A method for evaluating the hand function of stroke patients

Abstract

The present invention discloses a method for evaluating the hand function of stroke patients, which relates to the technical field of bioelectrical signal processing and specifically includes the following steps: signal acquisition and preprocessing; decomposition of brain and myoelectric signals; extraction of phase information using Hilbert transform; estimation of modulation index by phase-amplitude coupling; and upper limb rehabilitation grade assessment. The technical solution of the present invention proposes a method for evaluating the hand rehabilitation of stroke patients based on brain-muscle electrical signals through empirical mode decomposition and tensor-based phase-amplitude coupling-multiscale conditional mutual information, realizes the quantification of the rehabilitation degree, and combines with the stroke clinical assessment scale to quantitatively analyze the rehabilitation status of patients.

Description

Technical Field

[0001] The present invention relates to the technical field of bioelectrical signal processing, and particularly to a method for evaluating the hand function of stroke patients. Background Art

[0002] Stroke is a brain tissue injury caused by the interruption or abnormality of blood supply to the brain, usually divided into ischemic stroke (cerebral infarction) and hemorrhagic stroke (cerebral hemorrhage). Stroke not only brings great physical and psychological burdens to patients, but also brings huge economic pressure and nursing challenges to society and families. Clinical data shows that the recurrence rate of acute stroke patients is very high, reaching 17.7%, and the five-year cumulative recurrence rate exceeds 30%.

[0003] The rehabilitation process of stroke is complex and long-term, usually starting from after the acute-phase treatment and lasting through all stages of the patient's functional recovery. The goal of rehabilitation is to restore the patient's function to the greatest extent, improve the quality of life, and help the patient live as independently as possible. Nevertheless, there are still many challenges in the rehabilitation of stroke patients, among which the most prominent problems include the lack of quantitative means and timely feedback in the evaluation of the rehabilitation process. Clinically, sleep physicians and rehabilitation physicians mostly rely on empirical observation to judge the rehabilitation effect of patients, lacking scientific and standardized rehabilitation progress evaluation tools. At the same time, the rehabilitation effects of stroke patients with different etiologies vary greatly, and the rehabilitation progress of some patients is slow or fails to achieve the expected effect, which makes the overall rehabilitation management particularly complex.

[0004] Therefore, for the rehabilitation process of stroke patients, especially a portable monitoring system that can effectively evaluate and record the neuromuscular electrical signals of patients, is particularly important. The acquisition and analysis of physiological signals such as electroencephalogram (EEG), electromyogram (EMG), and electrooculogram (EOG) can not only provide real-time monitoring data for the rehabilitation of stroke patients, but also help medical staff comprehensively understand the physiological state of patients. In particular, the interaction between electroencephalogram signals and electromyogram signals (such as phase-amplitude coupling, PAC) provides a unique perspective for evaluating the transmission of neurocontrol information between the cerebral cortex and limb muscles.

[0005] Phase-Amplitude Coupling (PAC), as an important phenomenon in neural information transmission, is significantly enhanced in cognitive functions such as perception, learning, and memory. Research has shown that changes in PAC are closely related to the onset and clinical symptoms of some neurological diseases (such as Parkinson's disease, epilepsy, etc.). For stroke patients, changes in PAC can reflect the changes in the transmission of control signals between the cerebral cortex and the muscle system, thus providing a scientific basis for rehabilitation assessment. Therefore, an evaluation algorithm based on the coupling of electroencephalogram and electromyogram signals is expected to become an effective tool to help doctors accurately evaluate the rehabilitation status of patients, monitor the rehabilitation process in real time, and even provide autonomous rehabilitation means for patients without doctor supervision after discharge. Summary of the Invention

[0006] The technical solution of the present invention to solve the above technical problems is to provide a method for evaluating the hand function of stroke patients, including the following steps:

[0007] Step 1, signal acquisition and preprocessing: Set an action paradigm in combination with the Brunnstrom assessment scale, and simultaneously collect synchronous electroencephalogram signals and electromyogram signals of stroke patients at different levels, and preprocess the electroencephalogram signals and electromyogram signals respectively;

