A method and system for extracting features of electroencephalogram signals

Abstract

The present invention discloses a method and system for extracting features of an electroencephalogram (EEG) signal, and relates to the technical field of EEG signal feature extraction. The method comprises: acquiring an EEG signal; determining the autocorrelation characteristics of the EEG signal using an empirical mode decomposition algorithm optimized by the Pearson correlation coefficient and a multifractal detrended fluctuation analysis algorithm; determining the complexity characteristics of the EEG signal; determining the autocorrelation characteristics and the complexity characteristics as the characteristics of the EEG signal, and the characteristics of the EEG signal are used to classify the EEG signal as fatigue driving. The autocorrelation characteristics of the EEG signal are determined using an empirical mode decomposition algorithm optimized by the Pearson correlation coefficient and a multifractal detrended fluctuation analysis algorithm, so that the trend of the EEG signal is fully eliminated, and the hierarchical fuzzy entropy, asymmetric entropy characteristics, and asymmetric entropy index are comprehensively determined as the complexity characteristics, which fully reflects the overall and local entropy value differences of the EEG signal, and the feature extraction is more comprehensive.

Description

Technical Field

[0001] The present invention relates to the technical field of electroencephalogram (EEG) signal feature extraction, and in particular to a method and system for extracting EEG signal features. Background Art

[0002] Currently, fatigue driving has become the fifth leading cause of road safety hazards. Monitoring drivers' EEG signals and objectively assessing their fatigue status have become a hot topic of research. EEG signals (Electroencephalogram, EEG) are the potential differences generated by the electrophysiological activity of nerve cells in the cerebral cortex and contain rich pathological information. Analyzing EEG signals can extract implicit and subtle features, enabling timely warnings to fatigued drivers. EEG feature extraction methods primarily include time-domain analysis, frequency-domain analysis, time-frequency analysis, and nonlinear analysis, such as Hjorth parameters, power spectrum estimation, wavelet transform, and fractal dimension.

[0003] As the output of a nonlinear complex system, EEG signals possess two important properties: autocorrelation and complexity. Existing technologies often focus on extracting either autocorrelation or complexity features. When extracting autocorrelation features, both the Multifractal Detrended Fluctuation Analysis (MFDFA) method and the Empirical Mode Decomposition-modified MFDFA algorithm (MFDFAemd) have issues with the selection of trend terms, resulting in insufficient trend elimination and irrational EEG feature extraction. When extracting complexity features, extracting only hierarchical fuzzy entropy is insufficient and fails to fully reflect the differences in global and local entropy values ​​in EEG signals. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and system for extracting features of EEG signals, so that the features of EEG signals can be extracted more comprehensively.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A method for extracting features from an electroencephalogram (EEG) signal, comprising:

[0007] Acquire EEG signals;

[0008] The autocorrelation characteristics of the EEG signal are determined using an empirical mode decomposition algorithm optimized by Pearson correlation coefficient and a multifractal detrended fluctuation analysis algorithm;

[0009] Determine the complexity characteristics of the EEG signal; the complexity characteristics include: hierarchical fuzzy entropy, asymmetric entropy characteristics and asymmetric entropy index; the hierarchical fuzzy entropy includes the hierarchical fuzzy entropy of the left brain in different frequency bands and the hierarchical fuzzy entropy of the right brain in the corresponding frequency bands, the asymmetric entropy characteristics include the difference between the hierarchical fuzzy entropy of the left brain in different frequency bands and the hierarchical fuzzy entropy of the right brain in the corresponding frequency bands, the asymmetric entropy index includes the left brain asymmetric entropy index and the right brain asymmetric entropy index, the left brain asymmetric entropy index is obtained based on the hierarchical fuzzy entropy of the left brain in all frequency bands, and the right brain asymmetric entropy index is obtained based on the hierarchical fuzzy entropy of the right brain in all frequency bands;

[0010] The autocorrelation feature and the complexity feature are determined as features of the EEG signal, and the features of the EEG signal are used to classify the EEG signal as fatigue driving.

[0011] Optionally, the method of determining the autocorrelation characteristics of the EEG signal using an empirical mode decomposition algorithm optimized by Pearson correlation coefficient and a multifractal detrended fluctuation analysis algorithm specifically includes:

[0012] Performing mean-summing on the EEG signal to obtain a mean-summing signal;

[0013] Decomposing the mean-decomposed sum signal into a residual term and a plurality of intrinsic mode functions using an empirical mode decomposition algorithm;

[0014] Fitting the de-meaned summation signal using the least squares method to obtain a fitting polynomial;

[0015] Calculating the Pearson correlation coefficient between each of the intrinsic mode functions and the fitting polynomial;

[0016] Sort all Pearson correlation coefficients from large to small to obtain a correlation coefficient sequence;

[0017] Determine the first-ranked Pearson correlation coefficient in the correlation coefficient sequence as a first correlation coefficient, and determine the second-ranked Pearson correlation coefficient in the correlation coefficient sequence as a second correlation coefficient;

[0018] Determining a first intrinsic mode function and a second intrinsic mode function; the first intrinsic mode function is the intrinsic mode function corresponding to the first correlation coefficient, and the second intrinsic mode function is the intrinsic mode function corresponding to the second correlation coefficient;

[0019] The multifractal detrended fluctuation analysis algorithm is used to obtain the autocorrelation characteristics of the EEG signal according to the residual term, the first intrinsic mode function and the second intrinsic mode function.

[0020] Optionally, determining the complexity feature of the EEG signal specifically includes:

[0021] Performing wavelet packet decomposition and reconstruction on the left brain signal in the EEG signal to obtain left brain frequency band signals in four frequency bands;

[0022] Performing wavelet packet decomposition and reconstruction on the right brain signal in the EEG signal to obtain right brain sub-frequency band signals in four frequency bands;

[0023] Calculate the fuzzy entropy of the left brain sub-band signal in each frequency band to obtain the left brain hierarchical fuzzy entropy of the four frequency bands; the left brain hierarchical fuzzy entropies of the four frequency bands are: the first left brain fuzzy entropy, the second left brain fuzzy entropy, the third left brain fuzzy entropy and the fourth left brain fuzzy entropy;

[0024] Calculate the fuzzy entropy of the right brain sub-band signal in each frequency band to obtain the right brain hierarchical fuzzy entropy of the four frequency bands; the right brain hierarchical fuzzy entropy of the four frequency bands are: the first right brain fuzzy entropy, the second right brain fuzzy entropy, the third right brain fuzzy entropy and the fourth right brain fuzzy entropy;

[0025] Calculating the asymmetric entropy feature and the asymmetric entropy index according to the left brain hierarchical fuzzy entropy of four frequency bands and the right brain hierarchical fuzzy entropy of four frequency bands;

[0026] The left brain hierarchical fuzzy entropy, the right brain hierarchical fuzzy entropy, the asymmetric entropy feature and the asymmetric entropy index are determined as complexity features of the electroencephalogram signal.

