CN113397571A - Myoelectric movement unit decomposition method based on prior template - Google Patents

Myoelectric movement unit decomposition method based on prior template Download PDF

Info

Publication number
CN113397571A
CN113397571A CN202110805279.0A CN202110805279A CN113397571A CN 113397571 A CN113397571 A CN 113397571A CN 202110805279 A CN202110805279 A CN 202110805279A CN 113397571 A CN113397571 A CN 113397571A
Authority
CN
China
Prior art keywords
motion unit
action potential
signal
electromyographic
prior
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202110805279.0A
Other languages
Chinese (zh)
Inventor
汪婷
皮少军
席旭刚
王俊宏
罗志增
吕忠
李文国
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hangzhou Dianzi University
Original Assignee
Hangzhou Dianzi University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hangzhou Dianzi University filed Critical Hangzhou Dianzi University
Priority to CN202110805279.0A priority Critical patent/CN113397571A/en
Publication of CN113397571A publication Critical patent/CN113397571A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/389—Electromyography [EMG]
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/389—Electromyography [EMG]
    • A61B5/397—Analysis of electromyograms
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/02—Preprocessing
    • G06F2218/04—Denoising
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Veterinary Medicine (AREA)
  • Public Health (AREA)
  • Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • Signal Processing (AREA)
  • Artificial Intelligence (AREA)
  • Physiology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Psychiatry (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)

Abstract

The invention relates to a myoelectric motion unit decomposition method based on a prior template. Firstly, fitting the waveform of the action potential of a motion unit by using a plurality of prior templates with variable expressions; then, an iterative matching strategy is provided, and the layer-by-layer peeling of the motion unit is realized; and then, a peak triggering average algorithm is introduced to re-determine the action potential waveform of the movement unit so that the movement unit is more consistent with a real electromyographic movement unit. And finally, establishing an evaluation index based on the real signal decomposition result for evaluating the performance of the decomposition algorithm. The invention can detect and identify effective motion units under low noise level, and has better performance.