[0008] Step 2, decomposition of brain and electromyogram signals: Based on wavelet packet transform, synchronously decompose the preprocessed brain and electromyogram signals, including constructing a wavelet packet function and selecting a wavelet packet basis function, performing a decomposition operation on the original signal, calculating wavelet coefficients at each level using a multi-resolution decomposition algorithm, obtaining sub-band signals of the original signal in each frequency according to the wavelet coefficients, reconstructing the wavelet approximation coefficients and wavelet coefficients using a multi-resolution reconstruction algorithm, separating the reconstructed signals according to the active bands of the electroencephalogram and electromyogram, and saving the individual frequency domain signals;

[0009] Step 3, use Hilbert transform to extract phase information: Convert the decomposed brain and electromyogram signals into time series, use the discrete Hilbert transform to convert the time series into complex signals, obtain the instantaneous phase by calculating the phase angle of the complex signal, calculate the instantaneous amplitude, and perform phase verification;

[0010] Step 4, adopt phase-amplitude coupling to estimate the modulation index: Cut the phase φ(n) into n slices, the phase is assigned to n intervals, and each interval is 360 / n°; Take the average value of the amplitude a(t) in each delimited interval, denoted by <a(t) φ to represent, divide the amplitude in each interval by the sum of the intervals to obtain the probability distribution P; Use the Kullback-Leibler distance to obtain the modulation index;

[0011] Step 5, upper limb rehabilitation level assessment: Based on the Spearman correlation coefficient, map the fitting relationship between the MI of each patient and the Brunnstrom assessment scale, analyze the FCMC characteristic indicators and change trends that are significantly correlated with the levels of the Brunnstrom assessment scale, and quantitatively assess the disease levels of stroke patients.

[0012] Further, in Step 1, collect the electroencephalogram (EEG) signals of stroke patients at the leads Fp1, Fp2, F3, F4, C3, C4, Cz, P3, P4, A1, and A2, and collect the electromyogram (EMG) signals at the anterior deltoid, middle deltoid, biceps brachii, triceps brachii, and extensor digitorum of the forearm.

[0013] The preprocessing operations of the EEG signals include: rereferencing, baseline correction, filtering, removing electrooculogram, and EEG data segmentation; the preprocessing operations of the EMG signals include: baseline correction, filtering, and EMG data segmentation.

[0014] Further, in Step 2, wavelet transform performs multi-resolution analysis on the signal through the scaling function φ(t) and the wavelet function ψ(t); satisfies the orthogonality condition, and generates wavelet basis functions through scaling and translation:

[0015] φ j,k (t) = 2 j / 2 φ(2 j t - k), ψ j,k (t) = 2 j / 2 ψ(2 j t - k);

[0016] where j is the scale parameter and k is the translation parameter;

[0017] At the j-th layer, the wavelet packet basis function is defined as:

[0018]

[0019] The wavelet packet function of the n-th node in the j-th layer;

[0020] h k (t): low-pass filter coefficient;

[0021] g k (t): high-pass filter coefficient;

[0022] Gradually disassemble the original signal using the decomposition algorithm, compare it with the wavelet function. Assume the original function is F, and gradually decompose the original function into F1, F2, F3, F4...; there is F1 = FJ - 1 + WJ - 1 with the wavelet part W1, W2, W3, W4..., and decompose to the 0-th level;

[0023] The signals of each layer are used to calculate the decomposition coefficients through low-pass and high-pass filters:

[0024]

[0025] The signal value of the nth node in the jth layer;

[0026] h[m], g[m]: Low-pass and high-pass filter coefficients;

[0027] The wavelet coefficients of each level are calculated using the multi-resolution decomposition algorithm;

[0028] Based on the wavelet coefficients, the sub-band signals of the original signal in each frequency are obtained;

[0029] The multi-resolution reconstruction algorithm is used to reconstruct the wavelet approximation coefficients and wavelet coefficients in the form of discrete filters; The reconstruction of the signal uses the linear combination of wavelet packet coefficients:

[0030]

[0031] Reconstruction is performed layer by layer through the inverse filter;

[0032]

[0033] The reconstructed signal is separated according to the active bands of electroencephalogram and electromyogram; And the electromyogram signal is decomposed into the same frequency bands;

[0034] The individual frequency domain signals are saved.