[0027] Optionally, the calculating the asymmetric entropy feature and the asymmetric entropy index according to the left brain layered fuzzy entropy of four frequency bands and the right brain layered fuzzy entropy of four frequency bands specifically includes:

[0028] Determine a first asymmetric fuzzy entropy feature based on the first left-brain fuzzy entropy and the first right-brain fuzzy entropy, determine a second asymmetric fuzzy entropy feature based on the second left-brain fuzzy entropy and the second right-brain fuzzy entropy, determine a third asymmetric fuzzy entropy feature based on the third left-brain fuzzy entropy and the third right-brain fuzzy entropy, and determine a fourth asymmetric fuzzy entropy feature based on the fourth left-brain fuzzy entropy and the fourth right-brain fuzzy entropy;

[0029] determining the first asymmetric fuzzy entropy feature, the second asymmetric fuzzy entropy feature, the third asymmetric fuzzy entropy feature, and the fourth asymmetric fuzzy entropy feature as the asymmetric entropy features;

[0030] Calculating the left brain asymmetric entropy index according to the first left brain fuzzy entropy, the second left brain fuzzy entropy, the third left brain fuzzy entropy, and the fourth left brain fuzzy entropy; calculating the right brain asymmetric entropy index according to the first right brain fuzzy entropy, the second right brain fuzzy entropy, the third right brain fuzzy entropy, and the fourth right brain fuzzy entropy;

[0031] The left brain asymmetric entropy index and the right brain asymmetric entropy index are determined as asymmetric entropy indices.

[0032] A feature extraction system for an electroencephalogram signal, comprising:

[0033] An EEG signal acquisition module, used to acquire EEG signals;

[0034] An autocorrelation feature determination module is used to determine the autocorrelation feature of the EEG signal using an empirical mode decomposition algorithm optimized by Pearson correlation coefficient and a multifractal detrended fluctuation analysis algorithm;

[0035] a complexity feature determination module, configured to determine the complexity features of the EEG signal; the complexity features comprising: hierarchical fuzzy entropy, asymmetric entropy features, and an asymmetric entropy index; the hierarchical fuzzy entropy comprising the hierarchical fuzzy entropy of the left brain in different frequency bands and the hierarchical fuzzy entropy of the right brain in corresponding frequency bands; the asymmetric entropy feature comprising the difference between the hierarchical fuzzy entropy of the left brain in different frequency bands and the hierarchical fuzzy entropy of the right brain in corresponding frequency bands; the asymmetric entropy index comprising the left brain asymmetric entropy index and the right brain asymmetric entropy index; the left brain asymmetric entropy index being obtained based on the hierarchical fuzzy entropy of the left brain in all frequency bands; and the right brain asymmetric entropy index being obtained based on the hierarchical fuzzy entropy of the right brain in all frequency bands;

[0036] A feature determination module is used to determine the autocorrelation feature and the complexity feature as features of the EEG signal, and the features of the EEG signal are used to classify the EEG signal as fatigue driving.

[0037] Optionally, the autocorrelation feature determination module specifically includes:

[0038] a de-meaning summing unit, configured to perform de-meaning summing on the EEG signal to obtain a de-meaning summing signal;

[0039] An empirical mode decomposition unit, configured to decompose the mean-decomposed sum signal into a residual term and a plurality of intrinsic mode functions using an empirical mode decomposition algorithm;

[0040] A fitting polynomial determination unit, configured to fit the de-meaned summation signal using a least squares method to obtain a fitting polynomial;

[0041] a correlation coefficient calculation unit, configured to calculate the Pearson correlation coefficient between each of the intrinsic mode functions and the fitting polynomial;

[0042] The sorting unit is used to sort all Pearson correlation coefficients from large to small to obtain a correlation coefficient sequence;

[0043] a sub-correlation coefficient determining unit, configured to determine the first-ranked Pearson correlation coefficient in the correlation coefficient sequence as a first correlation coefficient, and determine the second-ranked Pearson correlation coefficient in the correlation coefficient sequence as a second correlation coefficient;

[0044] an intrinsic mode function determining unit, configured to determine a first intrinsic mode function and a second intrinsic mode function; the first intrinsic mode function is the intrinsic mode function corresponding to the first correlation coefficient, and the second intrinsic mode function is the intrinsic mode function corresponding to the second correlation coefficient;

[0045] The autocorrelation feature determination unit is used to obtain the autocorrelation feature of the EEG signal according to the residual term, the first intrinsic mode function and the second intrinsic mode function by using the multifractal detrended fluctuation analysis algorithm.

[0046] Optionally, the complexity feature determination module specifically includes:

[0047] a left brain frequency band signal acquisition unit, configured to perform wavelet packet decomposition and reconstruction on the left brain signal in the EEG signal to obtain left brain frequency band signals in four frequency bands;

[0048] a right brain frequency band signal acquisition unit, configured to perform wavelet packet decomposition and reconstruction on the right brain signal in the EEG signal to obtain right brain frequency band signals in four frequency bands;

[0049] The left brain hierarchical fuzzy entropy acquisition unit is used to calculate the fuzzy entropy of the left brain sub-band signal in each frequency band to obtain the left brain hierarchical fuzzy entropies of the four frequency bands; the left brain hierarchical fuzzy entropies of the four frequency bands are: a first left brain fuzzy entropy, a second left brain fuzzy entropy, a third left brain fuzzy entropy and a fourth left brain fuzzy entropy;

[0050] The right brain hierarchical fuzzy entropy acquisition unit is used to calculate the fuzzy entropy of the right brain sub-band signal in each frequency band to obtain the right brain hierarchical fuzzy entropies of four frequency bands; the right brain hierarchical fuzzy entropies of the four frequency bands are: a first right brain fuzzy entropy, a second right brain fuzzy entropy, a third right brain fuzzy entropy and a fourth right brain fuzzy entropy;

[0051] an asymmetric entropy calculation unit, configured to calculate the asymmetric entropy feature and the asymmetric entropy index according to the left brain hierarchical fuzzy entropies of four frequency bands and the right brain hierarchical fuzzy entropies of four frequency bands;

[0052] The complexity feature determination unit is used to determine the left brain hierarchical fuzzy entropy, the right brain hierarchical fuzzy entropy, the asymmetric entropy feature and the asymmetric entropy index as the complexity feature of the electroencephalogram signal.

