Lower limb muscle fatigue factor analysis method, system and storage medium fusing multi-muscle morphological characteristics

Abstract

The application discloses a lower limb muscle fatigue factor analysis method fusing multiple muscle shape features, and comprises the following steps: acquiring muscle shape features of a measured position based on three-dimensional coordinates of marker points of motion capture, constructing a multidimensional feature vector through the muscle shape features, and screening sensitive features sensitive to a fatigue state through variance analysis; obtaining combined features with strong sensitivity to the fatigue state and high interpretability through fatigue correlation analysis and joint feature screening; fusing the combined features into multidimensional fusion features through principal component analysis dimension reduction processing, and obtaining a fatigue factor based on the multidimensional fusion features. The application discloses a lower limb muscle fatigue factor analysis method fusing multiple muscle shape features, a system and a storage medium, synergic shape changes of the vastus lateralis muscle and the vastus medialis muscle are fused through a three-dimensional cone body volume model, fatigue degree quantitative indexes and fatigue cause contribution degrees are output in combination with an interpretable statistical model, and the lower limb muscle fatigue cause is explained.

Description

Technical Field

[0001] This invention relates to the field of biomechanics and muscle fatigue detection technology in smart healthcare, specifically to a method, system, and storage medium for analyzing lower limb muscle fatigue factors by integrating multiple muscle morphological characteristics. Background Technology

[0002] Lower limb muscle fatigue monitoring is a core component of safety control in human-computer interaction (such as exoskeleton robots and rehabilitation training equipment).

[0003] Existing muscle fatigue assessment techniques have the following main shortcomings: 1. Insufficient analysis of multiple muscle coordination: For example, in the existing technology (patent: method and system for recognizing lower limb muscle fatigue state based on multimodal feature fusion), only two-dimensional morphological parameters of a single muscle are provided, without considering the three-dimensional spatial relationship when multiple muscles coordinate to move, which results in the features not being able to fully reflect the overall shape of muscle coordinated contraction / stretching, and low utilization of spatial information.

[0004] 2. Lack of reliable quantitative indicators for muscle fatigue: For example, existing technologies for monitoring lower limb muscle fatigue lack reliable quantitative indicators. Traditional electromyography parameters are easily interfered with, and most technologies rely on machine learning, deep learning, and other models to mechanically classify muscle fatigue, which cannot effectively quantify the degree of fatigue and its causes, making it difficult to support the need for precise intervention. Summary of the Invention

[0005] This invention overcomes the shortcomings of existing technologies and provides a method, system, and storage medium for analyzing lower limb muscle fatigue factors by integrating multiple muscle morphological features. It integrates the synergistic morphological changes of the vastus lateralis and vastus medialis muscles through a three-dimensional frustum volume model, and combines an interpretable statistical model to output quantitative indicators of fatigue degree and contribution of fatigue causes, thereby realizing the explanation of the causes of lower limb muscle fatigue and providing a basis for muscle fatigue assessment and prediction.

[0006] To achieve the above objectives, the technical solution adopted by this invention is: a method for analyzing lower limb muscle fatigue factors by integrating multiple muscle morphological characteristics, comprising the following steps: Motion capture markers are placed at the measured location, and the three-dimensional coordinates of the motion capture markers are collected. The muscle morphology features at the measured location are obtained based on three-dimensional coordinates. These muscle morphology features include: longitudinal vertical distance of the cross section, radius of the muscle cross section circle, tensile length of the muscle surface, and area of ​​the triangle at the marked point. A multidimensional feature vector is constructed based on the muscle morphology features; the multidimensional feature vector includes the frustum volume and derived features, the derived features including: the difference in stretching of the inner and outer muscle surfaces, the difference in the area of ​​the inner and outer marker points, and the difference in the cross-sectional radius of the inner and outer muscles; Based on the multidimensional feature vector, features sensitive to fatigue state are selected through variance analysis and used as sensitive features. The correlation strength between each sensitive feature and different fatigue state levels is quantified by fatigue correlation analysis. Joint features were obtained through joint feature screening, which are highly sensitive to fatigue state and have high interpretability. The joint features are fused into multidimensional fused features by dimensionality reduction through principal component analysis (PCA), and the fatigue factor is obtained based on the multidimensional fused features.

[0007] In a preferred embodiment of the present invention, the arrangement of motion capture markers includes: attaching three motion capture markers to the center of the belly of each of the bilateral vastus lateralis (the main muscle responsible for knee extension) and vastus medialis (a synergistic muscle assisting in patellar stabilization), with the three markers at each muscle belly arranged equidistantly (laterally). Further, three motion capture markers are attached to the center of each muscle belly, with the three markers at each muscle belly arranged equidistantly laterally. Specifically, the y-axis in the coordinate system is parallel to the long axis of the femur, i.e., the femur is parallel to the y-axis. In a preferred embodiment of the present invention, obtaining the longitudinal vertical distance of the cross section includes: the projected distance between the center of the lateral muscle characteristic cross section and the center of the medial muscle characteristic cross section in the longitudinal direction of the measured position; Obtaining the radius of the muscle cross-section circle includes: determining the triangle enclosed by marker points on the muscle surface at the measured location, determining the circumcircle based on the triangle, and determining the radius of the muscle cross-section circle based on the circumcircle. The radius of the muscle cross-section circle is: Area is the area of ​​the triangle marked at point ; The acquisition of the muscle surface stretching length includes: the muscle surface stretching length is the arc length. R is the radius of the muscle cross-section circle, and C is the chord length. For the central angle, ; Obtaining the area of ​​the triangle at the marked point includes: , a, b, and c are the lengths of the three sides of the triangle. The triangle formed by three marked points serves as the smallest deformable unit, used to quantify the dynamic deformation of the muscle.