[0035] Furthermore, in step 3, the decomposed brain and electromyogram signals are transformed into time series;

[0036] The time series is transformed into a complex signal using the discrete Hilbert transform;

[0037]

[0038] The instantaneous phase is obtained by calculating the phase angle of the complex signal;

[0039] φ(t) = arg(z(t));

[0040] The instantaneous amplitude is obtained by calculating the formula;

[0041]

[0042] Compensation is performed between the phase differences;

[0043] φ unwrapped (n) = φ unwrapped (n - 1) + Δφ(n);

[0044] Wherein:

[0045] Δφ(n) = φ unwrapped (n) - φ unwrapped (n - 1);

[0046] To ensure phase continuity, adjustment is required;

[0047]

[0048] The finally unwrapped phase is:

[0049] φ unwrapped (n) = φ unwrapped (n - 1) + Δφ adjusted (n);

[0050] Perform phase verification.

[0051] The technical solution of the present invention proposes a method for evaluating the hand rehabilitation of stroke patients based on electroencephalogram and electromyogram signals through empirical mode decomposition and tensor-based phase-amplitude coupling - multi-scale conditional mutual information, realizing the quantification of the rehabilitation degree. Combining with the stroke clinical assessment scale, it has a certain significance for quantitatively analyzing the rehabilitation status of patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on the structures shown in these drawings.

[0053] Figure 1 It is a flowchart of the method for evaluating the hand function of stroke patients described in the present invention;

[0054] Figure 2 It is a schematic diagram of the action paradigm of the present invention;

[0055] Figure 3 It is a flowchart of the execution of the action paradigm of the present invention;

[0056] Figure 4 It is a schematic diagram of the electroencephalogram leads used in the present invention;

[0057] Figure 5 It is a schematic diagram of the electromyogram tissue used in the present invention;

[0058] Figure 6 It is a step flowchart of the method for evaluating the hand function of stroke patients described in the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0059] The present invention proposes a method for evaluating the hand function of stroke patients, aiming to propose a rehabilitation assessment method for stroke patients based on electroencephalogram (EEG) and electromyogram (EMG) signals.

[0060] The method for evaluating the hand function of stroke patients proposed by the present invention will be described in the following specific embodiments:

[0061] In the technical solution of this embodiment, as Figure 6 shown, a method for evaluating the hand function of stroke patients includes the following steps:

[0062] Step 1, signal acquisition and preprocessing: Set the movement paradigm in combination with the Brunnstrom assessment scale, and simultaneously collect the synchronous EEG signals and EMG signals of stroke patients at different levels. Then, preprocess the EEG signals and EMG signals respectively.

[0063] Step 2, decomposition of EEG and EMG signals: Synchronously decompose the preprocessed EEG and EMG signals based on wavelet packet transform, including constructing a wavelet packet function and selecting a wavelet packet basis function, performing a decomposition operation on the original signal, calculating the wavelet coefficients at each level using a multi-resolution decomposition algorithm, obtaining the sub-band signals of the original signal in each frequency according to the wavelet coefficients, reconstructing the wavelet approximation coefficients and wavelet coefficients using a multi-resolution reconstruction algorithm, separating the reconstructed signals according to the active frequency bands of EEG and EMG, and saving the individual frequency-domain signals.

[0064] Step 3, extracting phase information using Hilbert transform: Convert the decomposed EEG and EMG signals into time series, use the discrete Hilbert transform to convert the time series into complex signals, calculate the instantaneous phase by calculating the phase angle of the complex signals, calculate the instantaneous amplitude, and perform phase verification.