[0053] Optionally, the asymmetric entropy calculation unit specifically includes:

[0054] a fuzzy entropy feature calculation subunit, configured to determine a first asymmetric fuzzy entropy feature according to the first left-brain fuzzy entropy and the first right-brain fuzzy entropy, determine a second asymmetric fuzzy entropy feature according to the second left-brain fuzzy entropy and the second right-brain fuzzy entropy, determine a third asymmetric fuzzy entropy feature according to the third left-brain fuzzy entropy and the third right-brain fuzzy entropy, and determine a fourth asymmetric fuzzy entropy feature according to the fourth left-brain fuzzy entropy and the fourth right-brain fuzzy entropy;

[0055] a fuzzy entropy feature determining subunit, configured to determine the first asymmetric fuzzy entropy feature, the second asymmetric fuzzy entropy feature, the third asymmetric fuzzy entropy feature, and the fourth asymmetric fuzzy entropy feature as the asymmetric entropy feature;

[0056] a fuzzy entropy index calculation subunit, configured to calculate the left brain asymmetric entropy index according to the first left brain fuzzy entropy, the second left brain fuzzy entropy, the third left brain fuzzy entropy, and the fourth left brain fuzzy entropy, and calculate the right brain asymmetric entropy index according to the first right brain fuzzy entropy, the second right brain fuzzy entropy, the third right brain fuzzy entropy, and the fourth right brain fuzzy entropy;

[0057] The fuzzy entropy index determining subunit is configured to determine the left brain asymmetric entropy index and the right brain asymmetric entropy index as asymmetric entropy indices.

[0058] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0059] The present invention discloses a method and system for extracting features of an EEG signal, the method comprising: acquiring an EEG signal; determining the autocorrelation characteristics of the EEG signal using an empirical mode decomposition algorithm optimized by the Pearson correlation coefficient and a multifractal detrended fluctuation analysis algorithm; determining the complexity characteristics of the EEG signal; the complexity characteristics comprising: hierarchical fuzzy entropy, asymmetric entropy characteristics and asymmetric entropy index; determining the autocorrelation characteristics and complexity characteristics as the characteristics of the EEG signal, and the characteristics of the EEG signal are used to classify the EEG signal as fatigue driving. The autocorrelation characteristics of the EEG signal are determined using an empirical mode decomposition algorithm optimized by the Pearson correlation coefficient and a multifractal detrended fluctuation analysis algorithm, so that the trend of the EEG signal is fully eliminated, and the hierarchical fuzzy entropy, asymmetric entropy characteristics and asymmetric entropy index are comprehensively determined as the complexity characteristics, which fully reflects the overall and local entropy value differences of the EEG signal. Therefore, the extraction of EEG signal features is more comprehensive. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0061] Figure 1 Flowchart of a method for extracting features from EEG signals provided by an embodiment of the present invention;

[0062] Figure 2 This is a block diagram of the EEG signal feature extraction system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0063] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0064] The purpose of the present invention is to provide a method and system for extracting features of EEG signals, aiming to extract the features of EEG signals more comprehensively, which can be applied to the technical field of EEG signal feature extraction.

[0065] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0066] Figure 1 Flowchart of the feature extraction method of EEG signal provided by the embodiment of the present invention. Figure 1 As shown, the feature extraction method of the EEG signal in this embodiment includes:

[0067] Step 101: Acquire EEG signals.

[0068] Step 102: using the Pearson correlation coefficient optimized empirical mode decomposition algorithm and the multifractal detrended fluctuation analysis algorithm to determine the autocorrelation characteristics of the EEG signal.

[0069] Step 103: Determine the complexity characteristics of the EEG signal; the complexity characteristics include: hierarchical fuzzy entropy, asymmetric entropy characteristics, and asymmetric entropy index. The hierarchical fuzzy entropy includes the hierarchical fuzzy entropy of the left brain in different frequency bands and the hierarchical fuzzy entropy of the right brain in the corresponding frequency bands. The asymmetric entropy characteristics include the difference between the hierarchical fuzzy entropy of the left brain in different frequency bands and the hierarchical fuzzy entropy of the right brain in the corresponding frequency bands. The asymmetric entropy index includes the left brain asymmetric entropy index and the right brain asymmetric entropy index. The left brain asymmetric entropy index is obtained based on the hierarchical fuzzy entropy of the left brain in all frequency bands, and the right brain asymmetric entropy index is obtained based on the hierarchical fuzzy entropy of the right brain in all frequency bands.

[0070] Step 104: The autocorrelation feature and the complexity feature are determined as features of the EEG signal. The features of the EEG signal are used to classify the EEG signal as fatigue driving.

[0071] Specifically, the autocorrelation features include: the maximum value of the generalized Hurst spectrum, the minimum value of the scaling exponential spectrum, the two-segment slopes of the scaling exponential spectrum, and the width of the multifractal spectrum, which constitute the five-dimensional fractal features.

[0072] As an optional implementation, step 102 specifically includes:

[0073] The EEG signal is de-meaned and summed to obtain a de-meaned summed signal.

[0074] The empirical mode decomposition algorithm optimized by Pearson correlation coefficient is used to decompose the mean-summed signal into a residual term and multiple intrinsic mode functions.

[0075] The least square method is used to fit the de-meaned sum signal to obtain a fitting polynomial.