[0008] In a preferred embodiment of the present invention, the process of screening features sensitive to fatigue state based on the multidimensional feature vector through variance analysis includes the following steps: A mathematical model was established using one-way ANOVA: ; in, Let μ be the feature value of the j-th sample for the i-th fatigue level; μ is the mean of the overall sample set. The treatment effect for fatigue level i; The error is random and follows a set pattern. ; Sum of squares decomposition: ; ; ; Wherein, SST is the sum of squares of the deviations of all observations from the overall mean, a measure of the total variability of the data; SSB is the weighted sum of squares of the deviations of each group mean from the overall mean, reflecting the variability caused by different fatigue levels; SSW is the sum of squares of the deviations of each group's observations from the mean of that group, reflecting the variability caused by random errors within the group. The arithmetic mean of all observations. Let be the mean of the i-th group; The calculation of statistic F includes: Where N is the total sample size; k is the number of fatigue levels; The F-statistic was calculated using ANOVA to analyze whether there are significant differences in each sensitive feature among different fatigue state levels, and to quantify the relationship between the variation and random error caused by fatigue on each feature.

[0009] In a preferred embodiment of the present invention, quantifying the correlation strength between various sensitive features and different fatigue state levels through fatigue correlation analysis includes: the rank correlation coefficient formula for fatigue correlation analysis includes: ; where d i This represents the rank difference between the eigenvalue and the fatigue level.

[0010] In a preferred embodiment of the present invention, joint features that are highly sensitive to fatigue states and have high interpretability are obtained through joint feature screening, including: Through inter-group difference analysis and fatigue correlation analysis, we selected joint features that are highly sensitive to fatigue status and have high explanatory power. =P(F>F 观测 |H0 is true); where, F represents the probability of significance of the difference between groups, indicating the probability of observing the current data or a more extreme case when the assumption that the means of all fatigue level characteristics are equal is true; 观测 It is the actual calculated F value, and H0 is the null hypothesis that the mean values ​​of each fatigue level are equal; The selection criteria are: the selection conditions for the selected features are... ; in, With a significance level of 5%, The correlation is moderate to high.

[0011] In a preferred embodiment of the present invention, joint features are fused into multidimensional fused features through PCA dimensionality reduction processing, including: Z-score standardization is performed on each feature vector: Where z is the sample value after Z-score; x is the original sample value; μ z The mean of the original sample set; The standard deviation of the original sample set; Covariance matrix decomposition: , where W is the eigenvector matrix; First principal component extraction: ;in, This is the eigenvector corresponding to the largest eigenvalue.

[0012] In a preferred embodiment of the present invention, obtaining the fatigue factor based on multidimensional fusion features includes: By extracting the modulus of the fatigue factor, it can be determined that the larger the fatigue factor, the more severe the fatigue. when , but ; ; ; Where FatigueFactor is the original fatigue factor, which is calculated by PCA dimensionality reduction; Y is the fatigue level label.

[0013] In a preferred embodiment of the present invention, a computer system includes: Memory, used to store computer programs / instructions; A processor is used to execute the computer program / instructions to implement the steps of the method for analyzing lower limb muscle fatigue factors by incorporating multiple muscle morphological features.

[0014] In a preferred embodiment of the present invention, a computer program product includes a computer program / instruction that, when executed by a processor, implements the steps of a method for analyzing lower limb muscle fatigue factors by incorporating multiple muscle morphological features.

[0015] This invention addresses the deficiencies in the technical background, and the beneficial technical effects of this invention are: A method, system, and storage medium for analyzing lower limb muscle fatigue factors by integrating multiple muscle morphological features are proposed. The method integrates the synergistic morphological changes of the vastus lateralis and vastus medialis muscles through a three-dimensional frustum volume model, and combines interpretable statistical models to output quantitative indicators of fatigue degree and contribution of fatigue causes, thereby explaining the causes of lower limb muscle fatigue and providing a basis for muscle fatigue assessment and prediction.

