Method for calculating average turning amplitude of surface electromyogram signals based on clustering method

Abstract

The invention provides a method for calculating the average turning amplitude of surface electromyographic signals based on a clustering method. The method is implemented through the following steps that S1, original electromyographic signals are collected and preprocessed; s2, automatically grouping signal amplitude features through a clustering algorithm; s3, calculating an average turning amplitude; and S4, removing the relaxed state data. According to the method for calculating the average turning amplitude of the surface electromyographic signals based on the clustering method, effective signals and noise can be distinguished, the noise is removed, and electromyographic baseline drift interference can be effectively avoided; and a clustering algorithm is applied, so that electromyographic signal amplitudes can be automatically classified according to similarities, and manual intervention errors are avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of surface electromyography signal processing, and in particular to a method for calculating an average turning amplitude of a surface electromyography signal based on a clustering method. Background Art

[0002] Surface electromyography (sEMG) is the combined effect of electrical activity in superficial muscles and nerve trunks on the surface of the skin. It reflects the activity of the neuromuscular system and is typically in the frequency range of 30-350Hz. Muscles generate bioelectricity during movement or contraction. This electrical activity is recorded by electrodes on the surface of the skin and then amplified and processed to produce sEMG signals.

[0003] Surface EMG signal turning point analysis mainly involves analyzing the turning points of surface EMG signals to evaluate the changes and characteristics of muscle activity.

[0004] Traditional surface electromyography (SEMG) turning amplitude analysis relies on manual identification, which is highly subjective and inefficient. While some turning amplitude analysis methods that integrate automated algorithms, such as the threshold method, have emerged on the market, they are easily affected by noise, resulting in inaccurate amplitude calculations. For example, changes in muscle contraction intensity lead to amplitude fluctuations, and the fixed threshold method is prone to misjudgment (e.g., misjudging weak contraction signals as noise). Summary of the Invention

[0005] In view of the above-mentioned shortcomings of the prior art, the purpose of the present invention is to provide a method for calculating the average turning amplitude of surface electromyography signals based on clustering method, so as to solve the problems in the prior art of being easily interfered by noise and having inaccurate amplitude calculation.

[0006] To achieve the above and other related purposes, the present invention provides a method for calculating the average turning amplitude of surface electromyographic signals based on a clustering method, which is implemented by the following steps: S1, collect the original electromyographic signal and preprocess it; S2, automatically grouping signal amplitude features through clustering algorithm; S3. Calculate the average turning amplitude; S4. Eliminate the relaxed state data.

[0007] In some embodiments of the present invention, in step S1, the preprocessing includes at least the following steps: S11, reduce baseline drift noise through filtering; S12, pre-processing of 50Hz notch; S13. Eliminate interference signals that exceed the range.

[0008] In some embodiments of the present invention, in step S11, the filter is a high-pass filter, which can reduce or even eliminate low-frequency noise such as baseline drift.

[0009] In some embodiments of the present invention, step S12 is implemented using a notch filter. A notch filter (BSF) is a type of band-stop filter. A notch filter is specifically designed to suppress specific narrowband frequency components, and is typically used to remove 50Hz power frequency interference. Its operating principle is to design a notch filter centered at 50Hz to effectively suppress signals at and near 50Hz frequencies while having little impact on other frequency components.

[0010] In some embodiments of the present invention, in step S13, the interference signals mainly include two types: (1) Assuming that the EMG value range of a normal person's muscle contraction does not exceed M millivolts, the EMG value exceeding M millivolts is determined to be an interference signal, where 3≤M≤5; (2) Assuming that a certain EMG value is within the EMG value range P of muscle contraction, and there are N data before and after it that are both within the EMG value amplitude range Y of muscle relaxation, then the EMG value is determined to be an interference signal, where 0.03 mV ≤ P ≤ 2, 0 mV < Q ≤ 0.03 mV, and N ≥ 3 and is an integer.

[0011] In some embodiments of the present invention, step S2 specifically includes the following steps: S21, performing a modulo operation on the data pre-processed in step S1; S22. Use a clustering algorithm to classify the initial data.

[0012] In some embodiments of the present invention, in step S22, the clustering algorithm uses the DBSCAN clustering algorithm of Python software; The main code of the clustering algorithm is as follows: dbscan = DBSCAN(eps=1.2, min_samples = 5) yhat = dbscan.fit_predict(X) in: EPS represents the size of the neighborhood, which is used to determine the core points and boundary points; min_samples represents the minimum number of points required to form a high-density area.