Description

Myoelectric movement unit decomposition method based on prior template
Technical Field
The invention belongs to the field of signal processing and pattern recognition, relates to a decomposition method of an electromyographic signal, and particularly relates to a method for decomposing the electromyographic signal into movement unit component information based on a prior template.
Background
When muscles contract, motor neurons in the motor units begin to excite, all muscle fibers dominated by the motor neurons are caused to contract, so that action potentials of the motor units are generated, and action potentials emitted by a plurality of motor units form electromyographic signals at the detection electrodes through a complex comprehensive superposition process. The decomposition of the electromyographic signals is actually the reverse process of signal generation, and the fundamental purpose of the decomposition is to separate action potentials issued by each motion unit and acquire the relevant information of the motion units. The relevant information of the motor units, such as the release waveform, the interval release mode, the recruitment rule and the like, can provide decisive details about clinical diagnosis of the muscle diseases which are vital to the health of the nervous system, is beneficial to fundamentally exploring the physiological mechanism of the neuromuscular system, and has great value in the aspects of clinical diagnosis of the muscle diseases, research of the motion coordination relationship among muscle tissues, exploration of the control mechanism of the central neuron and the like. Therefore, the study of electromyographic signal decomposition has important significance for reflecting the control mechanism of the motor-nerve-muscular system and clinical application diagnosis.
In the last two decades, many researchers are dedicated to the decomposition of electromyographic signals, and various decomposition technologies are proposed, so that certain breakthrough achievements are achieved. According to the difference of the basic ideas of the achievement decomposition, the electromyographic signal decomposition method can be summarized into two types of system identification method and morphological decomposition method. The system identification method mainly takes the neuromuscular system of a human body as a multi-input multi-output system, and decomposes the electromyographic signals into components according with the action potential distribution characteristics of the movement units by adopting a blind source separation method. Based on this, many scholars make a great contribution, Zazula and Holobar et al apply blind deconvolution and convolution kernel compensation in combination to electromyographic signal decomposition and obtain reliable results. Chen-Luck et al constructed a progressive Fast ICA stripping framework for high-density electromyographic signal decomposition, and verified the reliability through multiple tests and simulations. However, the current system identification method is based on high-density electromyographic signals, and has certain limitation on single-channel electromyographic signals.
The morphological decomposition method is to directly extract the action potential sequence of the movement unit from the electromyographic signal by an optimization algorithm according to the inherent morphological characteristics of the action potential waveforms of different movement units. Since 1982, Deluca group has been engaged in electromyogram signal morphological Decomposition work, and recently a Precision Decomposition III system based on morphological Decomposition was established and the Decomposition result was compared with the intramuscular electromyogram signal to evaluate the Decomposition reliability. Subsequently, myoelectric signals under 75% -100% of the maximum autonomous contraction are decomposed by the system, and the accuracy rate reaches 92.5%. These studies lay a solid foundation for electromyographic signal decomposition and provide convenience for further exploring the influence of the movement unit on the movement.
Disclosure of Invention
The invention provides a myoelectric signal decomposition method based on a prior template, aiming at the problems that the existing myoelectric signal decomposition method has few general mathematical expressions, the decomposition process is complicated and complex, and the flexibility is lacked in practical application.
Firstly, fitting the waveform of the action potential of a motion unit by using a plurality of prior templates with variable expressions; then, an iterative matching strategy is provided, and the layer-by-layer peeling of the motion unit is realized; and then, a peak triggering average algorithm is introduced to re-determine the action potential waveform of the movement unit so that the movement unit is more consistent with a real electromyographic movement unit. And finally, establishing an evaluation index based on the real signal decomposition result for evaluating the performance of the decomposition algorithm.
The invention comprises the following steps:
step 1: myoelectric signals of 15% of volunteers in a resting state under continuous action of maximum voluntary contractility are collected.
Step 2: and (3) carrying out wavelet packet denoising processing on the electromyographic signals obtained in the step (1) to obtain relatively pure electromyographic signals.
And step 3: the myoelectric signal is decomposed by a motion unit, and the method comprises the following specific steps:
step 3.1: four prior templates are determined based on morphological characteristics of the motion unit firing sequence waveforms. The n-order Hermite-Rodriguez function is selected as a prior template to simulate the action potential waveform of the moving unit, the action potential waveform of the moving unit is mostly in a two-phase or three-phase structure, and the n-order Hermite-Rodriguez function is in direct proportion to the n-order derivative of the normal function, so that the action potential waveform of the moving unit can be effectively simulated.
Step 3.2: the electromyographic signals under the resting state are used as reference signals to carry out effective peak value detection on the electromyographic signals under the continuous action of 15% of maximum autonomic contraction force: the mean value of the reference signal is selected as a detection threshold, then the effective peak values (wave crest and wave trough) of the electromyographic signals under the maximum autonomic contraction continuous acting force of 15 percent are detected based on the detection threshold, and the effective peak values are arranged in descending order according to the absolute value of the amplitude.