[0016] The method for calculating the approximate frustum volume feature of the lower limb by integrating muscle morphology information, and the fatigue factor analysis method for quantifying the degree of lower limb muscle fatigue, are applicable to scenarios that require real-time monitoring of muscle fatigue, such as human-computer interaction and sports rehabilitation training. Attached Figure Description

[0017] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0018] Figure 1 This is a flowchart of the lower limb muscle fatigue factor analysis method integrating multiple muscle morphological features in a preferred embodiment of the present invention. Figure 2 This is a schematic diagram of the program flow for a lower limb muscle fatigue factor that integrates multiple muscle morphological features, according to a preferred embodiment of the present invention. Figure 3 This is a schematic diagram of the arrangement of muscle surface markers in a preferred embodiment of the lower limb muscle fatigue factor analysis method that integrates multiple muscle morphological features according to the present invention. Figure 4 This is a schematic diagram of the cone fitting of two muscles (vasomotor and vastus lateralis) in the lower limb muscle fatigue factor analysis method that integrates multiple muscle morphological features in a preferred embodiment of the present invention. Figure 5 This is a flowchart of statistical analysis and fatigue factor calculation in the lower limb muscle fatigue factor analysis method that integrates multiple muscle morphological features in a preferred embodiment of the present invention. Figure 6 This is a distribution diagram of the first principal component coefficients in the covariance matrix decomposition of the lower limb muscle fatigue factor analysis method that integrates multiple muscle morphological features in a preferred embodiment of the present invention. Figure 7 This is a distribution diagram of fatigue factors under different fatigue states in a preferred embodiment of the lower limb muscle fatigue factor analysis method that integrates multiple muscle morphological features; Figure 8 This is a schematic diagram of muscle morphology features in the lower limb muscle fatigue factor analysis method that integrates multiple muscle morphology features in a preferred embodiment of the present invention. Figure 9 This is a schematic diagram of the marking point arrangement of a lower limb muscle cross-section that incorporates multiple muscle morphological features in a preferred embodiment of the present invention. Figure 10 This is a schematic diagram of the sampling area of ​​the marker points of the lower limb muscles that integrate multiple muscle morphological features in a preferred embodiment of the present invention. Figure 11 This is a t-SNE visualization of the feature distribution under the condition of wearing an exoskeleton in a preferred embodiment of the present invention; Figure 12This is a t-SNE visualization of the feature distribution under the condition of not wearing an exoskeleton in a preferred embodiment of the present invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. These drawings are simplified schematic diagrams, which are only used to illustrate the basic structure of the present invention and therefore only show the components relevant to the present invention.

[0020] It should be noted that if directional indicators (such as up, down, bottom, top, etc.) are involved in the embodiments of the present invention, these directional indicators are only used to explain the relative positional relationship and movement of the components in a specific posture. If the specific posture changes, the directional indicators will also change accordingly. The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, features defined with "first" and "second" may explicitly or implicitly include one or more of that feature. Unless otherwise explicitly specified and limited, the terms "set," "connected," and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal connection of two components. For those skilled in the art, the specific meaning of the above terms in the present invention can be understood according to the specific circumstances.

[0021] Example 1: A method for analyzing lower limb muscle fatigue factors by integrating multiple muscle morphological features, comprising the following steps: Motion capture markers are placed at the measured location, and the three-dimensional coordinates of the motion capture markers are collected. The muscle morphology features at the measured location are obtained based on three-dimensional coordinates. These features include: longitudinal vertical distance of the cross section, radius of the muscle cross section circle, tensile length of the muscle surface, and area of ​​the triangle at the marker point. A multidimensional feature vector is constructed based on the muscle morphology features. The multidimensional feature vector includes the frustum volume and derived features. The derived features include: the difference in stretching of the inner and outer muscle surfaces, the difference in the area of ​​the inner and outer marker points, and the difference in the cross-sectional radius of the inner and outer muscles. Based on multidimensional feature vectors, features sensitive to fatigue state are selected through variance analysis and used as sensitive features. The correlation strength between each sensitive feature and different fatigue state levels is quantified by fatigue correlation analysis. Joint features were obtained through joint feature screening, which are highly sensitive to fatigue state and have high interpretability. The joint features are fused into multidimensional fused features by dimensionality reduction through principal component analysis (PCA), and the fatigue factor is obtained based on the multidimensional fused features.

[0022] Example 2: A method for analyzing lower limb muscle fatigue factors by integrating multiple muscle morphological features, comprising the following steps: Step S1: Place motion capture markers M at the measured location and collect the three-dimensional coordinates of the motion capture markers.

[0023] Specifically, the arrangement of motion capture markers includes: attaching three motion capture markers to the center of the belly of the vastus lateralis (the main muscle responsible for knee extension) and vastus medialis (a synergistic muscle that helps stabilize the patella) on both sides, with the three markers at each muscle belly arranged equidistantly laterally. In this embodiment, the preferred spacing between the motion capture markers is 4 cm.

[0024] Step S2: Obtain the muscle morphological features of the measured position based on three-dimensional coordinates. The muscle morphological features include: longitudinal vertical distance of the cross section, radius of the muscle cross section circle, tensile length of the muscle surface, and area of ​​the marked point triangle.

[0025] Specifically, obtaining the longitudinal vertical distance of the cross section includes: the projected distance between the center of the lateral muscle characteristic cross section and the center of the medial muscle characteristic cross section in the longitudinal direction of the measured position.

[0026] Specifically, obtaining the radius of the muscle cross-section circle includes: determining the triangle enclosed by marker points on the muscle surface at the measured location, determining the circumcircle based on the triangle, and determining the radius of the muscle cross-section circle based on the circumcircle. The radius of the muscle cross-section circle is: Area represents the area of ​​the triangle marked at point .