[0013] In some embodiments of the present invention, the specific steps of step S3 are: Calculate the average turning amplitude in each cluster generated by clustering; The specific steps of step S4 are: When judging whether the average turning amplitude of each class is less than the preset value R, if the average turning amplitude of one class or several classes is less than the preset value R, then the data of this class or these classes are judged as relaxed state data and are eliminated; Among them, 0.02 mV ≤ R ≤ 0.05 mV.

[0014] In some embodiments of the present invention, after step S4, the following steps are further included: S5. Extract data: extract motion status data.

[0015] In some embodiments of the present invention, after step S5, the following steps are further included: S6. Result analysis: Based on the motion state data extracted in step S5, the time domain index and frequency domain index of the motion process are calculated and the result analysis is performed.

[0016] As described above, the method for calculating the average turning amplitude of surface electromyographic signals based on the clustering method of the present invention has the following beneficial effects: 1. The present invention provides a method for calculating the average turning amplitude of surface electromyographic signals based on a clustering method. The method can distinguish effective signals from noise, remove noise, and effectively avoid electromyographic baseline drift interference. 2. The method for calculating the average turning amplitude of surface electromyographic signals based on clustering method of the present invention applies a clustering algorithm, which can automatically classify the amplitudes of electromyographic signals according to similarity, thus avoiding manual intervention errors; 3. The average range of muscle relaxation amplitudes varies among different groups of people. The two sets of data calculated from the clustering results, exercise state and relaxation state, can be dynamically adjusted based on the smaller set of data in the clustering results. The clustering-based surface electromyography signal average turning amplitude calculation method of the present invention applies a dynamic threshold design, adaptively adjusts the effective amplitude range based on the clustering results, and improves anti-interference ability. 4. The method for calculating the average turning amplitude of surface electromyographic signals based on clustering method of the present invention has good clinical value: The surface electromyography (EMG) values ​​of the gastrocnemius or soleus muscles of rehabilitation patients are measured in real time during a period of exercise. The surface electromyography values ​​of the patients during exercise are identified using a clustering method. Muscle fatigue is quantitatively assessed by analyzing time-domain and frequency-domain indicators. The surface electromyography (EMG) values ​​of the deltoid and biceps brachii muscles were measured in real time during a stroke patient's exercise. The surface electromyography values ​​during exercise were identified using a clustering method. The degree of muscle activation, fatigue resistance, and rehabilitation training effects were monitored by analyzing time-domain and frequency-domain indicators. In addition, the analysis results can be displayed as soon as the patient finishes exercising, which makes the analysis time short and efficient (single analysis takes ≤1 second). BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 The surface electromyography values ​​collected with a sampling interval of 100 milliseconds are clustered after modulo and normalization.

[0018] Figure 2 Based on Figure 1 The classified results are plotted as muscle lines using the original coordinate system.

[0019] Figure 3 The surface electromyography values ​​collected with a sampling interval of 70 milliseconds are clustered after modulo and normalization.

[0020] Figure 4 Based on Figure 3 The classified results are plotted as muscle lines using the original coordinate system. DETAILED DESCRIPTION

[0021] The following describes the embodiments of the present invention through specific examples. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments. The details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other unless they conflict.

[0022] It should be noted that the illustrations provided in the following embodiments are merely schematic illustrations of the basic concept of the present invention. Therefore, the illustrations only show components related to the present invention and are not drawn according to the number, shape, and size of components in actual implementation. In actual implementation, the type, quantity, and proportion of each component may be changed arbitrarily, and the component layout may also be more complex.

[0023] See also Figure 1-Figure 4 The present invention provides a method for calculating the average turning amplitude of surface electromyographic signals based on the clustering method in the following embodiment. The method for calculating the average turning amplitude of surface electromyographic signals based on the clustering method belongs to the technical field of surface electromyographic signal processing, and can solve the problems of the existing technology that it is easily affected by noise interference and the amplitude calculation is inaccurate.

[0024] The present invention provides a method for calculating the average turning amplitude of surface electromyographic signals based on a clustering method.

[0025] The method for calculating the average turning amplitude of the surface electromyography signal based on the clustering method is implemented by the following steps: S1, collect the original electromyographic signal and preprocess it; S2, automatically grouping signal amplitude features through clustering algorithm; S3. Calculate the average turning amplitude; S4. Eliminate the relaxed state data.

[0026] In this embodiment, in step S1, the raw electromyographic signal is collected by using an existing surface electromyographic device in conjunction with an electrode sheet, and the preprocessing includes at least the following steps: S11, reduce baseline drift noise through filtering; S12, pre-processing of 50Hz notch; S13. Eliminate interference signals that exceed the range.