[0065] Step 4, estimating the modulation index using phase-amplitude coupling: Divide the phase φ(n) into n slices, and the phase is assigned to n intervals, each interval being 360 / n°. Take the average value of the amplitude a(t) in each delimited interval, denoted by <a(t)> φ , and obtain the probability distribution P by dividing the amplitude in each interval by the sum of the intervals. Use the Kullback-Leibler distance to obtain the modulation index.

[0066] Step 5, upper limb rehabilitation level assessment: Map the fitting relationship between the MI of each patient and the Brunnstrom assessment scale based on the Spearman correlation coefficient, analyze the FCMC characteristic indexes and their changing trends that are significantly correlated with the levels of the Brunnstrom assessment scale, and quantitatively assess the disease levels of stroke patients.

[0067] Further, in step 1, electroencephalogram (EEG) signals of stroke patients are collected at leads Fp1, Fp2, F3, F4, C3, C4, Cz, P3, P4, A1, and A2, and electromyogram (EMG) signals of the anterior deltoid, middle deltoid, biceps brachii, triceps brachii, and extensor digitorum of the forearm are collected.

[0068] The preprocessing operations of EEG signals include: rereferencing, baseline correction, filtering, electrooculogram removal, and EEG data segmentation; the preprocessing operations of EMG signals include: baseline correction, filtering, and EMG data segmentation.

[0069] Further, in step 2, wavelet transform performs multi-resolution analysis on the signal through the scaling function φ(t) and the wavelet function ψ(t); satisfies the orthogonality condition, and generates wavelet basis functions through scaling and translation:

[0070] φ j,k (t) = 2 j / 2 φ(2 j t - k), ψ j,k (t) = 2 j / 2 ψ(2 j t - k);

[0071] where j is the scale parameter and k is the translation parameter;

[0072] At the j-th layer, the wavelet packet basis function is defined as:

[0073]

[0074] The wavelet packet function of the n-th node in the j-th layer;

[0075] h k (t): low-pass filter coefficient;

[0076] g k (t): high-pass filter coefficient;

[0077] The original signal is gradually disassembled using the decomposition algorithm and compared with the wavelet function. Assuming the original function is F, the original function is decomposed into F1, F2, F3, F4... step by step; compared with the wavelet parts W1, W2, W3, W4..., there is F1 = FJ - 1 + WJ - 1, and it is decomposed to the 0-th level;

[0078] The signals of each layer are used to calculate the decomposition coefficients through low-pass and high-pass filters:

[0079]

[0080] The signal value of the n-th node in the j-th layer;

[0081] h[m], g[m]: Low-pass and high-pass filter coefficients;

[0082] Calculate wavelet coefficients at each level using a multi-resolution decomposition algorithm;

[0083] Derive sub-band signals of the original signal in each frequency from the wavelet coefficients;

[0084] Use a multi-resolution reconstruction algorithm to reconstruct the wavelet approximation coefficients and wavelet coefficients in the form of discrete filter implementation; The reconstruction of the signal uses a linear combination of wavelet packet coefficients:

[0085]

[0086] Perform layer-by-layer reconstruction through the inverse filters h k and g k ;

[0087]

[0088] Separate the reconstructed signal according to the active frequency bands of electroencephalogram and electromyogram; And decompose the electromyogram signal into the same frequency bands;

[0089] Save the individual frequency domain signals.

[0090] Furthermore, in step 3, convert the decomposed brain and electromyogram signals into time series;

[0091] Use the discrete Hilbert transform to convert the time series into a complex signal;

[0092]

[0093] Obtain the instantaneous phase by calculating the phase angle of the complex signal;

[0094] φ(t) = arg(z(t));

[0095] Calculate the instantaneous amplitude through the formula;

[0096]

[0097] Perform compensation between the phase differences;

[0098] φ unwrapped (n) = φ unwrapped (n - 1) + Δφ(n);

[0099] Where:

[0100] Δφ(n) = φ unwrapped (n) - φ unwrapped (n - 1);

[0101] To ensure phase continuity, it is necessary to adjust Δφ(n);

[0102]

[0103] The finally unwrapped phase is:

[0104] φ unwrapped (n) = φ unwrapped (n - 1)+Δφ adjusted (n);

[0105] Perform phase verification.