[0027] Specifically, obtaining the muscle surface stretch length includes: the muscle surface stretch length is the arc length. R is the radius of the muscle cross-section circle, and C is the chord length. For the central angle, ;in Figure 3 , Figure 8 , Figure 9 In the middle, D is the longitudinal vertical distance of the cross section; N1 is the vastus medialis muscle; N2 is the vastus lateralis muscle; and N3 is the femur.

[0028] Specifically, obtaining the area of ​​the triangle marked by the point includes: , a, b, and c are the lengths of the three sides of the triangle. The triangle formed by three marked points serves as the smallest deformable unit, used to quantify the dynamic deformation of the muscle.

[0029] Step S3: Construct a multi-dimensional feature vector based on muscle morphology features; the multi-dimensional feature vector includes the frustum volume and derived features, including: the difference in surface stretching between the inner and outer sides of the muscle, the difference in area of ​​the inner and outer marked points, and the difference in cross-sectional radius between the inner and outer sides of the muscle. For example... Figure 4As shown, the upper end face of the cone is the cross-sectional circle of the vastus lateralis muscle, and the lower end face is the cross-sectional circle of the vastus medialis muscle; the vertical line in the middle of the cone is the longitudinal vertical distance of the cross-section.

[0030] Step S4: Based on the multidimensional feature vector, use variance analysis to screen features that are sensitive to fatigue state as sensitive features.

[0031] Specifically, the process of screening features sensitive to fatigue state based on multidimensional feature vectors through variance analysis includes the following steps: A mathematical model was established using one-way ANOVA: ;in, Let μ be the feature value of the j-th sample for the i-th fatigue level; μ is the mean of the overall sample set. The treatment effect for fatigue level i; The error is random and follows a set pattern. σ is the within-group standard deviation in the analysis of variance model, which is for samples within a certain fatigue level. Physically, it represents the random fluctuation range between different samples within the same fatigue level (same group). If σ is small, it indicates that the data is highly concentrated; if σ is large, it indicates that the data is dispersed. The model parameter is estimated from the data and used to determine whether the factor (fatigue level) is significant.

[0032] Sum of squares decomposition: ; ; ; Where, n i Meaning: The number of samples in the i-th group (i.e., the i-th fatigue level). SST is the sum of squares of the deviations of all observations from the overall mean, a measure of the total variability of the data; SSB is the weighted sum of squares of the deviations of each group mean from the overall mean, reflecting the variability caused by different fatigue levels; SSW is the sum of squares of the deviations of each group's observations from the mean of that group, reflecting the variability caused by random errors within the group. The arithmetic mean of all observations. Let be the mean of the i-th group.

[0033] The calculation of statistic F includes: Where N is the total sample size; k is the number of fatigue levels; The statistic F was calculated using ANOVA to analyze whether there are significant differences in each sensitive feature among different fatigue state levels, and to quantify the relationship between the variation and random error caused by fatigue on each feature, providing a significant basis for subsequent feature selection.

[0034] The determination of "whether there is a significant difference" is achieved through hypothesis testing, by comparing the magnitudes of "differences caused by different fatigue levels" and "differences caused by random errors." First, the null hypothesis (assuming that the mean values ​​of characteristics are equal across all fatigue states) needs to be proposed. Then, the F-statistic is calculated to obtain the corresponding p-value. If the p-value is less than the preset significance level (usually 0.05), it indicates that the probability of observing a difference between groups is extremely low; therefore, the null hypothesis is rejected, and the conclusion is that there are statistically significant differences in the characteristic values ​​between different fatigue states. Conversely, if the p-value is large, it indicates that the difference between groups is likely due to random fluctuations and cannot be considered significant.

[0035] Among them, the relationship between the statistic F and significance analysis is that the F value quantitatively describes the ratio between the "fatigue grade effect" and the "random error", which is the core criterion for judging whether the difference between groups is significant.

[0036] According to the formula for calculating the statistic F, when fatigue state has a significant impact on the eigenvalue, the between-group variation (SSB) will be much greater than the within-group random variation (SSW), resulting in an F value much greater than 1; conversely, if the eigenvalue is not affected by fatigue state, the between-group variation mainly comes from random error, and the F value is close to 1.

[0037] When the F value is large enough (i.e., the P value is less than 0.05), it indicates that the difference between different fatigue levels exceeds the range of random fluctuations, and this feature has significant differences between different fatigue states; conversely, it indicates that this feature is not sensitive to fatigue states.

[0038] Step S5: Quantify the correlation strength between each sensitive feature and different fatigue state levels through fatigue correlation analysis.

[0039] Specifically, fatigue correlation analysis (Spearman's method) quantifies the strength of the correlation between each sensitive feature and different fatigue state levels, including the rank correlation coefficient formula for fatigue correlation analysis: ; where d i ρ represents the rank difference between the eigenvalue and the fatigue level. In fatigue correlation analysis, ρ is used to measure the strength of the monotonic correlation between the eigenvalue and the fatigue level.

[0040] Step S6: Obtain joint features that are highly sensitive to fatigue state and have high interpretability through joint feature screening.