Step 3.3: intercepting a signal segment near an effective peak value and carrying out matching processing with a prior template, wherein the process is as follows:
1. and after the effective peak value is determined, intercepting the vicinity of each effective peak by adopting a sliding window with a variable window length to obtain a part of signal segments to be analyzed as the action potential of the alternative motion unit.
2. Calculating the distance D between the action potential of each alternative motion unit and the four prior templates by adopting a dynamic time warping algorithm, and respectively calculating the average amplitude value mean of the four prior templatesT。
3. And determining whether the action potential of the candidate motion unit is matched with the prior template or not by combining a dynamic time warping algorithm and the average amplitude of the prior template. The specific matching decision rule is as follows: if the distance D is less than one-half meanTIf so, the action potential of the alternative motion unit is considered to be matched with the prior template; otherwise there is no match.
4. And after the decision is finished, selecting the starting time of the action potential of the alternative motion unit as the issuing moment according to the action potential of the alternative motion unit matched with the prior template.
Step 3.4: the pulse triggering average technology is adopted to reconstruct the action potential waveform of the motion unit, and the process is as follows:
and taking the issuing moment of the action potential of the motion unit identified by each prior template as a trigger time point of the original sEMG signal, sequentially triggering to obtain a plurality of trigger waveforms, and averaging the trigger waveforms to obtain a reconstructed waveform corresponding to the action potential of the motion unit. These reconstructed waveforms can be considered as the true motion unit action potential waveforms delivered by each motion unit.
Step 3.5: and judging whether the loop iteration process continues or not by comparing the amplitude factor with the average value of the peak points of the reference signal. In the process of stripping the action potential of the motion unit in each iteration, if the amplitude is higher than the peak point average value of the reference signal, the visual residual signal is an updated electromyographic signal, and the step 3.2 is returned to continue the iteration; otherwise, the loop iteration ends.
The technical scheme provided by the invention has the beneficial effects that: the advantage of decomposing the electromyographic signals by using the decomposition method comprises a simple mathematical expression of the template, thereby reducing the complexity and improving the robustness of the algorithm. Meanwhile, a simulation electromyographic signal is constructed according to a system identification model, the effectiveness of the proposed decomposition method is evaluated, and the result shows that the precision of the decomposition method is 90.9 percent and the method is sensitive to low signal-to-noise ratio. And finally, two evaluation indexes are provided according to the decomposition result of the real myoelectricity, the evaluation is carried out by a reverse bisection calculation method, the result shows that the change of the evaluation indexes completely accords with the expectation, and the feasibility of decomposition is proved again. Furthermore, the final analysis shows that large muscle motor unit activation provides more information than small muscle motor unit activation during muscle contraction, revealing significantly the internal physiological mechanisms and main features of muscle information.
Drawings
FIG. 1 is a flow chart of an embodiment of the present invention;
FIG. 2 is a diagram of prior templates for first and second order Hermite-Rodriguez function simulations;
FIG. 3 is a diagram of the decomposition process and the result of the electromyographic signals collected under the action of 15% of the maximum autonomic contraction force;
FIG. 4 is a graph showing the variation of correlation coefficient between the action electric potential sequence of the exercise unit and the original electromyographic signal for different combinations of the number of the exercise units.
Detailed Description
The invention provides a single-channel electromyographic signal decomposition method based on a morphological decomposition idea. The main technical concept is that several fixed prior templates of the motion unit are determined according to the past research results, and the peak information of the actual electromyographic signal is extracted to set the template parameters and extract the effective peak. And then, intercepting a signal segment near the effective peak to perform template matching, and determining the release time of the motion unit. The motion unit firing waveform is then re-determined according to a pulse triggered averaging technique. And then deleting the identified action potential of the motion unit from the electromyographic signal to obtain a residual signal, taking the residual signal as a new electromyographic signal, repeating the steps, and continuously iterating until a termination condition is met to obtain a final decomposition result. And finally, establishing a new cross-correlation coefficient index based on the decomposition result to verify the effectiveness and feasibility of the decomposition method.
The following detailed description will be made with reference to the accompanying drawings, and as shown in fig. 1, this embodiment includes the following steps:
step 1, collecting electromyographic signals of a volunteer under the continuous acting force of 15% of maximum autonomous contractility and under a resting state, wherein the specific mode is as follows:
the electromyographic signals are transmitted through DELSYS TrignoTMThe wireless electromyography system performs acquisition, the acquisition sampling frequency is 2000Hz, the testee is 4 healthy right-handed men, and the electromyography acquisition channels are all superficial flexors. The specific flow of the experiment is as follows, the screen display of the first 2 seconds is 'ready', when the prompt tone 'ding' is played, the testee gradually grips the grip force sensor, the action time is 2 seconds, when the grip force strength reaches the designated autonomous contraction force, the prompt tone 'di' can be played to remind the testee to maintain the stable force for 5 seconds, when the testee hears the prompt tone 'dong', the hands are released within 2 seconds, and then the experiment flow is finished after 2 minutes of rest. Grip tracking is required throughout the experiment to ensure that muscles can maintain a stable contraction state during 5 seconds of continuous force application. Record myoelectricity for 5 secondsAnd denoising the signal by adopting a wavelet packet to obtain a purer signal.