[0106] Example 1:

[0107] A method for upper limb rehabilitation assessment of stroke patients, the specific process is as Figure 1 shown, including the following steps:

[0108] Step 1: Signal acquisition and preprocessing. Under the guidance of a clinician, setting the flexion of the upper arm as the action paradigm in combination with the Brunnstrom assessment scale, collect the synchronous electroencephalogram signals and surface electromyogram signals of several stroke patients when performing the forward flexion of the arm, ensuring that there are at least 5 subjects in each level of Brunnstrom among several stroke subjects.

[0109] In the above-mentioned action paradigm, it includes: two actions of upper arm flexion and arm opening, as Figure 2 shown. When the subject performs the upper arm flexion, place a prompt in front of the subject to give the subject picture and sound prompts. During the signal acquisition process, the subject needs to sit upright on the chair, with the arm hanging naturally, keeping the body stable, and reducing unnecessary movements, including but not limited to: head turning, swallowing, blinking, etc.; as Figure 3 shown as the specific action execution process. The 0 - 10s is the preparation stage, giving the subject voice prompts; during the action execution stage, the subject needs to slowly straighten the arm forward until it is parallel to the ground, and then slowly lift the forearm until it is parallel to the body and keep it stable. The action execution stage lasts for 16s; during the arm opening stage, the subject needs to slowly open the arm to both sides of the body until the arm is fully opened and parallel to the ground. The arm opening stage lasts for 16s. Each subject needs to complete 5 test group actions. One test group action contains 5 upper arm flexion actions and 5 arm opening actions. After the subject completes 1 test group action, they will get a 10 - minute rest to prevent the subject from experiencing mental and muscle fatigue.

[0110] During the acquisition of the EEG, in strict accordance with the international 10-20 system, the subject needs to complete scalp cleaning in advance, correctly wear the EEG cap, select the central area lead Cz as the reference point. To avoid distortion of the collected EEG signals, the impedance between the electrode and the scalp must be less than 5 kΩ, and then the EEG signals are collected. As Figure 4 shown, the EEGs at the Fp1, Fp2, F3, F4, C3, C4, Cz, P3, P4, A1, and A2 leads are collected respectively.

[0111] During the acquisition of the sEMG, the skin of the subject's upper arm is cleaned with alcohol, and if necessary, the local skin of the subject is polished with a scrub to reduce the skin impedance. As Figure 5 shown, the sEMGs at the anterior deltoid, middle deltoid, biceps brachii, triceps brachii, and extensor digitorum of the forearm are collected respectively.

[0112] The preprocessing operations of the EEG include but are not limited to rereferencing, baseline correction, filtering, electrooculogram removal, and EEG data segmentation, etc. In the rereferencing operation, A1 and A2 are selected as the new reference points; in the baseline correction operation, the baseline drift of the EEG is reduced by the method of removing the mean value; in the filtering operation shown, the EEG is subjected to band-pass filtering of 0.1 - 100 Hz and notch filtering of 50 Hz; in the electrooculogram removal operation, the independent component analysis (ICA) algorithm is used to remove the electrooculogram artifacts of the subject; in the EEG data segmentation operation, based on the marks in the EEG acquisition system, the time range of the action execution stage is determined, and the corresponding data segments are extracted.

[0113] The preprocessing operations of the sEMG include but are not limited to baseline correction, filtering, and EMG data segmentation, etc. In the baseline correction operation shown, the baseline drift of the sEMG is reduced by the method of taking the mean value; in the filtering operation, the sEMG is subjected to band-pass filtering of 0.1 - 100 Hz and notch filtering of 50 Hz; in the EMG data segmentation operation, based on the marks in the EMG acquisition system, the time range of the action execution stage is determined, and the corresponding data segments are extracted.