[0041] Specifically, joint features that are highly sensitive to fatigue state and have high interpretability are obtained through joint feature screening, including: Through inter-group difference analysis and fatigue correlation analysis, we selected joint features that are highly sensitive to fatigue status and have high explanatory power. =P(F>F 观测|H0 is true); where, F is the probability of significance of the difference between groups, i.e., P, which represents the probability of observing the current data or a more extreme case when the assumption that the means of all fatigue level characteristics are equal is true; 观测 It is the actual calculated F value, and H0 is the null hypothesis that the mean values ​​of each fatigue level are equal; The selection criteria are: the selection conditions for the selected features are... ;in, With a significance level of 5%, For a moderate to high correlation, a threshold p < 0.05 is set to select features that show statistically significant differences between different fatigue levels, ensuring that the selected features can effectively distinguish different fatigue states. Spearman's correlation analysis is used for fatigue correlation; ρSpearman is the rank correlation coefficient between the feature value and the fatigue level, measuring the strength and direction of the monotonic relationship.

[0042] Step S7: The joint features are fused into multidimensional fused features by principal component analysis (PCA) dimensionality reduction, and the fatigue factor is obtained based on the multidimensional fused features.

[0043] Among them, analysis of variance is used to quantify the relationship between the variation and random error caused by fatigue on each feature, specifically expressed as the ratio of between-group variation to within-group variation, i.e., the statistic F. For example... Figure 5 As shown, the F-value is converted into the corresponding p-value and compared with a preset significance level (e.g., 0.05). If P < 0.05, it indicates that the feature difference between different fatigue levels is statistically significant, and this feature can be retained as a sensitive feature for subsequent fatigue state identification models; if P ≥ 0.05, it indicates that the feature difference is not significant, and this feature cannot effectively distinguish different fatigue states, so it is removed. Figure 7 An asterisk between different fatigue states indicates a significant difference between the two states. Figure 7 The Chinese text explains the independent samples t-test: "*" indicates p < 0.05; "**" indicates p < 0.01; "***" indicates p < 0.001. Figure 7 The "***" indicates that the p-value between all fatigue states is less than 0.001. Furthermore, as... Figure 7 As shown, the horizontal axis represents the actual fatigue level groupings, divided into three groups: no fatigue, moderate fatigue, and severe fatigue; the vertical axis represents the constructed fatigue factor values. The contour is violin-shaped, with the width representing the sample distribution density of the corresponding fatigue factor value: the wider the contour, the more samples fall within that value range. The nested box plots demonstrate the statistical characteristics of the data; the boxes cover samples from the 25th to 75th percentiles within each group, and the median line represents the group median, showing the central tendency and dispersion range of the three fatigue factor groups. Figure 7The distribution shows that the fatigue factor score is clearly positively correlated with the actual degree of fatigue (i.e., the higher the degree of fatigue, the higher the fatigue factor value). The fatigue factor in the no-fatigue group is generally negative, with a median of around -1.5; the moderate fatigue group rises to the positive range, with a median of about 0.5; and the severe fatigue group further rises to around 1.3, indicating that the fatigue factor can well reflect the degree of fatigue in the sample.

[0044] Specifically, PCA dimensionality reduction is used to fuse joint features into multidimensional fused features, including: By normalizing features to the same scale, preserving the maximum variation information through covariance matrix decomposition, eliminating multicollinearity among features through principal component analysis, and extracting and calculating fatigue factors, the selected multidimensional features are fused into a single fatigue factor, preserving the maximum variation information and solving the collinearity problem. To eliminate the influence of different dimensions, the comparability of each feature is ensured.

[0045] Specifically, Z-score standardization is performed on each feature vector: Where z is the sample value after Z-score; x is the original sample value; μ z σ is the mean of the original sample set; z The standard deviation of the original sample set; Covariance matrix decomposition: Where W is the eigenvector matrix; the covariance matrix Cov(Z) is calculated for the standardized multidimensional fatigue feature data Z; subsequently, the characteristic equation Cov(Z)⋅w is solved through eigenvalue decomposition. i =λ i w i This yields an orthogonal matrix W composed of eigenvectors and a diagonal matrix Λ composed of eigenvalues. Each column of W represents a principal component direction, revealing the linear combination of the original fatigue features; the diagonal elements λ of Λ... i This indicates the magnitude of variance along the corresponding principal component direction, quantifying the degree to which each comprehensive index explains the variation in fatigue state.

