A brain-muscle coupling analysis method based on bidirectional backtracking maximum information coefficient

Abstract

The present invention discloses a method for analyzing brain-muscle coupling based on the maximum information coefficient of bidirectional retrieval. The method first synchronously collects and preprocesses EEG and surface electromyography (SEM) signals. Next, a Chebyshev II bandpass filter is used to obtain a specific frequency band. The maximum information coefficient of bidirectional retrieval is then used to calculate the EMG signals in the specific frequency band. Finally, bidirectional brain-muscle coupling analysis is performed across different characteristic frequency bands. This method provides an effective method for exploring the mechanisms controlling motor function and has promising application prospects.

Description

Technical Field

[0001] The present invention belongs to the research field of the movement control mechanism of the nervous system, and particularly relates to a method for analyzing bidirectional coupling characteristics of different characteristic frequency bands based on the calculation of the maximum information coefficient of bidirectional backtracking. Background Art

[0002] During voluntary movement, the motor cortex of the brain issues commands to control muscle movement via motor neural pathways, and sensory information from the muscles is fed back to the cortex via sensory neural pathways to ensure accurate execution of the movement. This information interaction can be quantified by the coupling relationship between EEG signals and surface EMG signals of effector muscles. Therefore, brain-muscle coupling has become an important approach to reveal the control-feedback mechanism of the nervous system and to evaluate motor function and rehabilitation outcomes in patients with neurological diseases such as stroke.

[0003] A key challenge in modeling the interrelationships between complex neurophysiological signals is accurately capturing the information flow between them, which includes two important metrics: direction and strength. Coherence is one of the main methods for quantifying brain-muscle coupling. However, the lack of ability to identify the direction of information interaction limits its application. Granger causality and its extension methods, as directional methods for measuring causal relationships between time series, are based on linear systems. However, neurophysiological signals have been shown to be nonlinear. Therefore, the effectiveness of Granger causality in analyzing relationships between nonlinear neurophysiological signals has been questioned. To address the nonlinearity between neurophysiological signals, transfer entropy was proposed and has been recognized in recent years as an effective method for detecting causal relationships between neurophysiological signals. However, when the time series is not long enough, transfer entropy cannot accurately detect coupling in practical applications.

[0004] Reshef et al. proposed the maximum information coefficient method, whose commonality property satisfies the requirements of measuring different functional relationships; its fairness property ensures that different functional relationships obtain similar measurement values ​​under the same noise level. The maximum information coefficient has been widely used in the field of neuroscience. Chen et al. proposed a new method for calculating the maximum information coefficient value - ChiMIC, which terminates the grid optimization through the χ2 test, eliminating the limitation of the original algorithm on the maximum grid size. Dan et al. proposed an algorithm that adds a search backtracking process on the equally divided axis to eliminate the limitation of equal division and obtain a better grid partition. However, it is also symmetrical due to the symmetry of mutual information, and therefore cannot identify the direction of information interaction between signals.

[0005] To overcome this limitation, this paper proposes a new Bidirectional Backtracking Maximum Information Coefficient (BBMIC) algorithm, which captures the information transmission delay between two time series by introducing a time delay parameter to infer the causal relationship. Summary of the Invention

[0006] The object of the present invention is to provide an analysis method for obtaining bidirectional coupling characteristics of different characteristic frequency bands, comprising the following steps:

[0007] Step (1), synchronously collecting EEG and surface electromyography signals; preprocessing the EEG and surface electromyography signals to obtain preprocessed signals;

[0008] Step (2), processing the preprocessed signal with a Chebyshev II bandpass filter to obtain a beta band subband signal and a gamma band subband signal; wherein the beta band is 15 Hz to 30 Hz, and the gamma band is 31 Hz to 60 Hz;

[0009] Step (3), representing the EEG and surface EMG signals collected in step (1) as time series to obtain an EEG time series and a surface EMG time series; representing the relationship between the EEG time series and the surface EMG time series in the form of a scatter plot to obtain a scatter plot distributed in a two-dimensional space; dividing the horizontal and vertical axes of the two-dimensional space into a grid, and using a χ2 test method to find a gridding scheme that maximizes mutual information;

[0010] The bidirectional maximum information coefficient (BBMIC) of the beta band sub-band signal and the gamma band sub-band signal is calculated separately. The calculation method is as follows:

[0011] Define the EEG signal in any one of the beta-band subband signal and the gamma-band subband signal obtained in step (2) as X, and the surface electromyography signal as Y; by introducing the time lag parameter τ to determine the direction of information transmission between the two signals, the bidirectional maximum information coefficient BBMIC is calculated according to the following formula:

[0012]

[0013] Where I(X,Y,τ) is the mutual information value of X and Y under the delay τ, x and y are the number of grids on the horizontal and vertical axes respectively. When BBMIC reaches its maximum value, the corresponding τ is negative, indicating the direction is from X to Y, and positive, indicating the direction is from Y to X.