[0114] Step 2: Decomposition of the brain-muscle electrical signals. In this embodiment, wavelet transform is used to achieve synchronous decomposition of the brain-muscle electrical signals, which specifically includes the following contents:

[0115] 2.1 Construct a wavelet packet function and select a wavelet packet basis function:

[0116] Wavelet transform performs multi-resolution analysis (MRA) on signals through the scaling function φ(t) and the wavelet function ψ(t).

[0117] Scaling function: Represents the low-frequency components of the signal.

[0118] Wavelet function: Represents the high-frequency components of the signal.

[0119] They satisfy the orthogonality condition and generate wavelet basis functions through scaling and translation:

[0120] φ j,k (t) = 2 j / 2 φ(2 j t - k), ψ j,k (t) = 2 j / 2 ψ(2 j t - k);

[0121] where j is the scale parameter and k is the translation parameter

[0122] At the j-th level, the wavelet packet basis functions are defined as:

[0123]

[0124] The wavelet packet function of the n-th node in the j-th level;

[0125] h k (t): low-pass filter coefficients;

[0126] g k (t): high-pass filter coefficients;

[0127] 2.2 Perform a decomposition operation on the original signal:

[0128] Gradually disassemble the original signal using the decomposition algorithm, compare it with the wavelet function. Assume the original function is F and decompose the original function into F1, F2, F3, F4... and the wavelet parts W1, W2, W3, W4... successively, that is, F1 = FJ - 1 + WJ - 1, and continue to decompose like this until it finally terminates at the 0-th level.

[0129] The signal of each layer passes through the low-pass filter h k and the high-pass filter g k to calculate the decomposition coefficients:

[0130]

[0131] The signal value of the n-th node in the j-th level;

[0132] h[m], g[m]: low-pass and high-pass filter coefficients;

[0133] 2.3 Use the multi-resolution decomposition algorithm to calculate the wavelet coefficients at each level.

[0134] 2.4 Obtain the sub-band signals of the original signal in each frequency according to the calculated wavelet coefficients.

[0135] The result of wavelet packet decomposition is a binary tree structure: each node represents the component of the signal in a specific frequency band; the depth of the tree determines the decomposition level, and the deeper the level, the higher the resolution of the signal frequency. Therefore, the initially set value represents that each wavelet packet layer represents a sub-band signal.

[0136] 2.5 Use the multi-resolution reconstruction algorithm to reconstruct the approximation coefficients and wavelet coefficients in the form of discrete filters. The reconstruction of the signal uses a linear combination of wavelet packet coefficients:

[0137]

[0138] Through the inverse filters h k and g k realize layer-by-layer reconstruction;

[0139]

[0140] 2.6 Separate the reconstructed signal according to the active frequency bands of electroencephalogram and electromyogram.

[0141] The frequency bands of electroencephalogram (EEG) signals are usually divided into several typical frequency bands according to different frequency ranges:

[0142] Delta 0.5Hz - 4Hz;

[0143] Theta 4Hz - 8Hz;

[0144] Alpha 8Hz - 13Hz;

[0145] Beta 13Hz - 30Hz;

[0146] Gamma 30Hz - 100Hz;

[0147] They correspond to different activity states and functions of the brain respectively. Decompose the reconstructed EEG signal into five frequency bands and decompose the EMG signal into the same frequency bands.

[0148] 2.7 Save the individual frequency domain signals.

[0149] Step 3: Use the Hilbert transform to extract the phase information, which specifically includes the following:

[0150] 3.1 Convert the split signal into a time series x(t);

[0151] 3.2 Use the Hilbert transform to convert the time series into a complex signal.

[0152]

[0153] The Hilbert transform is usually simplified by the Fourier transform. The manifestation of the Hilbert transform in the frequency domain is to shift the phase of the negative frequency component of the signal by -90 degrees and the phase of the positive frequency component by +90 degrees.