[0046] First principal component extraction: ;in, Let Z be the eigenvector corresponding to the largest eigenvalue. Here, Z is a standardized, preprocessed multidimensional sensitive feature data matrix. Specifically, Z is an n×p matrix, where n is the total number of samples and p is the number of sensitive feature dimensions retained after variance analysis and correlation screening. Each column of matrix Z undergoes Z-score standardization, i.e., subtracting the global mean of the feature and dividing by the standard deviation, ensuring that each feature dimension has a uniform dimension and scale. Standardization ensures that subsequent principal component analysis is not affected by the original magnitude of the features, allowing the first principal component direction w1 to accurately reflect the direction of maximum data variation. By linearly combining the standardized feature matrix Z with the first principal component direction vector w1, a comprehensive quantitative index reflecting the overall fatigue state, FatigueFactor, can be obtained. Furthermore, the first principal component is the first comprehensive index that can explain the total difference / variation of the original data to the greatest extent. The horizontal axis represents all the original features used; the vertical axis represents the linear weight coefficient of each original feature when constructing the first principal component, visually demonstrating the magnitude and direction of the influence of each original feature on the first principal component. Figure 6 The first principal component coefficients reflect the contribution of different features to the fatigue factor (the higher the absolute value of the coefficient, the greater the contribution). The sign of the coefficient represents the direction of change of the original feature and the fatigue factor score. Positive coefficient: The original feature and the fatigue factor score change in the same direction: the larger the value of the original feature, the higher the final score of the fatigue factor, and the two are positively correlated. Negative coefficient: The original feature and the fatigue factor score change in opposite directions: the larger the value of the original feature, the lower the final score of the fatigue factor, and the two are negatively correlated.

[0047] Specifically, fatigue factors obtained based on multidimensional fusion features include: By extracting the modulus of the fatigue factor, it can be determined that the larger the fatigue factor, the more severe the fatigue. when , but ; ; ; Where FatigueFactor is the original fatigue factor, which is calculated by PCA dimensionality reduction; Y is the fatigue level label. Right now The corrected fatigue factor; W adj This is the corrected PCA weight eigenvector; This represents the Pearson correlation coefficient between the original fatigue factor and the fatigue level label. A value less than 0 indicates a positive correlation between the fatigue factor and fatigue severity (a higher fatigue factor value indicates more severe fatigue); a value greater than 0 indicates a negative correlation between the fatigue factor and fatigue severity (a higher fatigue factor value indicates less severe fatigue). When the model approaches -FatigueFactor, the weights w need to be reversed simultaneously to maintain model consistency. In one embodiment, the fatigue level labels are defined as follows: no fatigue after 0 squats; moderate fatigue after 40 squats; and severe fatigue after 80 squats.

[0048] Furthermore, in this embodiment, the fatigue factor distribution of multiple subjects is comprehensively analyzed, and a fatigue factor below -0.5 is defined as no fatigue state; a fatigue factor between -0.5 and 0.5 is a moderate fatigue state; and a fatigue factor greater than 0.5 is a severe fatigue state.

[0049] Example 3, based on Example 2, further includes: In step S1, three markers at the belly of each muscle are arranged equidistantly in the horizontal direction, with a spacing of 40 mm in this embodiment.

[0050] In obtaining the radius of the muscle cross-section circle in step S2: the radius of the muscle cross-section circle is fitted with the contour of the marker points to quantify the change in the cross-sectional area of ​​the muscle belly. The triangle enclosed by the marker points on the muscle surface at the measured location is determined, and the circumcircle is determined based on this triangle. The radius of the muscle cross-section circle is then calculated based on this circumcircle. Let R be the radius of the cross-section circle of the left vastus lateralis muscle belly. L-VL Let R be the radius of the circle formed by the cross-section of the left vastus medialis muscle belly. L-VM Let R be the radius of the circle formed by the cross-section of the right vastus lateralis muscle belly. R-VL Let R be the radius of the circle formed by the cross-section of the left vastus medialis muscle belly. R-VM The radius of the muscle cross-section circle is determined based on the circumcircle. The radius of the muscle cross-section circle is: Area represents the area of ​​the triangle marked at point .

[0051] In obtaining the muscle surface stretch length in step S2: the deformation displacement between the marker points for quantifying the muscle surface stretch length reflects the passive stretching state of the muscle. The arc length determined by the three marker points on the muscle surface is calculated based on the aforementioned circumcircle geometry; this arc length is considered the muscle surface stretch length. Let Arc be the stretch length of the left vastus lateralis muscle. L-VL Let Arc be the length of the surface stretch of the left vastus medialis muscle. L-VM Let Arc be the length of the surface stretch of the right vastus lateralis muscle. R-VL Let Arc be the length of the surface stretch of the left vastus medialis muscle. R-VM The stretching length of the muscle surface is the arc length: R is the radius of the muscle cross-section circle, and C is the chord length. For the central angle, .

[0052] In obtaining the area of ​​the triangle formed by the three marker points in step S2: the triangle is used as the smallest deformable unit to quantify the dynamic deformation of the muscle. Let the area of ​​the triangle formed by the left vastus lateralis muscle be denoted as Area. L-VL Let the area of ​​the deltoid region of the left vastus medialis muscle be denoted as Area. L-VM Let the area of ​​the right vastus lateralis deltoid muscle be Area. R-VL Let the area of ​​the deltoid region of the left vastus medialis muscle be denoted as Area. R-VM . , a, b, and c are the lengths of the three sides of the triangle. The triangle formed by three marked points serves as the smallest deformable unit, used to quantify the dynamic deformation of the muscle.