[0076] The Pearson correlation coefficient between each intrinsic mode function and the fitting polynomial is calculated.

[0077] Sort all Pearson correlation coefficients from large to small to obtain a correlation coefficient sequence;

[0078] The first-ranked Pearson correlation coefficient in the correlation coefficient sequence is determined as the first correlation coefficient, and the second-ranked Pearson correlation coefficient in the correlation coefficient sequence is determined as the second correlation coefficient.

[0079] A first intrinsic mode function and a second intrinsic mode function are determined; the first intrinsic mode function is the intrinsic mode function corresponding to the first correlation coefficient, and the second intrinsic mode function is the intrinsic mode function corresponding to the second correlation coefficient.

[0080] The multifractal detrended fluctuation analysis algorithm is used to obtain the autocorrelation characteristics of the EEG signal based on the residual term, the first intrinsic mode function and the second intrinsic mode function.

[0081] As an optional implementation, step 103 specifically includes:

[0082] The left brain signal in the EEG signal is decomposed and reconstructed by wavelet packets to obtain the left brain frequency band signals in four frequency bands.

[0083] The right brain signal in the EEG signal is decomposed and reconstructed by wavelet packets to obtain right brain frequency band signals in four frequency bands.

[0084] The fuzzy entropy of the left brain sub-band signal of each frequency band is calculated to obtain the left brain hierarchical fuzzy entropy of the four frequency bands; the left brain hierarchical fuzzy entropies of the four frequency bands are: the first left brain fuzzy entropy, the second left brain fuzzy entropy, the third left brain fuzzy entropy and the fourth left brain fuzzy entropy.

[0085] The fuzzy entropy of the right brain sub-band signal of each frequency band is calculated to obtain the right brain hierarchical fuzzy entropy of four frequency bands; the right brain hierarchical fuzzy entropies of the four frequency bands are: the first right brain fuzzy entropy, the second right brain fuzzy entropy, the third right brain fuzzy entropy and the fourth right brain fuzzy entropy.

[0086] The asymmetric entropy features and asymmetric entropy index were calculated based on the left brain hierarchical fuzzy entropy of four frequency bands and the right brain hierarchical fuzzy entropy of four frequency bands.

[0087] The left brain hierarchical fuzzy entropy, right brain hierarchical fuzzy entropy, asymmetric entropy feature and asymmetric entropy index were identified as the complexity features of EEG signals.

[0088] As an optional implementation, the asymmetric entropy feature and the asymmetric entropy index are calculated based on the left brain layered fuzzy entropy of the four frequency bands and the right brain layered fuzzy entropy of the four frequency bands, specifically including:

[0089] A first asymmetric fuzzy entropy feature is determined according to the first left brain fuzzy entropy and the first right brain fuzzy entropy, a second asymmetric fuzzy entropy feature is determined according to the second left brain fuzzy entropy and the second right brain fuzzy entropy, a third asymmetric fuzzy entropy feature is determined according to the third left brain fuzzy entropy and the third right brain fuzzy entropy, and a fourth asymmetric fuzzy entropy feature is determined according to the fourth left brain fuzzy entropy and the fourth right brain fuzzy entropy.

[0090] The first asymmetric fuzzy entropy feature, the second asymmetric fuzzy entropy feature, the third asymmetric fuzzy entropy feature, and the fourth asymmetric fuzzy entropy feature are determined as asymmetric entropy features.

[0091] The left brain asymmetric entropy index is calculated according to the first left brain fuzzy entropy, the second left brain fuzzy entropy, the third left brain fuzzy entropy and the fourth left brain fuzzy entropy, and the right brain asymmetric entropy index is calculated according to the first right brain fuzzy entropy, the second right brain fuzzy entropy, the third right brain fuzzy entropy and the fourth right brain fuzzy entropy.

[0092] The left brain asymmetric entropy index and the right brain asymmetric entropy index were determined as the asymmetric entropy index.

[0093] Related calculation formulas and steps:

[0094] Autocorrelation feature extraction:

[0095] The calculation principle of the EEG signal (a nonlinear time series) x(k) (k = 1, 2, ..., N) is as follows:

[0096] (1) Discretize the time series (EEG signal) to obtain the mean-free summation sequence y(t).

[0097]

[0098] in, <x>represents the mean value of the entire EEG signal time series, t represents the value of k, N represents the maximum value of k, y(1)=x(1)- <x>;y(2)=x(1)- <x> +x(2)- <x>; ...and so on, we get the mean-devalued summation sequence y(t).

[0099] (2) The empirical mode decomposition (EMD) algorithm is used to decompose the mean-summed sequence y(t) into a series of intrinsic mode functions (IMFs) and a residual term.

[0100]

[0101] Among them, c i (t) is the i-th IMF, n is the number of IMFs, and r(t) is the residual term.

[0102] (3) Divide the mean-summed sequence y(t) into N non-overlapping s subsequences, and use the least squares method to fit the polynomial trend of each subsequence to obtain the fitting polynomial y v (t).

[0103] (4) Calculate c i (t) and y v (t) Pearson correlation coefficient between.

[0104]

[0105] Among them, p i Represents the Pearson correlation coefficient between and , cov() represents the calculated covariance, σ represents the standard deviation, μ represents the mean, and E() represents the expectation.

[0106] (5) The fitting trend term of the signal is expressed as:

[0107]

[0108] Combining formulas (2) and (4), the detrended signal is:

[0109]

[0110] (6) The detrended signal Y s (t) Divide into N non-overlapping s To avoid data loss, we divide the time series again in the reverse direction (the positive direction is from left to right, and the reverse direction is from right to left), and get a total of 2N s subsequences.

[0111] (7) Calculate the mean square error F of each subsequence 2 (v,s).

[0112] When v=1,2,…,N s When F 2 (v,s) is calculated as follows:

[0113]

[0114] When v=N s +1,N s +2,…,2N s When F 2 The solution of (v,s) is as shown in formula (7).

[0115]

[0116] Among them, v represents the subscript of each segment, s represents the size of each segment, and j represents the subscript of the time point within each segment.

[0117] (8) Calculate the order fluctuation function F of the entire sequence of all subsequences q (s).

[0118]

[0119] The order q represents different degrees of fluctuation, and the value of q is generally an arbitrary real number, for example, [-5, 0) ∪ (0, 5].