[0154]

[0155] The analytic signal z(t) is a complex-valued signal composed of the original real-valued signal x(t) and its Hilbert transform constituting a complex-valued signal

[0156] 3.3 The instantaneous phase is obtained by calculating the phase angle of the complex signal.

[0157]

[0158] 3.4 The instantaneous amplitude is obtained by calculating the formula.

[0159]

[0160] 3.5 Phase unwrapping is performed to ensure phase continuity

[0161] φ unwrapped (n)=φ unwrapped (n - 1)+Δφ(n);

[0162] where:

[0163] Δφ(n)=φ unwrapped (n)-φ unwrapped (n - 1);

[0164] To ensure phase continuity, Δφ(n) needs to be adjusted:

[0165]

[0166] The finally unwrapped phase is:

[0167] φ unwrapped (n)=φ unwrapped (n - 1)+Δφ adjusted (n);

[0168] 3.5 Phase verification is performed. The calculated phase is compared with the phase of the original signal to judge the accuracy of the phase of the reconstructed signal.

[0169] Step 4: The modulation index (MI) is estimated using phase-amplitude coupling (PAC), which specifically includes the following:

[0170] 4.1 To better measure the adaptability of PAC, first, the phase φ(n) is cut into n slices, and the phase is binned into n intervals, each interval being 360 / n°.

[0171] 4.2 The average value of the amplitude a(t) is taken within each defined interval and is denoted by <a φ (t). The probability distribution P is obtained by dividing the amplitude within each interval by the sum of the intervals.

[0172]

[0173] Where P(j) represents the normalized amplitude within an interval. This distribution can be used to use the modulation index.

[0174] 4.3 The modulation index (MI) is calculated through the Kullback-Leibler distance, which is used to measure the deviation between the amplitude probability distribution P and the uniform distribution Q (i.e., the amplitude values are uniformly distributed within the range [-1, 1]):

[0175]

[0176] Where:

[0177]

[0178] For the uniform distribution, Q(k) = 1 / n, and it can also be obtained that

[0179]

[0180] Step 5: In the upper limb rehabilitation level assessment, explore the consistency relationship between the modulation parameters estimated by phase-amplitude coupling and the clinical assessment scale based on the Spearman method. The calculation process of the Spearman correlation coefficient is as follows:

[0181]

[0182] In the formula, ρ is the Spearman correlation coefficient, and h i is the rank difference of each pair of observed values. In the present invention, each pair of observed values is divided into MI and Brunnstrom levels, num is the total number of observed values, and after calculating two ρ values respectively, the mean value ρ mean is obtained to represent the fitting relationship between the MI of each patient and the Brunnstrom assessment scale, analyze the FCMC characteristic indexes and change trends that are significantly correlated with the Brunnstrom assessment scale level, and complete the quantitative assessment of the disease level of stroke patients.

[0183] As described above, it is only the preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A method for evaluating hand function in stroke patients, characterized in that: The following steps are involved: Step 1, signal acquisition and preprocessing: Combined with the Brunnstrom rating scale, the action paradigm was set, and the synchronous EEG signals and EMG signals of patients with different levels of stroke were collected at the same time, and the EEG signals and EMG signals were preprocessed respectively; Step 2, decomposition of brain and electromyographic signals: Synchronously decompose the preprocessed brain and electromyographic signals based on wavelet packet transform, including constructing a wavelet packet function and selecting a wavelet packet basis function, decomposing the signal, using a multi-resolution decomposition algorithm to calculate the wavelet coefficients of each level, and obtaining the sub-band signals of the original signal in each frequency according to the wavelet coefficients. The wavelet approximation coefficients and the wavelet coefficients are reconstructed using a multi-resolution reconstruction algorithm, and the reconstructed signals are separated according to the active bands of EEG and EMG, and the individual frequency domain signals are saved; Step 3, extract phase information using Hilbert transform: convert the decomposed brain and electromyographic signals into time series, use discrete Hilbert transform to convert the time series into complex signals, calculate the instantaneous phase by calculating the phase angle of the complex signal, calculate the instantaneous amplitude, and perform phase verification; Step 4: Use phase-amplitude coupling to estimate the modulation index: Cut the phase φ(n) into n slices, and the phase is distributed into n intervals, each interval is 360 / n°; take the average value of the amplitude a(t) in each demarcated interval, and use <a(t) φ Indicates that the probability distribution P is obtained by dividing the amplitude in each interval by the sum of the intervals; the modulation index is obtained using the Kullback-Leibler distance; Step 5, assessment of upper limb rehabilitation level: Based on the Spearman correlation coefficient, the fitting relationship between each patient's MI and the Brunnstrom rating scale was mapped, and the FCMC characteristic indicators and change trends that were significantly correlated with the Brunnstrom rating scale level were analyzed to quantitatively assess the disease level of stroke patients.

