a group of real-time heart rate data corresponding to the previous Q wave occurrence time in the adjacent Q wave occurrence times; for each sub-time window, the ratio of the standard deviation of multiple groups of real-time heart rate data corresponding to multiple Q wave occurrence times in the sub-time window to the average value is determined as the coefficient of variation corresponding to the sub-time window, and if the coefficient of variation is less than the preset coefficient of variation threshold, it is determined that the local signal of the independent signal in the sub-time window is an ECG artifact signal.
7. The method according to claim 1, characterized in that Removing artifact signals from the multi-dimensional independent signals, comprising: For the independent signal of each dimension, if the independent signal is determined to be a global artifact signal, an electrooculographic artifact signal, or an electrocardiographic artifact signal, the independent signal is assigned a value of 0; If the local signal corresponding to at least one sub-time window in the independent signal is determined to be a local artifact signal or an ECG artifact signal, the local signal in the independent signal is smoothed using a preset window function, wherein the type of the window function includes at least one of the following: a rectangular window function, a trapezoidal window function, a sinusoidal window function, and a maximum-normalized kernel function.
8. An electroencephalogram signal data processing device, characterized in that: include: An acquisition module, used to acquire multi-channel physiological signal observation data within a preset time window, perform independent component analysis on the physiological signal observation data, and obtain a confusion matrix corresponding to the physiological signal observation data and multi-dimensional independent signals; The first determination module is used to determine the artifact signal from the multi-dimensional independent signals, including: dividing the preset time window into multiple sub-time windows, restoring the physiological signal of each channel in each sub-time window according to the confusion matrix and the independent signal of each dimension, determining whether the independent signal of each dimension is a global artifact signal according to the amplitude of each physiological signal and a preset global amplitude threshold, or determining the local artifact signal in the independent signal of each dimension according to the amplitude of each physiological signal and a preset local amplitude threshold; and / or, determining whether the independent signal of each dimension is an electrooculographic artifact signal according to the target row element corresponding to the independent signal of each dimension in the confusion matrix; and / or, determining the ECG artifact signal in the independent signal of each dimension according to a preset ECG channel signal or a Q-wave recognition algorithm; The second determination module is used to remove the artifact signal from the multi-dimensional independent signal, and determine the target EEG signal according to the confusion matrix and the independent signal after the artifact signal is removed.
9. A computer program product, characterized in that include: A computer program, wherein when the computer program is executed by a processor, the method for processing electroencephalogram signal data described in any one of claims 1 to 7 is implemented.
10. An electronic device, characterized in that: include: A memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the EEG signal data processing method according to any one of claims 1 to 7 through the computer program.
Citation Information
Patent Citations
Method and system for quickly and automatically removing electromyographic artifacts, storage medium and computer equipment
CN112294340A
Epileptic seizure prediction method and system based on electroencephalogram signals, medium and equipment
CN118252473A
Device and method for denoising noise of electroencephalogram signal using series independent component analysis
KR1020160044641A
System and Method For Signal Denoising Using Independent Component Analysis and Fractal Dimension Estimation
US20090264786A1
Method and apparatus for denoising of physiological signals
US20130070792A1
Cited By
Ubiquitous network multi-mode privacy data protection method
CN120671189A
An ubiquitous network multi-modal privacy data protection method
CN120671189B
