Composite muscle action potential decomposition method based on maximum random contraction myoelectricity

Abstract

The invention discloses a composite muscle action potential decomposition method based on maximum random contraction myoelectricity. The method comprises the following steps: acquiring maximum random contraction myoelectricity signals and functional electrical stimulation response signals; performing maximum random contraction muscle electrolysis to obtain action potential waveforms, corresponding demixing matrixes and issuing sequences of the target muscle movement units; and performing M-wave decomposition based on the priori knowledge constraint of the target muscle motion unit. According to the invention, on the premise of obtaining a section of maximum random force contraction electromyographic signals, the function electrical stimulation of the same target muscle can be automatically extracted to induce the motion unit issuing information of the M wave, namely the decomposition of the M wave is completed. In occasions where M-wave decomposition needs to be carried out in clinical or scientific research, the method can be used for reducing the labor burden and improving the working efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of electromyographic signal processing, and specifically relates to a compound muscle action potential (M wave) decomposition method based on maximum voluntary contraction electromyography, which is mainly used in the exploration of electrical stimulation mechanism. Background Art

[0002] EMG is a bioelectric signal recorded by electrodes when a nerve excitation impulse is transmitted to the muscle, causing the muscle to contract. It carries neuromuscular control information and can reflect the activation of the muscle to a certain extent. Voluntary contraction EMG is composed of the time and space superposition of multiple motor units (MU) activated and released by the central control, which propagate along the muscle fibers at the detection electrode. It can reflect the performance of the downward instructions of the nervous system on the skeletal muscle system. In the process of analyzing microscopic neural drive information, it is necessary to obtain the activity information of the most basic component unit (motor unit) of the neuromuscular system, and EMG signal decomposition technology is often used to complete this task.

[0003] EMG decomposition technology is to restore the EMG signal to its most basic component, the form of MUAP sequence, to help obtain neural drive information from the microscopic level. The MUAP sequence contains the release and waveform information of the motor unit action potential, among which the release information is the most important and can be represented as a 0-1 release sequence. In terms of the decomposition of voluntary contraction EMG signals, scholars in this field have done a lot of in-depth research. The most representative voluntary contraction EMG decomposition algorithms include the Convolution Kernel Compensation (CKC) algorithm and the Progressive Fast ICAPeel-off (PFP) algorithm. The CKC algorithm extracts the release information of the motor unit by estimating the cross-correlation vector between the motor unit release sequence and the surface EMG signal. The PFP algorithm uses the classic blind source separation algorithm, the Fast Independent Component Analysis (FastICA) algorithm, to decompose the surface EMG signal twice, and uses the results of the first decomposition to guide the second decomposition. The reliable motor unit is determined based on whether the results of the two decompositions are consistent. The reliable motor unit obtained is stripped from the original signal through the stripping strategy, and the above algorithm is executed again until no more motor units are found. The algorithm has been successfully applied in many studies and its performance has been fully verified.

[0004] Functional electrical stimulation (FES) belongs to the category of neuromuscular electrical stimulation (NES). It uses a low-frequency (1-100Hz) pulse current of a certain intensity to stimulate one or more groups of muscles through a pre-set program to induce muscle movement or simulate normal voluntary movement, so as to improve or restore the function of the stimulated muscle or muscle group. The compound muscle action potential (M wave) is a special electromyographic signal generated by directly inducing muscle contraction through electrical stimulation of motor nerves. It has a specific morphology and characteristics and can be used to measure the response of the stimulated muscle to electrical stimulation, the degree of local fatigue, and the control of the nervous system over the muscle. It is one of the most commonly used bases for studying changes in the excitability of the sarcolemma.

[0005] In order to explore the mechanism of action of functional electrical stimulation and analyze the M wave in detail and accurately, it is an indispensable step to decompose the motor unit action potential that constitutes the M wave. M wave decomposition can be regarded as the reverse process of M wave formation, that is, to restore the aliased M wave to its most basic component - the form of MUAP sequence. Since electrical stimulation of motor nerves can induce a large number of synchronous recruitment of motor units, different types of MUAPs in the M wave are severely aliased, and voluntary contraction electromyography decomposition algorithms such as CKC and PFP are difficult to apply to M wave decomposition. Summary of the invention

[0006] In order to solve the above technical problems, the present invention provides a compound muscle action potential decomposition method based on maximum voluntary contraction electromyography, which collects the maximum voluntary force contraction electromyography signal that theoretically contains the information of all motor units constituting the target muscle; decomposes the maximum voluntary force contraction electromyography signal to obtain the motor unit waveform, release sequence and unmixing matrix and other prior knowledge of the target muscle; and carries out M-wave decomposition based on the prior knowledge constraints of the target muscle motor unit, which is suitable for decomposing M waves in clinical or scientific research to obtain motor unit release information and other occasions.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] Step 1, collecting the electromyographic signal of the target muscle in the state of maximum voluntary force and the functional electrical stimulation response signal in the state of relaxation;

[0009] Step 2, decomposing the electromyographic signal under the maximum voluntary force state to obtain prior information including the target muscle unmixing matrix, motor unit waveform and firing sequence;

[0010] Step 3: Use the prior information of the target muscle as a constraint and combine it with the electrical stimulation response signal to perform M-wave decomposition.