2. The method for evaluating hand function of stroke patients according to claim 1, characterized in that: In step 1, the EEG signals of stroke patients at leads Fp1, Fp2, F3, F4, C3, C4, Cz, P3, P4, A1, and A2 are collected respectively, and the EMG signals of the anterior bundle of the deltoid muscle, the middle bundle of the deltoid muscle, the biceps brachii, the triceps brachii, and the extensor digitorum of the forearm are collected. The preprocessing operations of EEG signals include: re-referencing, baseline correction, filtering, removal of electrooculogram and EEG data segmentation; the preprocessing operations of EMG signals include: baseline correction, filtering and EMG data segmentation.

3. The method for evaluating hand function of stroke patients according to claim 1, characterized in that: In step 2, the wavelet transform performs multi-resolution analysis on the signal through the scaling function φ(t) and the wavelet function ψ(t); the orthogonality condition is satisfied, and the wavelet basis function is generated by scaling and translating: f j,k (t)=2 j / 2 φ(2 j tk),ψ j,k (t)=2 j / 2 ψ(2 j tk); Among them, j is the scale parameter and k is the translation parameter; At the jth layer, the wavelet packet basis function Defined as: In the jth layer, the wavelet packet function of the nth node; h k (t): low-pass filter coefficient; g k (t): high-pass filter coefficient; The original signal is gradually disassembled using a decomposition algorithm and compared with the wavelet function. Assuming the original function is F, the original function is decomposed into F1, F2, F3, F4... step by step; and with the wavelet part W1, W2, W3, W4..., F1=FJ-1+WJ-1, decomposing to the 0th level; The signal of each layer passes through a low-pass filter h k and high pass filter g k Calculate the decomposition coefficients: The signal value of the nth node in the jth layer; h[m], g[m]: low-pass and high-pass filter coefficients; The wavelet coefficients at each level are calculated using a multiresolution decomposition algorithm; According to the wavelet coefficients, the sub-band signals of the original signal in each frequency are obtained; The wavelet approximation coefficients and wavelet coefficients are reconstructed using a multi-resolution reconstruction algorithm in the form of discrete filters; the signal is reconstructed using a linear combination of wavelet packet coefficients: Through the inverse filter h k and g k Realize layer-by-layer reconstruction; Separate the reconstructed signal according to the active bands of EEG and EMG; and decompose the EMG signal into the same frequency bands; Save the individual frequency domain signals.

4. The method for evaluating hand function of stroke patients according to claim 1, characterized in that: In step 3, the decomposed brain and myoelectric signals are converted into time series; Use discrete Hilbert transform to transform the time series into a complex signal; The instantaneous phase is obtained by calculating the phase angle of the complex signal; φ(t)=arg(z(t)); The instantaneous amplitude is calculated by the calculation formula; Compensate between phase differences; f unwrapped (n)=φ unwrapped (n-1)+Δφ(n); in: Δφ(n)=φ unwrapped (n)-φ unwrapped (n-1); In order to ensure phase continuity, Δφ(n) needs to be adjusted; The final unwrapped phase is: f unwrapped (n)=φ unwrapped (n-1)+Δφ adjusted (n); Perform a phase check.

Images