And 2, the electromyographic signals with high quality can obtain better decomposition effect, and the key points are that aliasing noise signals in the signals are effectively eliminated, the noise influence is reduced, and the signal-to-noise ratio is improved. Therefore, the invention adopts a wavelet packet denoising algorithm to preprocess the electromyographic signals, and the main idea is as follows: the electromyographic signals containing noise are decomposed into components of different frequency bands through wavelet packet transformation, electromyographic signal components are generally concentrated in a large wavelet coefficient, the noise components are distributed in a wavelet domain, and then wavelet packet coefficients of different scales are utilized for reconstruction. The reference signal refers to an electromyographic signal acquired in a resting state.
And step 3: the myoelectric signal is decomposed by a motion unit, and the method comprises the following specific steps:
step 3.1, determining four prior templates according to morphological characteristics of the waveform of the motion unit issuing sequence: the action potential waveform of the motion unit is mostly in a two-phase or three-phase structure, and the Hermite-Rodriguez function of the n order is in direct proportion to the derivative of the n order of the normal function, so that the action potential waveform of the motion unit can be effectively simulated. The basic expression of the Hermite-Rodriguez function of order n is as follows:
Figure BDA0003166120840000051
where n is 1,2, the order of the HR function, and α represents a time scale factor related to the amplitude of the MUAP waveform, the larger the amplitude, the larger the time scale factor.
Figure BDA0003166120840000052
Is an n-th order hermitian polynomial. The first-order and second-order Hermite-Rodriguez functions respectively correspond to a two-phase waveform and a three-phase waveform, and the expressions are as follows:
Figure BDA0003166120840000053
Figure BDA0003166120840000054
wherein A is1,A2Is a normalized amplitude factor of the motion unit action potential. The motion unit action potential sequence issued by the motion units with different sizes is approximately simulated by a plurality of morphological waveforms obtained by adjusting parameters such as amplitude factors, time scale factors and the like in the prior template, wherein the value range of a dependent variable t in the template is set to be-20 ms, the variation range of the time scale factor alpha is set to be 5-20 ms, so that the motion unit action potential waveform is more effectively approximated, and FIG. 2 is a simulation waveform schematic diagram of first-order and second-order HR functions.
Step 3.2, the electromyographic signals in the resting state are used as reference signals to carry out effective peak value detection on the electromyographic signals under the continuous action of the 15% maximum autonomic contraction force, and an amplitude factor A is initialized, wherein the method specifically comprises the following steps: firstly, selecting the mean value of the reference signals as a detection threshold, then detecting effective peak values (wave crest and wave trough) of the electromyographic signals under the action force of 15% of the maximum autonomic contraction duration based on the threshold, and arranging the effective peak values in descending order according to the absolute value of the amplitude values. Then, initializing the amplitude factor A of the template according to the following constraint rule:
Figure BDA0003166120840000061
arc max m
s.t.|peak1|-|peakm|≤2sd(rest_signal)
wherein, | peakkI represents the absolute value of the amplitude of the kth peak point in the electromyographic signals collected under a certain MVC power level and satisfies | peak1|≥|peak2|≥…≥|peakk|≥…，A0Denotes the initial value of the amplitude factor, and sd (rest _ signal) is the standard deviation of the reference signal.
The amplitude factor a determines the magnitude of the MUAP waveform, and in the present invention, the amplitude factor varies in the form of an equal ratio sequence with a common ratio of 1/2, and is intended to describe the amplitude levels of different MUAPs, with the following constraints:
Figure BDA0003166120840000062
arc max i
s.t.Ai≤mean(rest_peaks)
in the formula, AiRepresents the amplitude factor for the ith iteration and mean (rest _ peaks) represents the average of all the peak amplitudes in the reference signal.
Effective peak values (wave crest and wave trough) of the electromyographic signals under the maximum autonomous contraction continuous acting force of 15% are detected based on the constraint conditions, and are arranged in a descending order according to the absolute value of the amplitude.
Step 3.3, intercepting a signal segment near the effective peak value and carrying out matching processing on the signal segment and a prior template, wherein the template matching comprises the following steps:
1. in determining the amplitude factor AiThen, all electromyographic signals meeting the peak are selectedi|-AiThe peak pulse point of 0 or more is used as the effective peak. The traditional method is that the effective peak position is used as a symmetrical point, a signal segment is intercepted through a window with a fixed length, and then the signal segment is matched and analyzed with a prior template. However, the durations of the time intervals for which the motion unit action potentials of different amplitudes are delivered are different from the delivery intervals, and the motion unit action potentials of different shapes at the same amplitude level are not necessarily symmetrical about the peak point. Therefore, the embodiment adopts a sliding window with a variable window length to intercept the effective peaks, and obtains a part of signal segments to be analyzed as the action potential of the alternative motion unit.