[0053] In obtaining the longitudinal vertical distance of the cross-section in step S2, the projected distance between the center of the characteristic cross-section of the vastus lateralis muscle and the center of the characteristic cross-section of the vastus medialis muscle along the longitudinal axis of the lower limb (i.e., the direction parallel to the horizontal plane and extending along the long axis of the femur) is defined as the absolute value of the difference in the y-axis direction of the intermediate marker points at each muscle belly, for ease of calculation. High = |y VL -y VM |;wherein, y VL The y-coordinate of the midpoint marker point of the vastus lateralis muscle; y VM The y-coordinate value of the midpoint marker point of the vastus medialis muscle.

[0054] The tension difference δ between the inner and outer muscle surfaces in step S3 arc δ represents the absolute value of the difference in surface tension between the vastus medialis and vastus lateralis muscles on both sides. Let δ be the surface tension difference between the left and right vastus medialis and vastus lateralis muscles. L-arc The tensile difference between the inner and outer surfaces of the right side muscle is δ. R-arc ; δ arc =|Arc VL -Arc VM |;Among them, Arc VL The surface stretch length of the vastus lateralis muscle; Arc VM The length of the surface stretch of the vastus medialis muscle.

[0055] The area difference δ between the inner and outer marked points in step S3 area δ represents the absolute value of the difference in surface tension between the vastus medialis and vastus lateralis muscles on both sides. Let δ be the surface tension difference between the left and right vastus medialis and vastus lateralis muscles. L-area The tensile difference between the inner and outer surfaces of the right side muscle is δ. R-area ; δ area =|AreaVL -Area VM |. Among them, Area VL Area is the triangle marked by the vastus lateralis muscle. VM Let be the area of ​​the triangle marked by the vastus medialis muscle.

[0056] The difference in radii δ between the inner and outer muscle cross sections in step S3 R δ represents the absolute value of the difference in surface tension between the vastus medialis and vastus lateralis muscles on both sides. Let δ be the surface tension difference between the left and right vastus medialis and vastus lateralis muscles. L-R The tensile difference between the inner and outer surfaces of the right side muscle is δ. R-R ; δ R =|R VL -R VM |. Among them, R VL R is the radius of the circle formed by the cross-section of the vastus lateralis muscle. VM The radius of the cross-section of the vastus medialis muscle is denoted as .

[0057] The frustum volume V in step S3 is formed by the characteristic cross-sectional circle R of the vastus lateralis and vastus medialis muscles. VL R VM High is the vertical distance between the bottom surface and the two cross sections, and High is the height, forming the volume of the truncated cone.

[0058] .

[0059] Example 4: A computer system comprising: Memory, used to store computer programs / instructions; A processor is used to execute the computer program / instructions to implement the steps of the method for analyzing lower limb muscle fatigue factors by incorporating multiple muscle morphological features in any of the embodiments of Example 1 to Example 3.

[0060] Example 5: A computer program product, comprising: A computer program / instruction, when executed by a processor, implements the steps of the method for analyzing lower limb muscle fatigue factors that integrates multiple muscle morphological features in any of the embodiments 1 to 3.

[0061] Working principle: A method, system, and storage medium for analyzing lower limb muscle fatigue factors by integrating multiple muscle morphological features are disclosed. This method integrates the synergistic morphological changes of the vastus lateralis and vastus medialis muscles through a three-dimensional frustum volume model. Combined with an interpretable statistical model, it outputs quantitative indicators of fatigue level and the contribution of fatigue causes, thereby explaining the causes of lower limb muscle fatigue and providing a basis for muscle fatigue assessment and prediction. This invention, through a method for calculating the approximate frustum volume features of the lower limb by integrating muscle morphological information and a fatigue factor analysis method for quantifying lower limb muscle fatigue, is applicable to scenarios requiring real-time monitoring of muscle fatigue levels, such as human-computer interaction and sports rehabilitation training.

[0062] The above specific embodiments are specific support for the concept proposed in this invention, and should not be used to limit the scope of protection of this invention. Any equivalent changes or modifications made on the basis of this technical solution in accordance with the technical concept proposed in this invention shall still fall within the scope of protection of this invention.

Claims

1. A method for analyzing lower limb muscle fatigue factors by integrating multiple muscle morphological features, characterized in that: Includes the following steps: Motion capture markers are placed at the measured location, and the three-dimensional coordinates of the motion capture markers are collected. The muscle morphology features of the measured location are obtained based on three-dimensional coordinates. The muscle morphology features include: longitudinal vertical distance of the cross section, radius of the muscle cross section circle, tensile length of the muscle surface, and area of ​​the triangle of the marker point. A multidimensional feature vector is constructed based on the muscle morphology features; the multidimensional feature vector includes the frustum volume and derived features, the derived features including: the difference in stretching of the inner and outer muscle surfaces, the difference in the area of ​​the inner and outer marker points, and the difference in the cross-sectional radius of the inner and outer muscles; Based on the multidimensional feature vector, features sensitive to fatigue state are selected through variance analysis and used as sensitive features. The correlation strength between each sensitive feature and different fatigue state levels is quantified by fatigue correlation analysis. Joint features were obtained through joint feature screening, which are highly sensitive to fatigue state and have high interpretability. Principal component analysis is used to reduce the dimensionality of the joint features and fuse them into multidimensional fused features. The fatigue factor is then obtained based on the multidimensional fused features.