[0120] (9) Function F q (s) is a function of the scale s and order q of the time series, F q (s) and s have a high degree of stability, and if there is self-similarity, there is a law relationship between and:

[0121] F q (s)∝s h(q) (9)

[0122] Analysis F q The double logarithmic relationship between (s) and the time scale s is used to fit the generalized Hurst exponent h(q).

[0123] (10) For discrete time series, the scaling exponent τ(q) in the multifractal concept is:

[0124] τ(q)=qh(q)-1 (10)

[0125] By taking the derivative of both sides of the equation (10) with respect to q, we can obtain the singular index α and the multifractal spectrum f(α):

[0126] In summary, the implementation process of determining the autocorrelation characteristics of EEG signals is as follows: after the original EEG signals undergo data preprocessing (using the eeglab toolkit under MATLAB: importing data, locating electrodes, removing useless electrodes, re-referencing (whole brain average reference), filtering (0.5Hz-35Hz), RunICA, removing artifact components, and removing bad segments), a de-meaning and summation operation is performed; the EMD algorithm is used to calculate the IMFs of the de-meaning and summation sequence; and the Pearson correlation coefficient is then used to select the sum of the two terms with the largest correlation between each IMF modal component and the EEG signal fitting polynomial and the residual term, which is used as the EEG signal trend term and input into the MFDFA algorithm for calculation to obtain the autocorrelation characteristics of the EEG signal.

[0127] Complexity feature extraction:

[0128] Fuzzy entropy:

[0129] Fuzzy Entropy (FuzzyEn) is an improved algorithm proposed to address the discontinuous entropy values ​​obtained during the extraction of sample entropy and approximate entropy. By introducing a fuzzy membership function, an improved sample entropy, namely fuzzy entropy, is obtained. Fuzzy entropy has little dependence on the length of the time series and is robust to noise signals. The calculation steps are as follows:

[0130] (1) x(k) (k = 1, 2, ..., N) is a time series containing N sampling points (the original EEG signal), and the time series is reconstructed into an m-dimensional phase space to obtain a vector Defining Vectors and The maximum value of the difference between the corresponding elements is the distance between the two vectors; and the fuzzy membership function is introduced to define the similarity between the vector and , that is:

[0131]

[0132] Where i,j=1~Nm,i≠j,b represents the boundary gradient, which is 0.2;r represents the boundary width;b and r are known quantities; represents the distance between two vectors, corresponding to the maximum value of the difference between the two vector elements; n is the boundary gradient, which is 2; r is the boundary width, which is 0.2; n and r are known quantities.

[0133] (2) Define the first function

[0134]

[0135] m is the length of each reconstructed phase space, and the similarity D is calculated according to the above formula. ij m

[0136] (3) The fuzzy entropy value of a finite length time series is

[0137] F(m,n,r,N)=lnφ m (n,r)-lnφ m+1 (n,r) (14)

[0138] Hierarchical fuzzy entropy:

[0139] Hierarchical fuzzy entropy uses wavelet packet decomposition and reconstruction to decompose the original 35Hz EEG signal into four frequency bands: δ (1-4Hz), θ (4-8Hz), α (8-14Hz), and β (14-30Hz). Fuzzy entropy is then calculated for each frequency band. Hierarchical fuzzy entropy fully considers both the low- and high-frequency components of the signal, enabling it to extract more meaningful EEG dynamics information and better identify the complex characteristics of EEG signals.

[0140] Asymmetric entropy characteristics:

[0141] Dr. Sperry developed the "left-right division of labor theory" through his famous split-brain experiment, confirming the asymmetry between the left and right hemispheres. Many studies have also confirmed that the EEG signals of the left and right hemispheres are asymmetric under different conditions. Based on the asymmetric characteristics of left and right EEG signals and using hierarchical fuzzy entropy, this embodiment proposes asymmetric entropy characteristics for signals in each frequency band between symmetrical EEG leads, as shown in Equation (15):

[0142] ΔF=F L -F R (15)

[0143] Among them, F L and F R Represents the fuzzy entropy of EEG signals in each frequency band of the left and right brain, that is, F δ ,F θ ,F α ,F β . Asymmetric entropy index:

[0144] This example obtains the following abnormal EEG phenomena through analysis:

[0145] ① The left and right signals are obviously asymmetric.

[0146] ② It is mainly composed of δ waves, with a small amount of θ wave activity, or a small amount of low-amplitude fast waves in the δ or β frequency bands complexed on the slow waves.

[0147] ③θ is mainly a wave, with a small amount of δ, α, and β scattered waves.

[0148] The above phenomenon shows that there are obvious differences in the frequency of occurrence of different bands of abnormal EEG signals: the frequency band of wave and wave is significantly more than that of wave and wave. Combined with the hierarchical fuzzy entropy theory, this embodiment proposes an asymmetric entropy index of EEG signals, and the calculation formula is as shown in formula (16):

[0149]

[0150] In summary, the process of extracting EEG signal complexity features is as follows: after preprocessing the original EEG signal data (the same as the autocorrelation feature extraction part), wavelet packet decomposition and reconstruction are performed to obtain four frequency band components: δ, θ, α, and β. The fuzzy entropy of each frequency band component is calculated to obtain the hierarchical fuzzy entropy (the fuzzy entropy of each frequency band component is the hierarchical fuzzy entropy). Then, based on the asymmetric characteristics of EEG, the asymmetric entropy feature and the asymmetric entropy index are calculated. The hierarchical fuzzy entropy, asymmetric entropy feature, and asymmetric entropy index are fused to determine the complexity feature of the EEG signal.

[0151] Figure 2 This is a block diagram of a feature extraction system for EEG signals provided by an embodiment of the present invention. Figure 2 As shown, the feature extraction system of the EEG signal in this embodiment includes:

[0152] The EEG signal acquisition module 201 is used to acquire EEG signals.

[0153] The autocorrelation feature determination module 202 is used to determine the autocorrelation feature of the EEG signal by using an empirical mode decomposition algorithm optimized by the Pearson correlation coefficient and a multifractal detrended fluctuation analysis algorithm.