2. And calculating the similarity of the action potential of each alternative motion unit and the four templates by adopting a dynamic time warping algorithm, and normalizing the similarity. Based on the idea of dynamic programming, the dynamic time warping algorithm can measure the similarity of two time series with different lengths.
Given two time series: exercise of sportsCell action potential candidate xl(t)＝x1,x2,...xi,...,xnAnd experience template yk(t)＝y1,y2,...yi,...,ymTheir lengths are n and m, respectively. Dynamic time warping algorithms typically use dynamic programming algorithms, and to align the two sequences, an n x m matrix D is constructed, with matrix elements DijRepresenting the amplitude distance between the two points xi and yj. The algorithm can be understood as finding the smallest path among the paths from the lower left corner to the upper right corner in the matrix. The path is a cumulative addition of the grid points passed through, each matrix element d being, from a continuity and monotonicity point of viewijThere are only three directions of travel: (i +1, j), (i, j +1) or (i +1, j + 1). The goal of the dynamic time warping algorithm is to minimize the following equation:
Figure BDA0003166120840000071
wherein D represents the distance between the action potential of the alternative motion unit and the prior template, DijIs the value expressed by the matrix elements on the path, K is the number of the matrix elements passed by the path, and in addition, the average amplitude value mean of the four prior templates is calculated respectivelyT。
3. And determining whether the action potential of the candidate motion unit is matched with the prior template or not by combining a dynamic time warping algorithm and the average amplitude of the prior template. The specific matching decision rule is as follows:
if D (x) is satisfiedl(t),yk(t))<meanT2, directly considering the action potential of the alternative motion unit to be matched with the prior template; otherwise, directly considering that the action potential of the alternative motion unit is not matched with the prior template;
4. and after the decision is finished, selecting the starting time of the action potential of the alternative motion unit as the issuing moment information of each motion unit obtained by the iteration according to the action potential of the alternative motion unit matched with the prior template.
Step 3.4, reconstructing the action potential waveform of the motion unit by adopting a pulse triggering average technology, wherein the process is as follows:
and taking the issuing moment of the action potential of the motion unit identified by each prior template as a trigger time point of the original electromyographic signal, sequentially triggering to obtain a plurality of trigger waveforms, and averaging the trigger waveforms to obtain a reconstructed waveform corresponding to the action potential of the motion unit. These reconstructed waveforms can be considered as the true motion unit action potential waveforms delivered by each motion unit.
Thus, a set of motion unit distribution information and waveform information is obtained, and the decomposition process and the result of the electromyographic signals collected by the superficial flexor muscles under the action of 15% of the maximum voluntary contraction force are shown in fig. 3. Since this iteration is based on the same amplitude factor, the firing sequence waveforms for this set of motion units can be considered to be at the same amplitude level.
Step 3.5, by comparing the amplitude factor AiAnd the size of the mean value mean (rest _ peaks) of the peak points of the reference signals judges whether the loop iteration process continues. In the process of stripping action potential of motion unit in each iteration, if AiIf the mean is higher than the mean (rest _ peaks), the visual residual signal is the updated electromyographic signal, and the step 3.2 is returned to continue the iteration; otherwise, the loop iteration is ended, and the subsequent evaluation step is carried out.
The implementation also comprises a step 4 of establishing a cross-correlation coefficient evaluation index based on the decomposition result to verify the effectiveness and feasibility of the electromyography decomposition method:
and (3) carrying out linear combination on the plurality of motion unit action potential sequences obtained by decomposition step by step according to two strategies of increasing the amplitude level from large to small and increasing the amplitude level from small to large to form a plurality of action combination sequences, sequentially calculating the cross correlation coefficient between the motion unit action potential combination sequences and the original electromyographic signals, and evaluating the quality of the decomposition algorithm by observing the change trend of the cross correlation coefficient. If the motion waveform reconstructed by the motion unit release moment and the pulse trigger averaging technology conforms to the release rule of the real motion unit, when more motion sequences are combined together, the cross correlation between the original electromyographic signal and the motion unit motion potential combination sequence is also enhanced. On the contrary, the movement unit action potential combination sequence is updated by continuously adding new action sequenceIn the process, once the cross correlation coefficient between the updated action combination sequence and the original electromyographic signals is found to be reduced, the fact that the newly added action sequence cannot be matched with the information issued by the real motion unit is shown. Cross correlation coefficient r of ith combinationiThe calculation formula of (a) is as follows:
Figure BDA0003166120840000081
where x (t) represents the original electromyographic signal, xMU(i)(t) represents the motion unit motion potential sequence of the ith combination,
Figure BDA0003166120840000091
and
Figure BDA0003166120840000092
then x (t) and x are represented respectivelyMU(i)(T), where T is the signal sampling duration and k represents the number of decomposition iterations.
According to the motion unit action potential sequence of each amplitude level obtained by decomposition, a motion unit action potential combination sequence is constructed by two strategies from large to small and from small to large respectively, the cross correlation coefficient of the combination sequence and the original electromyographic signal is calculated, the effectiveness and the reliability of decomposition are evaluated according to the coefficient change trend, and the cross correlation coefficient result graph is shown in fig. 4.