2. The method for analyzing lower limb muscle fatigue factors by integrating multiple muscle morphological features according to claim 1, characterized in that: The arrangement of the motion capture markers includes: attaching multiple motion capture markers to the center of the belly of the vastus lateralis and vastus medialis muscles on both sides, with multiple markers at equal intervals at the belly of each muscle.

3. The method for analyzing lower limb muscle fatigue factors by integrating multiple muscle morphological features according to claim 2, characterized in that: The acquisition of the longitudinal vertical distance of the cross section includes: the projected distance between the center of the lateral muscle characteristic cross section and the center of the medial muscle characteristic cross section in the longitudinal direction of the measured position; Obtaining the radius of the muscle cross-section circle includes: determining the triangle enclosed by marker points on the muscle surface at the measured location, determining the circumcircle based on the triangle, and determining the radius of the muscle cross-section circle based on the circumcircle. The radius of the muscle cross-section circle is: Area is the area of ​​the triangle marked at point ; The acquisition of the muscle surface stretching length includes: the muscle surface stretching length is the arc length. R is the radius of the muscle cross-section circle, and C is the chord length. For the central angle, ; Obtaining the area of ​​the triangle at the marked point includes: , a, b, and c are the lengths of the three sides of the triangle. The triangle formed by three marked points serves as the smallest deformable unit, used to quantify the dynamic deformation of the muscle.

4. The method for analyzing lower limb muscle fatigue factors by integrating multiple muscle morphological features according to claim 3, characterized in that: The steps involved in screening features sensitive to fatigue state based on the multidimensional feature vector through variance analysis are as follows: A mathematical model was established using one-way ANOVA: ； in, Let μ be the feature value of the j-th sample for the i-th fatigue level; μ is the mean of the overall sample set. The treatment effect for fatigue level i; The error is random and follows a set pattern. ; Sum of squares decomposition: ； ； ； Wherein, SST is the sum of squares of the deviations of all observations from the overall mean, a measure of the total variability of the data; SSB is the weighted sum of squares of the deviations of each group mean from the overall mean, reflecting the variability caused by different fatigue levels; SSW is the sum of squares of the deviations of each group's observations from the mean of that group, reflecting the variability caused by random errors within the group. The arithmetic mean of all observations. Let be the mean of the i-th group; The calculation of statistic F includes: Where N is the total sample size; k is the number of fatigue levels; The F-statistic was calculated using ANOVA to analyze whether there are significant differences in each sensitive feature among different fatigue state levels, and to quantify the relationship between the variation and random error caused by fatigue on each feature.

5. The method for analyzing lower limb muscle fatigue factors by integrating multiple muscle morphological features according to claim 4, characterized in that: The correlation strength between various sensitive characteristics and different fatigue state levels is quantified through fatigue correlation analysis, including the rank correlation coefficient formula for fatigue correlation analysis: ; where d i This represents the rank difference between the eigenvalue and the fatigue level.

6. The method for analyzing lower limb muscle fatigue factors by integrating multiple muscle morphological features according to claim 5, characterized in that: Joint features with high sensitivity to fatigue state and high interpretability were obtained through joint feature screening, including: Through inter-group difference analysis and fatigue correlation analysis, we selected joint features that are highly sensitive to fatigue status and have high explanatory power. =P(F>F 观测 |H0 is true); where, F represents the probability of significance of the difference between groups, indicating the probability of observing the current data or a more extreme case when the assumption that the means of all fatigue level characteristics are equal is true; 观测 It is the actual calculated F value, and H0 is the null hypothesis that the mean values ​​of each fatigue level are equal; The selection criteria are: the selection conditions for the selected features are... ; in, With a significance level of 5%, The correlation is moderate to high.

7. The method for analyzing lower limb muscle fatigue factors by integrating multiple muscle morphological features according to claim 6, characterized in that: Principal component analysis (PCA) is used to reduce dimensionality and fuse joint features into multidimensional fused features, including: Z-score standardization is performed on each feature vector: Where z is the sample value after Z-score; x is the original sample value; μ z The mean of the original sample set; The standard deviation of the original sample set; Covariance matrix decomposition: , where W is the eigenvector matrix; First principal component extraction: ;in, This is the eigenvector corresponding to the largest eigenvalue.

8. The method for analyzing lower limb muscle fatigue factors by integrating multiple muscle morphological features according to claim 7, characterized in that: Fatigue factors obtained based on multidimensional fusion features include: By extracting the modulus of the fatigue factor, it can be determined that the larger the fatigue factor, the more severe the fatigue. when , but ; ； ； Where FatigueFactor is the original fatigue factor, which is calculated by PCA dimensionality reduction; Y is the fatigue level label.

9. A computer system, characterized in that, include: Memory, used to store computer programs / instructions; A processor for executing the computer program / instructions to implement the steps of the method for analyzing lower limb muscle fatigue factors that integrates multiple muscle morphological features, as described in any one of claims 1-8.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the steps of the method for analyzing lower limb muscle fatigue factors that integrates multiple muscle morphological features as described in any one of claims 1-8.

Images