[0011] The beneficial effects of the present invention are:

[0012] The M-wave decomposition method proposed in the present invention is highly operational. It only requires pre-collecting a section of maximum voluntary contraction electromyographic data and decomposing it to obtain prior information on the motor units of the target muscle. Subsequently, the M-wave induced by functional electrical stimulation can be automatically decomposed, which is faster and more reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is a flow chart of a method for decomposing compound muscle action potential based on maximum voluntary contraction electromyography of the present invention;

[0014] Figure 2 Schematic diagram of maximum voluntary contraction electromyographic signal and electrical stimulation response signal in an embodiment of the present invention;

[0015] Figure 3 This is a schematic diagram of the motor unit extraction result of the maximum voluntary contraction electromyography in an embodiment of the present invention;

[0016] Figure 4 This is a comparison diagram of the effects before and after M-wave extraction of the electrical stimulation response signal in an embodiment of the present invention;

[0017] Figure 5 It is the mixed signal obtained by splicing the maximum voluntary contraction electromyography and M wave;

[0018] Figure 6 A schematic diagram of a source signal and an extracted emission sequence obtained by using an unmixing matrix and a mixed signal;

[0019] Figure 7 is the reference signal obtained by replacing the emission sequence;

[0020] Figure 8 A schematic diagram of the source signal and the extracted emission sequence obtained by FastICA decomposing the mixed signal using the reference signal;

[0021] Fig. 9 Schematic diagram of the action potential waveform and firing sequence obtained by decomposing the M wave in an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0023] like Figure 1 As shown, for a certain target muscle of the object to be tested, the process of performing M wave decomposition using the present invention includes:

[0024] Step 1: Collect the electromyographic signal of the target muscle in the state of maximum voluntary force and the functional electrical stimulation response signal in the state of relaxation:

[0025] Step 1.1: Place an array of electrodes of r (rows) × c (columns) on the target muscle. First, let the subject arbitrarily contract the target muscle with maximum force and collect the length of Multi-channel surface electromyography ; Then, functional electrical stimulation was applied to the target muscle in a relaxed state, and the acquisition length was The electrical stimulation response signal .in The length should satisfy .in, represents the maximum voluntary contraction surface electromyographic signal of the i-th channel, is the electrical stimulation response signal of the i-th channel, U is the total number of channels and ;

[0026] In this embodiment, the data acquisition device uses a 128-channel high-density surface electromyography acquisition device. The array electrode is attached to the biceps brachii of the right arm of the human body, and the common reference electrode and ground electrode are attached to the wrists of the right and left hands respectively. The system sampling frequency is 2000Hz, and the electrodes are arranged in 4 rows × 4 columns, that is, r=4, c=4, U=16. The acquisition time of the multi-channel maximum voluntary contraction electromyography signal is 10 seconds, and the acquisition time of the electrical stimulation response signal is 2.5 seconds, that is, , the stimulation frequency was 30 Hz and the current intensity was 15 mA. Figure 2 The maximum voluntary contraction electromyographic signal and electrical stimulation response signal collected in this embodiment are shown.

[0027] Step 2: Decompose the electromyographic signal under the maximum voluntary force state to obtain the prior information including the target muscle unmixing matrix, motor unit waveform and firing sequence:

[0028] Decomposition of multi-channel surface electromyographic signals using an automated stepwise variable stripping algorithm , and get the unmixing matrix , waveform matrix of N motion units and the release sequence matrix . represents the unmixing vector corresponding to the i-th motion unit, represents the waveform corresponding to the i-th motion unit, represents the firing sequence corresponding to the i-th motor unit; the motor unit waveform and firing sequence extracted in this embodiment are as follows Figure 3 shown.

[0029] Step 3: Using the prior information of the target muscle as a constraint, the M-wave decomposition is performed in combination with the electrical stimulation response signal;

[0030] Step 3.1: Use FastICA denoising algorithm to remove the electrical stimulation response signal The electrical stimulation pulse artifact. Get the M wave signal ,in represents the M wave signal of the i-th channel. Figure 4 Shown is a schematic diagram of artifact removal effect;

[0031] Step 3.2: Splicing multi-channel surface electromyography signals and the M wave signal to obtain a mixed signal ,in Represents the mixed signal of the i-th channel. Figure 5 This is an example of a mixed signal obtained by splicing;

[0032] Mixed Signal It can be expressed as the convolution of the action potential waveform and the firing sequence as shown in formula (1),

[0033]

[0034] In the formula, represents the mixed signal of the ith channel, It represents the waveform obtained by sampling the action potential emitted by the jth motor unit at the ith channel. represents the jth release sequence, t, Parameters used in the convolution operation;

[0035] According to formula (2), the mixed signal The expanded electromyographic data can be obtained by expansion, that is, the expanded signal The purpose of matrix expansion is to convert convolution into matrix multiplication for blind source separation.