Claims (4)

1. A myoelectric motion unit decomposition method based on a prior template is characterized by comprising the following steps:
step 1: collecting myoelectric signals of 15% of volunteers under continuous action of maximum autonomic contractility and in a resting state;
step 2: denoising the electromyographic signals obtained in the step 1 by adopting wavelet packets;
and step 3: the myoelectricity simulation signal and the myoelectricity real signal are decomposed by a motion unit, and the method comprises the following specific steps:
step 3.1: selecting an n-order Hermite-Rodriguez function as a prior template to simulate the action potential waveform of the motion unit;
step 3.2: the electromyographic signals under the resting state are used as reference signals, and effective peak value detection is carried out on the electromyographic signals under the continuous action of 15% of maximum autonomic contraction force: selecting the mean value of the reference signals as a detection threshold, detecting effective peak values of the electromyographic signals under 15% of maximum autonomous contraction continuous acting force based on the detection threshold, and performing descending order according to the absolute value of the amplitude;
step 3.3: intercepting a signal segment near an effective peak value and carrying out matching processing on the signal segment and a prior template, wherein the matching processing is as follows:
A. after the effective peak value is determined, intercepting the vicinity of each effective peak by adopting a sliding window with a variable window length to obtain a part of signal segments to be analyzed as alternative motion unit action potentials;
B. calculating the distance D between the action potential of each alternative motion unit and four prior templates by adopting a dynamic time warping algorithm, and respectively calculating the average amplitude values of the four prior templatesmean T ；
C. Determining whether the action potential of the alternative motion unit is matched with the prior template or not by combining a dynamic time warping algorithm and the average amplitude of the prior template; the specific matching decision rule is as follows: if the distance D is less than one-halfmean T If so, the action potential of the alternative motion unit is considered to be matched with the prior template; otherwise, the signals are not matched;
D. selecting the starting time of the action potential of the alternative motion unit as a distribution moment according to the action potential of the motion unit matched with the prior template;
step 3.4: the pulse triggering average technology is adopted to reconstruct the action potential waveform of the motion unit, and the process is as follows:
taking the motion unit issuing moment identified by each prior template as a trigger time point of the original sEMG signal, and sequentially triggering to obtain a plurality of trigger waveforms;
averaging the trigger waveforms to obtain reconstructed waveforms corresponding to action potentials of the motion units;
step 3.5: and judging whether the loop iteration process continues or not by comparing the amplitude factor with the average value of the peak points of the reference signal: in the process of stripping the action potential of the motion unit in each iteration, if the amplitude is higher than the peak point average value of the reference signal, the visual residual signal is an updated electromyographic signal, and the step 3.2 is returned to continue the iteration; otherwise, the loop iteration ends.
2. The decomposition method for electromyographic motor units based on prior templates according to claim 1, wherein the first and second order Hermite-Rodriguez functions are adopted in step 3.1 to respectively correspond to the biphase waveform and the triphase waveform in the action potential waveform of the motor unit.
3. The a priori template based electromyographic motor unit decomposition method according to claim 1, further comprising initializing an amplitude factor in a Hermite-Rodriguez function of order n, the amplitude factor being a normalized motor unit action potential amplitude factor, in step 3.2.
4. A priori template based electromyographic motion unit decomposition method according to claim 3, wherein the amplitude factor is varied in an equal ratio series having a common ratio of 1/2 for describing the amplitude levels of different MUAPs.
CN202110805279.0A 2021-07-16 2021-07-16 Myoelectric movement unit decomposition method based on prior template Pending CN113397571A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110805279.0A CN113397571A (en) 2021-07-16 2021-07-16 Myoelectric movement unit decomposition method based on prior template