[0014] Select several BBMICs within a preset time interval and sum them to obtain the total information flow C between the EEG signal and the surface EMG signal. BBMIC :

[0015] Step (4): using the total information flow C between the EEG signal and the surface electromyography signal calculated in step (3) BBMIC Perform bidirectional brain-muscle coupling analysis in different characteristic frequency bands.

[0016] Preferably, in step (1), the surface electromyographic signals include: electromyographic signals of the flexor digitorum superficialis (FDS), the brachioradialis (B), the flexor carpi radialis (FCR), the flexor carpi ulnaris (FCU), the biceps brachii (BB), and the extensor digitorum (ED);

[0017] The sampling frequency of the synchronous acquisition is 1000 Hz;

[0018] In step (3), the preset time interval is 40 milliseconds, and the number of the plurality of BBMICs is 40.

[0019] Preferably, in step (1), the pretreatment includes:

[0020] The EEG signal is subjected to a 0-75 Hz low-pass filter process to eliminate 50 Hz power frequency interference; the surface electromyography signal is subjected to a 0-200 Hz low-pass filter process to eliminate 50 Hz power frequency interference.

[0021] Preferably, in step (4), the bidirectional brain-muscle coupling analysis includes: analyzing the coupling conditions of different characteristic frequency bands from the aspects of coupling strength and information flow direction, and BBMIC The value of C is used to analyze the coupling strength. BBMIC The larger the value, the greater the coupling strength; by placing two C BBMIC The relative size of the value is used to analyze the direction of information flow, C BBMIC The direction with a larger value represents the direction of information flow.

[0022] The present invention has the following beneficial effects: Traditional brain-muscle coupling analysis methods cannot effectively describe the coupling strength and signal transmission direction of signals at different characteristic frequency bands. To address this issue, the present invention proposes a brain-muscle coupling analysis method based on the maximum information coefficient of bidirectional backtracking. This method quantitatively describes the causal coupling relationship between brain and muscle at different characteristic frequency bands under different grip forces in healthy individuals. This provides a theoretical basis for further exploring the pathological mechanisms of motor dysfunction and evaluating rehabilitation function. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Flowchart of the method of the present invention.

[0024] Figure 2 Flowchart of the experimental paradigm.

[0025] Figure 3 This is a diagram illustrating the definition of the χ2 statistic.

[0026] Figure 4 The brain myoelectric coupling C of the subjects at different forces BBMIC Comparison chart of values. DETAILED DESCRIPTION

[0027] The following is a detailed description of an embodiment of the present invention in conjunction with the accompanying drawings: This embodiment is implemented on the premise of the technical solution of the present invention, and provides a detailed implementation plan and specific operation process.

[0028] The present invention proposes a coupling analysis method based on the maximum information coefficient of bidirectional backtracking, such as Figure 1 As shown, the implementation of the present invention mainly includes four steps: (1) synchronous acquisition of EEG and surface electromyography signals and preprocessing; (2) using a Chebyshev II bandpass filter to obtain a specific frequency band for the preprocessed signal; (3) calculating the maximum information coefficient value of the two-way backtracking; (4) performing two-way brain-muscle coupling analysis on different characteristic frequency bands based on the calculation results of step (3).

[0029] The following describes each step in detail.

[0030] Step 1: Synchronous acquisition and preprocessing of EEG and surface EMG signals