[0027] In this embodiment, in step S11, the filter is a high-pass filter, which can reduce or even eliminate low-frequency noise such as baseline drift.

[0028] In this embodiment, step S12 is implemented using a notch filter. A notch filter (BSF) is a type of band-stop filter. A notch filter is specifically designed to suppress specific narrowband frequency components, and is typically used to remove 50Hz power frequency interference. Its operating principle is to design a notch filter with a center frequency of 50Hz to effectively suppress signals at and near 50Hz while having little impact on other frequency components.

[0029] In this embodiment, in step S13, the interference signals mainly include two types: Type (1): Assuming that the myoelectric value range of a normal person's muscle contraction does not exceed M millivolts, the myoelectric value exceeding M millivolts is determined to be an interference signal, where 3≤M≤5, for example, M=3 in this embodiment; Type (2): Assuming that a certain EMG value is within the EMG value range P of muscle contraction, and N data before and after it are both within the EMG value amplitude range Q of muscle relaxation, then the EMG value is determined to be an interference signal, where 0.03 mV ≤ P ≤ 2, 0 mV < Q ≤ 0.03 mV, and N ≥ 3 and is an integer. For example, in this embodiment, P = 0.1 mV, Q = 0.03 mV, and N = 4. For example: Figure 2 The electromyographic value corresponding to the horizontal coordinate of about 3.5 can be determined as an interference signal.

[0030] In this embodiment, step S2 specifically includes the following steps: S21, performing a modulo operation on the data pre-processed in step S1; S22. Use a clustering algorithm to classify the initial data.

[0031] In this embodiment, in step S22, the clustering algorithm uses the DBSCAN clustering algorithm of Python software; The DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm is a very popular clustering algorithm, which is particularly suitable for discovering clusters of arbitrary shapes and can handle noise points.

[0032] Working principle of DBSCAN algorithm: DBSCAN algorithm divides the points in the data set into three states: core points (CorePoints), border points (BorderPoints) and noise points (NoisePoints).

[0033] The main code of the clustering algorithm is as follows: dbscan = DBSCAN(eps=1.2, min_samples = 5) yhat = dbscan.fit_predict(X) in: EPS represents the size of the neighborhood, which is used to determine the core points and boundary points; min_samples represents the minimum number of points required to form a high-density area.

[0034] In this embodiment, the specific steps of step S3 are: Calculate the average turning amplitude within each cluster generated by clustering.

[0035] In this embodiment, the specific steps of step S4 are: When judging whether the average turning amplitude of each class is less than the preset value R, if the average turning amplitude of one class or several classes is less than the preset value R, then the data of this class or these classes are judged as relaxed state data and are eliminated; Wherein, 0.02 mV≤R≤0.05 mV, for example, in this embodiment, R=0.03 mV.

[0036] In this embodiment, after step S4, the following steps are further included: S5. Extract data: extract motion status data.

[0037] In this embodiment, after step S5, the following steps are further included: S6. Result analysis: Based on the motion state data extracted in step S5, the time domain index and frequency domain index of the motion process are calculated and the result analysis is performed.

[0038] During muscle contraction, the sEMG data amplitude changes greatly, while during muscle relaxation, the sEMG data amplitude changes less. Figure 1 and Figure 2 is a set of experimental data, Figure 3 and Figure 4 A set of experimental data. Figure 1 and Figure 3 is the result produced after step S22 of the present invention, Figure 2 and Figure 4 It is a line graph that draws the clustering results using the original coordinate system.

[0039] The method for calculating the average turning amplitude of surface electromyographic signals based on the clustering method of the present invention has at least the following advantages over the prior art: 1. The present invention provides a method for calculating the average turning amplitude of surface electromyographic signals based on a clustering method. The method can distinguish effective signals from noise, remove noise, and effectively avoid electromyographic baseline drift interference. The modular design allows for easy integration into existing EMG equipment.