[0036]

[0037] Wherein, K is the expansion factor. In this embodiment, K=15; Indicates that The vector obtained by circularly shifting left by e bits;

[0038] Step 3.3, obtaining the first source signal and preliminary release sequence;

[0039] Step 3.3.1: For extended signal , using the unmixing matrix Solving the motion unit source signal matrix .in represents the source signal of the ith motor unit, Figure 6 The calculated source signal of one motor unit is shown;

[0040] Step 3.3.2: Source signal , calculated according to the Otsu method Threshold , extract the corresponding preliminary release sequence ; After extracting the release sequence of each source signal, the preliminary release sequence matrix is ​​obtained .

[0041] Step 3.4: Initially release the sequence The data at the position corresponding to the maximum voluntary contraction electromyography are used as the firing sequence Replace it and get the reference signal corresponding to the i-th motion unit . The length should be , in theory, the former The points are the firing sequence of the corresponding motor unit in the maximum voluntary contraction electromyography. The points are the firing sequences of the corresponding motor units in the M wave signal. In fact, the introduction of the M wave will lead to the preliminary firing sequence mentioned in step 3.3 Therefore, Before The points are replaced by the firing sequence of the i-th motor unit obtained by the maximum voluntary contraction electromyographic decomposition ; For each initial release sequence After replacing all, we get the reference signal matrix , used to guide the update of the source signal, Figure 7 A reference signal obtained in this embodiment is shown;

[0042] Step 3.5, reference signal guides source signal and release sequence update;

[0043] Step 3.5.1: Use FastICA algorithm to analyze the extended signal Decompose and use the model constraint output source signal as shown in formula (3) Similarity with reference signals to ensure reliability;

[0044]

[0045]

[0046] In the formula, represents the negative entropy function, E represents the expectation, represents a nonlinear function, represents random numbers that conform to the standard normal distribution, Reflects the i-th source signal to be extracted With reference signal The distance measure between is a preset correlation lower bound. In this embodiment , ;

[0047] Step 3.5.2: Define the upper limit of the number of iterations as , initialize the number of iterations d = 1, initialize the i-th source signal , the i-th unmixing vector , and iterate according to steps 3.5.3, 3.5.4 and 3.5.5 until d = ;

[0048] Step 3.5.3: Obtain the ith unmixing vector of the d+1th iteration by equation (4): :

[0049]

[0050] In the formula, Represents a nonlinear function The first derivative of is the distance metric function The first derivative of ;

[0051] Step 3.5.4: For the ith unmixing vector of the d+1th iteration Standardization: Endow ,in yes 2-norm of ;

[0052] Step 3.5.5: Update the i-th source signal of the d+1-th iteration ; Assign d+1 to d;

[0053] Step 3.5.6: Calculate according to Otsu's method Threshold , extract the corresponding release sequence And save. And for the source signal i+1th source signal and release sequence Continue updating until the source signal and release sequence corresponding to the last motor unit are updated, and the updated release sequence matrix is ​​obtained. ; Figure 8 A source signal and a reference signal updated in this embodiment are shown;

[0054] Step 3.6: Save the emission sequence matrix corresponding to the M wave , and as a final result, Release sequence matrix for update The part representing the M wave emission sequence; Fig. 9 The waveform and firing sequence of the motor unit in the M wave extracted in this embodiment are shown.

[0055] In the field of myoelectric decomposition, it is generally believed that one unmixing vector corresponds to one motor unit. Each release sequence is consistent with the waveform matrix obtained in step 2.1 The waveform information in corresponds one to one.

[0056] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above embodiments can be implemented by software, or by means of software plus necessary general hardware platforms. Based on such understanding, the technical solutions of the above embodiments can be embodied in the form of software products, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0057] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for decomposing compound muscle action potential based on maximum voluntary contraction electromyography, characterized in that: The steps include: Step 1, collecting the electromyographic signal of the target muscle in the state of maximum voluntary force and the functional electrical stimulation response signal in the state of relaxation; Step 2, decomposing the electromyographic signal under the maximum voluntary force state to obtain prior information including the target muscle unmixing matrix, motor unit waveform and firing sequence; Step 3: Use the prior information of the target muscle as a constraint and combine it with the electrical stimulation response signal to perform M-wave decomposition.