[0031] A total of 8 healthy subjects (aged 24-26, all right-handed) were recruited in this study. They did not have any history of neurological diseases. All subjects were introduced to the experimental procedures and possible effects of the experiment. They voluntarily signed consent forms and were informed one day in advance not to do any strenuous exercise and to keep their bodies relaxed. The experiment simultaneously collected EEG and myoelectric signals of the flexor digitorum superficialis (FDS), brachioradialis (B), flexor carpi radialis (FCR), flexor carpi ulnaris (FCU), biceps brachii (BB), and extensor digitorum (ED). Before collecting and placing the electrodes, the subjects' scalps were kept clean, the impedance was kept at a low level, and the skin surface of the upper arm was cleaned with alcohol cotton pads. This experiment used a spring gripper to test the maximum voluntary contraction grip (MVC) three times, and the average value was taken as the MVC for each subject. The grasping movements of the upper limb movements were performed in the order of 10%, 20%, and 30% of the MVC. The complete and detailed experimental paradigm process is as follows. Figure 2 As shown. The sampling frequency was 1000 Hz, and the sequential tasks for each subject were as follows: the first 20 seconds were a static state, 21-25 seconds were a 5-second upper limb hand grasping task, and 26-45 seconds were a resting static state. Finally, each subject repeated the above steps a total of 3 times. Because EEG and EMG signals are unstable and usually contain noise, the collected EEG signals were filtered to eliminate 50 Hz power frequency interference and low-pass filtered from 0 to 75 Hz; the EMG signals were filtered to eliminate 50 Hz power frequency interference and low-pass filtered from 0 to 200 Hz before subsequent EMG coupling analysis.

[0032] Step 2: Use Chebyshev II bandpass filter to obtain specific frequency bands for the preprocessed signal

[0033] The present invention further utilizes Chebyshev II bandpass filter to extract brain and myoelectric signals in beta frequency band (15Hz-30Hz) and gamma frequency band (31Hz-60Hz) for further analysis.

[0034] Step 3: Calculate the value of the bidirectional maximum information coefficient. The variables are represented by time series, and the relationship between the two variables is represented by a scatter plot. The two sets of data constitute multiple coordinate points distributed in a two-dimensional space. The two-dimensional space is divided into grids along the horizontal and vertical axes. There are many different gridding schemes, and the gridding scheme that maximizes the mutual information is found. The coupled analysis algorithm based on the bidirectional maximum information coefficient uses the χ2 test to terminate the grid optimization. Given an optimal segmentation point, if the χ2 test value of the optimal segmentation point is lower than a given threshold (the threshold value of the present invention = 0.01), the segmentation point is valid, and the present algorithm continues to search for the next optimal segmentation point. Otherwise, the present algorithm will stop the search process.

[0035] according to Figure 3 , with the kth optimal segmentation point SP on the horizontal axis k For example, suppose the vertical axis is divided into ny, SP k In SP k-2 and SP k-1 The horizontal axis is divided into the s-1th and sth columns. Figure 3 The gray part (ny×2) is called the SP of the χ2 test. k Detection area. The χ2 statistic is defined as follows:

[0036]

[0037] where n j,i is the number of data points in row j and column i, n *,i is the number of data points in column i, n j,* is the number of data points in the jth row, and N is the number of data points in the detection area.

[0038] The algorithm of the present invention divides the optimal partitioning into two stages. First, an equal partition is performed on one axis (the vertical axis, for example). Under the constraints of the χ2 test, a dynamic programming algorithm is used to locate the optimal partition on the horizontal axis. The obtained horizontal partition is fixed, and the optimal partition of the vertical axis is returned. The corresponding normalized mutual information under this grid partition is then calculated.

[0039] The present invention introduces a time lag parameter τ to determine the direction of information transmission between two signals, which is defined as follows:

[0040]

[0041] Where I(X,Y,τ) is the mutual information between signals X and Y at a delay of τ, and x and y are the number of grid cells on the horizontal and vertical axes, respectively. When BBMIC reaches its maximum value, a negative τ indicates a direction from X to Y, and a positive τ indicates a direction from Y to X.

[0042] In order to estimate the total information flow C between two physiological time series (EEG and EMG) BBMIC , represented by the sum of BBMICs within a certain delay D. In this study, the delay D was set to 40 data points with a step size of 1. The larger the value, the stronger the coupling strength between brain and muscle signals, and vice versa.

[0043] Step 4: Perform bidirectional coupling analysis of different characteristic frequency bands based on the calculation results of step 3.

[0044] Specifically: Analyze the coupling of different characteristic frequency bands from the two aspects of coupling strength and information flow direction. BBMIC The value of indicates the coupling strength. BBMIC The values ​​are compared, and the direction of the larger value indicates the direction of information flow.

[0045] This study investigates differences in brain-muscle bidirectional coupling under varying grip forces. Because the C3 channel, located in the primary sensorimotor cortex, is more sensitive to EEG signal changes during upper limb movement, and agonist muscles are more sensitive to EMG signal changes during grip force, this study analyzed experimental results using signals from the C3 EEG channel and the FDS.