[0154] The complexity feature determination module 203 is used to determine the complexity features of the EEG signal; the complexity features include: hierarchical fuzzy entropy, asymmetric entropy features and asymmetric entropy index; the hierarchical fuzzy entropy includes the hierarchical fuzzy entropy of the left brain in different frequency bands and the hierarchical fuzzy entropy of the right brain in the corresponding frequency bands, the asymmetric entropy feature includes the difference between the hierarchical fuzzy entropy of the left brain in different frequency bands and the hierarchical fuzzy entropy of the right brain in the corresponding frequency bands, the asymmetric entropy index includes the left brain asymmetric entropy index and the right brain asymmetric entropy index, the left brain asymmetric entropy index is obtained based on the left brain hierarchical fuzzy entropy of all frequency bands, and the right brain asymmetric entropy index is obtained based on the right brain hierarchical fuzzy entropy of all frequency bands.

[0155] The feature determination module 204 is used to determine the autocorrelation feature and the complexity feature as the features of the EEG signal, and the features of the EEG signal are used to classify the EEG signal as fatigue driving.

[0156] As an optional implementation, the autocorrelation feature determination module 202 specifically includes:

[0157] The mean-removed summing unit is used to perform mean-removed summing on the EEG signal to obtain a mean-removed summed signal.

[0158] The empirical mode decomposition unit is used to decompose the mean-decomposed sum signal into a residual term and multiple intrinsic mode functions by using an empirical mode decomposition algorithm optimized by the Pearson correlation coefficient.

[0159] The fitting polynomial determination unit is used to fit the mean-removed summation signal using the least squares method to obtain a fitting polynomial.

[0160] The correlation coefficient calculation unit is used to calculate the Pearson correlation coefficient between each intrinsic mode function and the fitting polynomial.

[0161] The sorting unit is used to sort all Pearson correlation coefficients from large to small to obtain a correlation coefficient sequence.

[0162] The sub-correlation coefficient determining unit is used to determine the first-ranked Pearson correlation coefficient in the correlation coefficient sequence as the first correlation coefficient, and to determine the second-ranked Pearson correlation coefficient in the correlation coefficient sequence as the second correlation coefficient.

[0163] The intrinsic mode function determination unit is used to determine a first intrinsic mode function and a second intrinsic mode function; the first intrinsic mode function is the intrinsic mode function corresponding to the first correlation coefficient, and the second intrinsic mode function is the intrinsic mode function corresponding to the second correlation coefficient.

[0164] The autocorrelation feature determination unit is used to obtain the autocorrelation feature of the EEG signal according to the residual term, the first intrinsic mode function and the second intrinsic mode function using a multifractal detrended fluctuation analysis algorithm.

[0165] As an optional implementation, the complexity feature determination module 203 specifically includes:

[0166] The left brain frequency band signal acquisition unit is used to perform wavelet packet decomposition and reconstruction on the left brain signal in the EEG signal to obtain left brain frequency band signals in four frequency bands.

[0167] The right brain frequency band signal acquisition unit is used to perform wavelet packet decomposition and reconstruction on the right brain signal in the EEG signal to obtain right brain frequency band signals in four frequency bands.

[0168] The left brain hierarchical fuzzy entropy acquisition unit is used to calculate the fuzzy entropy of the left brain sub-band signal in each frequency band to obtain the left brain hierarchical fuzzy entropies of four frequency bands; the left brain hierarchical fuzzy entropies of the four frequency bands are: the first left brain fuzzy entropy, the second left brain fuzzy entropy, the third left brain fuzzy entropy and the fourth left brain fuzzy entropy.

[0169] The right brain hierarchical fuzzy entropy acquisition unit is used to calculate the fuzzy entropy of the right brain sub-band signal in each frequency band to obtain the right brain hierarchical fuzzy entropies of four frequency bands; the right brain hierarchical fuzzy entropies of the four frequency bands are: the first right brain fuzzy entropy, the second right brain fuzzy entropy, the third right brain fuzzy entropy and the fourth right brain fuzzy entropy.

[0170] The asymmetric entropy calculation unit is used to calculate an asymmetric entropy feature and an asymmetric entropy index according to the left brain hierarchical fuzzy entropy of four frequency bands and the right brain hierarchical fuzzy entropy of four frequency bands.

[0171] The complexity feature determination unit is used to determine the left brain hierarchical fuzzy entropy, the right brain hierarchical fuzzy entropy, the asymmetric entropy feature and the asymmetric entropy index as the complexity features of the EEG signal.

[0172] As an optional implementation, the asymmetric entropy calculation unit specifically includes:

[0173] The fuzzy entropy feature calculation subunit is used to determine a first asymmetric fuzzy entropy feature according to the first left brain fuzzy entropy and the first right brain fuzzy entropy, determine a second asymmetric fuzzy entropy feature according to the second left brain fuzzy entropy and the second right brain fuzzy entropy, determine a third asymmetric fuzzy entropy feature according to the third left brain fuzzy entropy and the third right brain fuzzy entropy, and determine a fourth asymmetric fuzzy entropy feature according to the fourth left brain fuzzy entropy and the fourth right brain fuzzy entropy.

[0174] The fuzzy entropy feature determination subunit is configured to determine the first asymmetric fuzzy entropy feature, the second asymmetric fuzzy entropy feature, the third asymmetric fuzzy entropy feature, and the fourth asymmetric fuzzy entropy feature as asymmetric entropy features.

[0175] The fuzzy entropy index calculation subunit is used to calculate the left brain asymmetric entropy index according to the first left brain fuzzy entropy, the second left brain fuzzy entropy, the third left brain fuzzy entropy and the fourth left brain fuzzy entropy, and calculate the right brain asymmetric entropy index according to the first right brain fuzzy entropy, the second right brain fuzzy entropy, the third right brain fuzzy entropy and the fourth right brain fuzzy entropy.

[0176] The fuzzy entropy index determining subunit is used to determine the left brain asymmetric entropy index and the right brain asymmetric entropy index as an asymmetric entropy index.