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110805279.0A CN113397571A (en) 2021-07-16 2021-07-16 Myoelectric movement unit decomposition method based on prior template

Publications (1)

Publication Number Publication Date
CN113397571A true CN113397571A (en) 2021-09-17

Family

ID=77686644

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110805279.0A Pending CN113397571A (en) 2021-07-16 2021-07-16 Myoelectric movement unit decomposition method based on prior template

Country Status (1)

Country Link
CN (1) CN113397571A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114052750A (en) * 2021-12-22 2022-02-18 杭州电子科技大学 Method for extracting brain muscle information transfer rule based on standard template electromyography
CN114343680A (en) * 2021-12-24 2022-04-15 杭州电子科技大学 Real-time decomposition method for dynamic surface electromyographic signals
CN114386479A (en) * 2021-12-09 2022-04-22 首都医科大学附属北京友谊医院 Medical data processing method and device, storage medium and electronic equipment
CN115840906A (en) * 2023-02-13 2023-03-24 博睿康科技（常州）股份有限公司 Decomposition method, decomposition model and signal analysis device for action potential

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109359619A (en) * 2018-10-31 2019-02-19 浙江工业大学之江学院 A kind of high density surface EMG Signal Decomposition Based method based on convolution blind source separating
CN110236538A (en) * 2019-06-21 2019-09-17 西安交通大学 A kind of automatic real-time detection method of movement starting point based on electromyography signal
CN111329476A (en) * 2020-03-04 2020-06-26 中国科学技术大学 Method and device for estimating muscle strength based on microscopic nerve driving information

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109359619A (en) * 2018-10-31 2019-02-19 浙江工业大学之江学院 A kind of high density surface EMG Signal Decomposition Based method based on convolution blind source separating
CN110236538A (en) * 2019-06-21 2019-09-17 西安交通大学 A kind of automatic real-time detection method of movement starting point based on electromyography signal
CN111329476A (en) * 2020-03-04 2020-06-26 中国科学技术大学 Method and device for estimating muscle strength based on microscopic nerve driving information