[0040] 2. The method for calculating the average turning amplitude of surface electromyographic signals based on clustering method of the present invention applies a clustering algorithm, which can automatically classify the amplitudes of electromyographic signals according to similarity, thus avoiding manual intervention errors; 3. The average range of muscle relaxation amplitudes varies among different groups of people. The two sets of data calculated from the clustering results, exercise state and relaxation state, can be dynamically adjusted based on the smaller set of data in the clustering results. The clustering-based surface electromyography signal average turning amplitude calculation method of the present invention applies a dynamic threshold design, adaptively adjusts the effective amplitude range based on the clustering results, and improves anti-interference ability. 4. The method for calculating the average turning amplitude of surface electromyographic signals based on clustering method of the present invention has good clinical value: For example, during a rehabilitation patient's exercise, the surface EMG values ​​of the gastrocnemius or soleus muscles can be measured in real time. The surface EMG values ​​at the moment of exercise can be identified through clustering, and the muscle fatigue can be quantitatively assessed by analyzing time domain and frequency domain indicators. Another example: During a stroke patient's exercise, the surface EMG values ​​of the deltoid and biceps brachii are measured in real time. The surface EMG values ​​during exercise are identified using clustering methods. The degree of muscle activation, fatigue resistance, and rehabilitation training effects are monitored by analyzing time-domain and frequency-domain indicators. In addition, the analysis results can be displayed as soon as the patient finishes exercising, which makes the analysis time short and efficient (single analysis takes ≤1 second).

[0041] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the present invention. Anyone skilled in the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by one of ordinary skill in the art without departing from the spirit and technical principles disclosed herein are intended to be covered by the claims of the present invention.

Claims

1. A method for calculating the average turning amplitude of surface electromyographic signals based on clustering method, characterized in that: This is achieved by following these steps: S1, collect the original electromyographic signal and preprocess it; S2, automatically grouping signal amplitude features through clustering algorithm; S3. Calculate the average turning amplitude; S4. Eliminate the relaxed state data.

2. The method for calculating the average turning amplitude of surface electromyographic signals based on the clustering method according to claim 1, characterized in that: In step S1, the preprocessing includes at least the following steps: S11, reduce baseline drift noise through filtering; S12, pre-processing of 50Hz notch; S13. Eliminate interference signals that exceed the range.

3. The method for calculating the average turning amplitude of surface electromyographic signals based on the clustering method according to claim 2, characterized in that: In step S11 , the filter is a high-pass filter.

4. The method for calculating the average turning amplitude of surface electromyographic signals based on clustering method according to claim 2, characterized in that: Step S12 is implemented by a notch filter.

5. The method for calculating the average turning amplitude of surface electromyographic signals based on clustering method according to claim 2, characterized in that: In step S13, the interference signals mainly include two types: (1) Assuming that the EMG value range of a normal person's muscle contraction does not exceed M millivolts, the EMG value exceeding M millivolts is determined to be an interference signal, where 3≤M≤5; (2) Assuming that a certain EMG value is within the EMG value range P of muscle contraction, and there are N data before and after it that are both within the EMG value amplitude range Y of muscle relaxation, then the EMG value is determined to be an interference signal, where 0.03 mV ≤ P ≤ 2, 0 mV < Q ≤ 0.03 mV, and N ≥ 3 and is an integer.

6. The method for calculating the average turning amplitude of surface electromyographic signals based on clustering method according to any one of claims 1 to 5, characterized in that: Step S2 specifically includes the following steps: S21, performing a modulo operation on the data pre-processed in step S1; S22. Use a clustering algorithm to classify the initial data.

7. The method for calculating the average turning amplitude of surface electromyographic signals based on clustering method according to claim 6, characterized in that: In step S22, the clustering algorithm uses the DBSCAN clustering algorithm of Python software; The main code of the clustering algorithm is as follows: dbscan = DBSCAN(eps=1.2, min_samples = 5) yhat = dbscan.fit_predict(X) in: EPS represents the size of the neighborhood, which is used to determine the core points and boundary points; min_samples represents the minimum number of points required to form a high-density area.

8. The method for calculating the average turning amplitude of surface electromyographic signals based on clustering method according to claim 6, characterized in that: The specific steps of step S3 are: Calculate the average turning amplitude in each cluster generated by clustering; The specific steps of step S4 are: When judging whether the average turning amplitude of each class is less than the preset value R, if the average turning amplitude of one class or several classes is less than the preset value R, then the data of this class or these classes are judged as relaxed state data and are eliminated; Among them, 0.02 mV ≤ R ≤ 0.05 mV.

9. The method for calculating the average turning amplitude of surface electromyographic signals based on the clustering method according to claim 8, characterized in that: After step S4, the method further includes: S5. Extract data: extract motion status data.

10. The method for calculating the average turning amplitude of surface electromyographic signals based on clustering method according to claim 9, characterized in that: After step S5, the method further includes: S6. Result analysis: Based on the motion state data extracted in step S5, the time domain index and frequency domain index of the motion process are calculated and the result analysis is performed.

Images