[0177] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0178] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the device and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.< / x> < / x> < / x> < / x>

Claims

1. A method for extracting features from an EEG signal, characterized in that: The method comprises: Acquire EEG signals; The autocorrelation characteristics of the EEG signal are determined using an empirical mode decomposition algorithm optimized by Pearson correlation coefficient and a multifractal detrended fluctuation analysis algorithm; Determine the complexity characteristics of the EEG signal; the complexity characteristics include: hierarchical fuzzy entropy, asymmetric entropy characteristics and asymmetric entropy index; the hierarchical fuzzy entropy includes the hierarchical fuzzy entropy of the left brain in different frequency bands and the hierarchical fuzzy entropy of the right brain in the corresponding frequency bands, the asymmetric entropy characteristics include the difference between the hierarchical fuzzy entropy of the left brain in different frequency bands and the hierarchical fuzzy entropy of the right brain in the corresponding frequency bands, the asymmetric entropy index includes the left brain asymmetric entropy index and the right brain asymmetric entropy index, the left brain asymmetric entropy index is obtained based on the hierarchical fuzzy entropy of the left brain in all frequency bands, and the right brain asymmetric entropy index is obtained based on the hierarchical fuzzy entropy of the right brain in all frequency bands; determining the autocorrelation feature and the complexity feature as features of the EEG signal, wherein the features of the EEG signal are used to classify the EEG signal as fatigue driving; The method of determining the autocorrelation characteristics of the EEG signal using the empirical mode decomposition algorithm optimized by the Pearson correlation coefficient and the multifractal detrended fluctuation analysis algorithm specifically includes: Performing mean-summing on the EEG signal to obtain a mean-summing signal; Decomposing the mean-decomposed sum signal into a residual term and a plurality of intrinsic mode functions using an empirical mode decomposition algorithm; Fitting the de-meaned summation signal using the least squares method to obtain a fitting polynomial; Calculating the Pearson correlation coefficient between each of the intrinsic mode functions and the fitting polynomial; Sort all Pearson correlation coefficients from large to small to obtain a correlation coefficient sequence; Determine the first-ranked Pearson correlation coefficient in the correlation coefficient sequence as a first correlation coefficient, and determine the second-ranked Pearson correlation coefficient in the correlation coefficient sequence as a second correlation coefficient; Determining a first intrinsic mode function and a second intrinsic mode function; the first intrinsic mode function is the intrinsic mode function corresponding to the first correlation coefficient, and the second intrinsic mode function is the intrinsic mode function corresponding to the second correlation coefficient; Obtaining an autocorrelation feature of the EEG signal based on the residual term, the first intrinsic mode function, and the second intrinsic mode function using the multifractal detrended fluctuation analysis algorithm; The calculation formula of the asymmetric entropy index is: Among them, FAI is the asymmetric entropy index; F δ is the fuzzy entropy of the EEG signal in the δ wave frequency band; F θ is the fuzzy entropy of the EEG signal in the θ wave band; F α is the fuzzy entropy of the EEG signal in the α wave band; F β is the fuzzy entropy of the EEG signal in the β wave band.

2. The method for extracting features from an EEG signal according to claim 1, wherein: Determining the complexity feature of the EEG signal specifically includes: Performing wavelet packet decomposition and reconstruction on the left brain signal in the EEG signal to obtain left brain frequency band signals in four frequency bands; Performing wavelet packet decomposition and reconstruction on the right brain signal in the EEG signal to obtain right brain sub-frequency band signals in four frequency bands; Calculate the fuzzy entropy of the left brain sub-band signal in each frequency band to obtain the left brain hierarchical fuzzy entropy of the four frequency bands; the left brain hierarchical fuzzy entropies of the four frequency bands are: the first left brain fuzzy entropy, the second left brain fuzzy entropy, the third left brain fuzzy entropy and the fourth left brain fuzzy entropy; Calculate the fuzzy entropy of the right brain sub-band signal in each frequency band to obtain the right brain hierarchical fuzzy entropy of the four frequency bands; the right brain hierarchical fuzzy entropy of the four frequency bands are: the first right brain fuzzy entropy, the second right brain fuzzy entropy, the third right brain fuzzy entropy and the fourth right brain fuzzy entropy; Calculating the asymmetric entropy feature and the asymmetric entropy index according to the left brain hierarchical fuzzy entropy of four frequency bands and the right brain hierarchical fuzzy entropy of four frequency bands; The left brain hierarchical fuzzy entropy, the right brain hierarchical fuzzy entropy, the asymmetric entropy feature and the asymmetric entropy index are determined as complexity features of the electroencephalogram signal.

3. The method for extracting features from an EEG signal according to claim 2, wherein: The calculating of the asymmetric entropy feature and the asymmetric entropy index according to the left brain hierarchical fuzzy entropy of four frequency bands and the right brain hierarchical fuzzy entropy of four frequency bands specifically includes: Determine a first asymmetric fuzzy entropy feature based on the first left-brain fuzzy entropy and the first right-brain fuzzy entropy, determine a second asymmetric fuzzy entropy feature based on the second left-brain fuzzy entropy and the second right-brain fuzzy entropy, determine a third asymmetric fuzzy entropy feature based on the third left-brain fuzzy entropy and the third right-brain fuzzy entropy, and determine a fourth asymmetric fuzzy entropy feature based on the fourth left-brain fuzzy entropy and the fourth right-brain fuzzy entropy; determining the first asymmetric fuzzy entropy feature, the second asymmetric fuzzy entropy feature, the third asymmetric fuzzy entropy feature, and the fourth asymmetric fuzzy entropy feature as the asymmetric entropy features; Calculating the left brain asymmetric entropy index according to the first left brain fuzzy entropy, the second left brain fuzzy entropy, the third left brain fuzzy entropy, and the fourth left brain fuzzy entropy; calculating the right brain asymmetric entropy index according to the first right brain fuzzy entropy, the second right brain fuzzy entropy, the third right brain fuzzy entropy, and the fourth right brain fuzzy entropy; The left brain asymmetric entropy index and the right brain asymmetric entropy index are determined as asymmetric entropy indices.