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
罗万国: "基于经验模板的表面肌电运动单元动作电位序列分解方法研究", 《中国优秀硕士学位论文全文数据库 医药卫生科技辑》 *

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114386479A (en) * 2021-12-09 2022-04-22 首都医科大学附属北京友谊医院 Medical data processing method and device, storage medium and electronic equipment
CN114386479B (en) * 2021-12-09 2023-02-03 首都医科大学附属北京友谊医院 Medical data processing method and device, storage medium and electronic equipment
CN114052750A (en) * 2021-12-22 2022-02-18 杭州电子科技大学 Method for extracting brain muscle information transfer rule based on standard template electromyography
CN114052750B (en) * 2021-12-22 2024-04-30 杭州电子科技大学 Brain muscle information transfer rule extraction method based on standard template myoelectricity decomposition
CN114343680A (en) * 2021-12-24 2022-04-15 杭州电子科技大学 Real-time decomposition method for dynamic surface electromyographic signals
CN114343680B (en) * 2021-12-24 2024-04-19 杭州电子科技大学 Real-time decomposition method for dynamic surface electromyographic signals
CN115840906A (en) * 2023-02-13 2023-03-24 博睿康科技（常州）股份有限公司 Decomposition method, decomposition model and signal analysis device for action potential

Similar Documents

Publication Publication Date Title
CN113397571A (en) Myoelectric movement unit decomposition method based on prior template
Chen et al. Adaptive real-time identification of motor unit discharges from non-stationary high-density surface electromyographic signals
Rogers et al. A comparison of EMG-based muscle fatigue assessments during dynamic contractions
CN110598676B (en) Deep learning gesture electromyographic signal identification method based on confidence score model
Dai et al. Independent component analysis based algorithms for high-density electromyogram decomposition: Experimental evaluation of upper extremity muscles
CN111329476B (en) Method and device for estimating muscle strength based on microscopic nerve driving information
Sidek et al. ECG biometric recognition in different physiological conditions using robust normalized QRS complexes
Zhou et al. MUAP number estimates in surface EMG: template-matching methods and their performance boundaries
Zhang et al. Muscle force estimation based on neural drive information from individual motor units
Hansen et al. Unmixing oscillatory brain activity by EEG source localization and empirical mode decomposition
Meng et al. Evaluation of decomposition parameters for high-density surface electromyogram using fast independent component analysis algorithm
Ning et al. Improve computational efficiency and estimation accuracy of multi-channel surface EMG decomposition via dimensionality reduction
CN116089851A (en) Self-adaptive updated online motion unit identification method
Zhou et al. Can standard surface EMG processing parameters be used to estimate motor unit global firing rate?
Kramberger et al. On the prediction of motor unit filter changes in blind source separation of high-density surface electromyograms during dynamic muscle contractions
Kranjec et al. Improved assessment of muscle excitation from surface electromyograms in isometric muscle contractions
Veer A flexible approach for segregating physiological signals
Xu et al. A novel and efficient surface electromyography decomposition algorithm using local spatial information
CN114767132B (en) Surface myoelectricity decomposition-based movement unit identification method
CN114052750B (en) Brain muscle information transfer rule extraction method based on standard template myoelectricity decomposition
Zhao et al. Online Decomposition of Surface Electromyogram into Individual Motor Unit Activities Using Progressive FastICA Peel-off
Ning et al. Multichannel surface EMG decomposition based on measurement correlation and LMMSE
CN115251845A (en) Sleep monitoring method for processing brain wave signals based on TB-TF-BiGRU model
Chen et al. A fast ECG diagnosis using frequency-based compressive neural network
CN113827257A (en) Surface myoelectricity decomposition method based on FastICA and contractility

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20210917

WD01 Invention patent application deemed withdrawn after publication