[0046] In order to compare the differences in brain-muscle coupling in both directions and in each frequency band under different grip strengths, an independent sample T test was performed. Figure 4 In the figure, '*' represents p<0.05, '**' represents p<0.01, and '***' represents p<0.001. As can be seen from the figure, brain-myoelectric coupling is divided into EEG→EMG (downward) and EMG→EEG (upward), and the coupling between brain and muscle is similar, which is consistent with the conclusions of existing studies. The coupling strength of the downward beta band is always greater than that of the upward beta band, which reflects that during movement, the motor cortex needs to transmit more information to the muscles for control. The coupling strength of the beta band is relatively large, which confirms that when maintaining constant force output, brain-myoelectric coupling is mainly manifested in the beta band, and the oscillation of the beta band between the motor cortex and the muscle plays a major role, reflecting the information interaction between the cortex and the muscle. Under different grip forces, there is no significant difference in the coupling strength of the gamma band, while the coupling strength of the beta band decreases with increasing grip force.

[0047] The embodiments described above are merely preferred examples of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should fall within the scope of protection determined by the claims of the present invention.

Claims

1. A brain-muscle coupling analysis method based on bidirectional backtracking maximum information coefficient, characterized in that: The following steps are involved: Step (1), synchronously collecting EEG and surface EMG signals; Preprocessing the EEG and surface EMG signals to obtain preprocessed signals; Step (2), processing the preprocessed signal with a Chebyshev II bandpass filter to obtain a beta band subband signal and a gamma band subband signal; wherein the beta band is 15 Hz to 30 Hz, and the gamma band is 31 Hz to 60 Hz; Step (3), representing the EEG and surface EMG signals collected in step (1) as time series to obtain an EEG time series and a surface EMG time series; representing the relationship between the EEG time series and the surface EMG time series in the form of a scatter plot to obtain a scatter plot distributed in a two-dimensional space; dividing the horizontal and vertical axes of the two-dimensional space into a grid, and using a χ2 test method to find a gridding scheme that maximizes mutual information; The bidirectional maximum information coefficient (BBMIC) of the beta band sub-band signal and the gamma band sub-band signal is calculated separately. The calculation method is as follows: Define the EEG signal in any one of the beta-band subband signal and the gamma-band subband signal obtained in step (2) as X, and the surface electromyography signal as Y; by introducing the time lag parameter τ to determine the direction of information transmission between the two signals, the bidirectional maximum information coefficient BBMIC is calculated according to the following formula: Where I(X,Y,τ) is the mutual information value of X and Y under the delay τ, x and y are the number of grids on the horizontal and vertical axes respectively. When BBMIC reaches its maximum value, the corresponding τ is negative, indicating the direction is from X to Y, and positive, indicating the direction is from Y to X. Select several BBMICs within a preset time interval and sum them to obtain the total information flow C between the EEG signal and the surface EMG signal. BBMIC : Step (4): using the total information flow C between the EEG signal and the surface electromyography signal calculated in step (3) BBMIC Perform bidirectional brain-muscle coupling analysis in different characteristic frequency bands.

2. The brain-muscle coupling analysis method based on bidirectional backtracking maximum information coefficient according to claim 1, characterized in that: In the step (1), the surface electromyographic signals include: electromyographic signals of the flexor digitorum superficialis (FDS), the brachioradialis (B), the flexor carpi radialis (FCR), the flexor carpi ulnaris (FCU), the biceps brachii (BB), and the extensor digitorum (ED); The sampling frequency of the synchronous acquisition is 1000 Hz; In step (3), the preset time interval is 40 milliseconds, and the number of the plurality of BBMICs is 40.

3. The brain-muscle coupling analysis method based on bidirectional backtracking maximum information coefficient according to claim 2, characterized in that: In the step (1), the pretreatment includes: The EEG signal is subjected to a 0-75 Hz low-pass filter process to eliminate 50 Hz power frequency interference; the surface electromyography signal is subjected to a 0-200 Hz low-pass filter process to eliminate 50 Hz power frequency interference.

4. The brain-muscle coupling analysis method based on bidirectional backtracking maximum information coefficient according to claim 1, characterized in that: In the step (4), the bidirectional brain-muscle coupling analysis includes: analyzing the coupling conditions of different characteristic frequency bands from the two aspects of coupling strength and information flow direction, and BBMIC The value of C is used to analyze the coupling strength. BBMIC The larger the value, the greater the coupling strength; by placing two C BBMIC The relative size of the value is used to analyze the direction of information flow, C BBMIC The direction with a larger value represents the direction of information flow.

Images