4. A feature extraction system for EEG signals, characterized in that: include: An EEG signal acquisition module, used for acquiring EEG signals; An autocorrelation feature determination module is used to determine the autocorrelation feature of the EEG signal using an empirical mode decomposition algorithm optimized by Pearson correlation coefficient and a multifractal detrended fluctuation analysis algorithm; a complexity feature determination module, configured to determine the complexity features of the EEG signal; the complexity features comprising: hierarchical fuzzy entropy, asymmetric entropy features, and an asymmetric entropy index; the hierarchical fuzzy entropy comprising the hierarchical fuzzy entropy of the left brain in different frequency bands and the hierarchical fuzzy entropy of the right brain in corresponding frequency bands; the asymmetric entropy feature comprising the difference between the hierarchical fuzzy entropy of the left brain in different frequency bands and the hierarchical fuzzy entropy of the right brain in corresponding frequency bands; the asymmetric entropy index comprising the left brain asymmetric entropy index and the right brain asymmetric entropy index; the left brain asymmetric entropy index being obtained based on the hierarchical fuzzy entropy of the left brain in all frequency bands; and the right brain asymmetric entropy index being obtained based on the hierarchical fuzzy entropy of the right brain in all frequency bands; a feature determination module, configured to determine the autocorrelation feature and the complexity feature as features of the EEG signal, wherein the features of the EEG signal are used to classify the EEG signal as fatigue driving; The autocorrelation feature determination module specifically includes: a de-meaning summing unit, configured to perform de-meaning summing on the EEG signal to obtain a de-meaning summing signal; An empirical mode decomposition unit, configured to decompose the mean-decomposed sum signal into a residual term and a plurality of intrinsic mode functions using an empirical mode decomposition algorithm; A fitting polynomial determination unit, configured to fit the de-meaned summation signal using a least squares method to obtain a fitting polynomial; a correlation coefficient calculation unit, configured to calculate the Pearson correlation coefficient between each of the intrinsic mode functions and the fitting polynomial; The sorting unit is used to sort all Pearson correlation coefficients from large to small to obtain a correlation coefficient sequence; a sub-correlation coefficient determining unit, configured to determine the first-ranked Pearson correlation coefficient in the correlation coefficient sequence as a first correlation coefficient, and determine the second-ranked Pearson correlation coefficient in the correlation coefficient sequence as a second correlation coefficient; an intrinsic mode function determining unit, configured to determine a first intrinsic mode function and a second intrinsic mode function; the first intrinsic mode function is the intrinsic mode function corresponding to the first correlation coefficient, and the second intrinsic mode function is the intrinsic mode function corresponding to the second correlation coefficient; an autocorrelation feature determining unit, configured to obtain the autocorrelation feature of the EEG signal according to the residual term, the first intrinsic mode function, and the second intrinsic mode function using the multifractal detrended fluctuation analysis algorithm; The calculation formula of the asymmetric entropy index is: Among them, FAI is the asymmetric entropy index; F δ is the fuzzy entropy of the EEG signal in the δ wave frequency band; F θ is the fuzzy entropy of the EEG signal in the θ wave band; F α is the fuzzy entropy of the EEG signal in the α wave band; F β is the fuzzy entropy of the EEG signal in the β wave band.

5. The feature extraction system of EEG signal according to claim 4, characterized in that: The complexity feature determination module specifically includes: a left brain frequency band signal acquisition unit, configured to perform wavelet packet decomposition and reconstruction on the left brain signal in the EEG signal to obtain left brain frequency band signals in four frequency bands; a right brain frequency band signal acquisition unit, configured to perform wavelet packet decomposition and reconstruction on the right brain signal in the EEG signal to obtain right brain frequency band signals in four frequency bands; The left brain hierarchical fuzzy entropy acquisition unit is used to calculate the fuzzy entropy of the left brain sub-band signal in each frequency band to obtain the left brain hierarchical fuzzy entropies of the four frequency bands; the left brain hierarchical fuzzy entropies of the four frequency bands are: a first left brain fuzzy entropy, a second left brain fuzzy entropy, a third left brain fuzzy entropy and a fourth left brain fuzzy entropy; The right brain hierarchical fuzzy entropy acquisition unit is used to calculate the fuzzy entropy of the right brain sub-band signal in each frequency band to obtain the right brain hierarchical fuzzy entropies of four frequency bands; the right brain hierarchical fuzzy entropies of the four frequency bands are: a first right brain fuzzy entropy, a second right brain fuzzy entropy, a third right brain fuzzy entropy and a fourth right brain fuzzy entropy; an asymmetric entropy calculation unit, configured to calculate the asymmetric entropy feature and the asymmetric entropy index according to the left brain hierarchical fuzzy entropies of four frequency bands and the right brain hierarchical fuzzy entropies of four frequency bands; The complexity feature determination unit is used to determine the left brain hierarchical fuzzy entropy, the right brain hierarchical fuzzy entropy, the asymmetric entropy feature and the asymmetric entropy index as the complexity feature of the electroencephalogram signal.

6. The feature extraction system of EEG signal according to claim 5, characterized in that: The asymmetric entropy calculation unit specifically includes: a fuzzy entropy feature calculation subunit, configured to determine a first asymmetric fuzzy entropy feature according to the first left-brain fuzzy entropy and the first right-brain fuzzy entropy, determine a second asymmetric fuzzy entropy feature according to the second left-brain fuzzy entropy and the second right-brain fuzzy entropy, determine a third asymmetric fuzzy entropy feature according to the third left-brain fuzzy entropy and the third right-brain fuzzy entropy, and determine a fourth asymmetric fuzzy entropy feature according to the fourth left-brain fuzzy entropy and the fourth right-brain fuzzy entropy; a fuzzy entropy feature determining subunit, configured to determine the first asymmetric fuzzy entropy feature, the second asymmetric fuzzy entropy feature, the third asymmetric fuzzy entropy feature, and the fourth asymmetric fuzzy entropy feature as the asymmetric entropy feature; a fuzzy entropy index calculation subunit, configured to calculate the left brain asymmetric entropy index according to the first left brain fuzzy entropy, the second left brain fuzzy entropy, the third left brain fuzzy entropy, and the fourth left brain fuzzy entropy, and calculate the right brain asymmetric entropy index according to the first right brain fuzzy entropy, the second right brain fuzzy entropy, the third right brain fuzzy entropy, and the fourth right brain fuzzy entropy; The fuzzy entropy index determining subunit is configured to determine the left brain asymmetric entropy index and the right brain asymmetric entropy index as asymmetric entropy indices.

Images