2. The method for decomposing compound muscle action potential based on maximum voluntary contraction electromyography according to claim 1, characterized in that: The step 1 comprises adopting The array electrode patch is placed on the target muscle, and the target muscle is contracted at will with maximum force. Multi-channel surface electromyography ; Functional electrical stimulation is applied to the target muscle in a relaxed state, and the acquisition length is The electrical stimulation response signal ,in, represents the maximum voluntary contraction surface electromyographic signal of the i-th channel, is the electrical stimulation response signal of the i-th channel.

3. The method for decomposing compound muscle action potential based on maximum voluntary contraction electromyography according to claim 2, characterized in that: The acquisition length of the multi-channel surface electromyographic signal and the electrical stimulation response signal meets .

4. The method for decomposing compound muscle action potential based on maximum voluntary contraction electromyography according to claim 1, characterized in that: The step 2 includes decomposing the multi-channel surface electromyographic signal using an automated stepwise variable stripping algorithm , and get the unmixing matrix , waveform matrix of N motion units and the release sequence matrix ,in represents the unmixing vector corresponding to the i-th motion unit, represents the waveform corresponding to the i-th motion unit, Represents the firing sequence corresponding to the i-th motor unit.

5. The method for decomposing compound muscle action potential based on maximum voluntary contraction electromyography according to claim 1, characterized in that: The step 3 comprises: Step 3.1, using filtering or FastICA algorithm to remove electrical stimulation pulse artifacts in the electrical stimulation response to obtain an M wave signal; Step 3.2, splicing multi-channel surface electromyographic signals and M waves to obtain a mixed signal; Step 3.3: Use the unmixing matrix and the mixed signal to obtain the source signal S of the motion unit release sequence, and extract the release sequence of the source signal S to obtain the preliminary release sequence matrix ; Step 3.4, replace the part of the firing sequence corresponding to the maximum voluntary contraction electromyography with the known firing sequence under the maximum voluntary force contraction, and obtain the reference signal matrix R; Step 3.5, perform FastICA decomposition on the mixed signal, and use the reference signal to constrain the output of FastICA to guide the extraction of the source signal S, and obtain an updated source signal matrix; Step 3.6: Extract the release sequence of the source signal matrix to obtain the updated release sequence matrix , and obtain the M wave emission sequence matrix .

6. The method for decomposing compound muscle action potential based on maximum voluntary contraction electromyography according to claim 5, characterized in that: The step 3.2 includes splicing multi-channel surface electromyographic signals and M waves to obtain a mixed signal, wherein the mixed signal is represented by the convolution of the action potential waveform and the firing sequence; and expanding the mixed signal to obtain expanded electromyographic data, namely, the expanded signal.

7. The method for decomposing compound muscle action potential based on maximum voluntary contraction electromyography according to claim 5, characterized in that: The step 3.3 includes, for the extended signal, using the unmixing matrix to solve the motion unit source signal matrix, and calculating the threshold according to the Otsu method for the source signal of the i-th motion unit. , extract the corresponding release sequence ; After extracting the release sequence of each source signal, the preliminary release sequence matrix is ​​obtained .

8. The method for decomposing compound muscle action potential based on maximum voluntary contraction electromyography according to claim 5, characterized in that: The step 3.4 includes replacing the data at the position corresponding to the maximum voluntary contraction electromyography in the preliminary firing sequence matrix with the firing sequence, and obtaining the reference signal matrix R after replacing each firing sequence.

9. The method for decomposing compound muscle action potential based on maximum voluntary contraction electromyography according to claim 5, characterized in that: The step 3.5 comprises: Step 3.5.1: Use FastICA algorithm to analyze the extended signal To decompose; Step 3.5.2: Define the upper limit of the number of iterations as , initialize the number of iterations d = 1, initialize the i-th source signal , the i-th unmixing vector , and iterate according to steps 3.5.3, 3.5.4 and 3.5.5 until ; Step 3.5.3: Obtain the ith unmixing vector of the d+1th iteration by equation (4): : (4) In the formula, Represents a nonlinear function The first derivative of is the distance metric function The first derivative of , E represents expectation; Step 3.5.4: For the ith unmixing vector of the d+1th iteration To standardize; Step 3.5.5: Update the i-th source signal of the d+1-th iteration ; Assign d+1 to d; Step 3.5.6: Calculate according to Otsu's method Threshold , extract the corresponding release sequence And save, and for the source signal i+1 source signal and release sequence Continue updating until the source signal and release sequence corresponding to the last motor unit are updated, and the updated release sequence matrix is ​​obtained. .

10. The method for decomposing compound muscle action potential based on maximum voluntary contraction electromyography according to claim 5, characterized in that: In step 3.6, the emission sequence matrix corresponding to the M wave in the updated emission sequence U is saved. .

Images