Muscle mass measurement method, device, equipment, medium and product
By collecting and analyzing biceps ultrasound images and physiological information, extracting texture features and using neural network models, the problem that the existing technology cannot comprehensively evaluate muscle strength and endurance is solved, and accurate assessment and analysis of muscle mass is achieved.
Patent Information
- Application Number
- CN202410746696.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-11
- Publication Date
- 2024-11-12
AI Technical Summary
The existing fitness platforms and muscle assessment platforms cannot fully and accurately reflect the detailed information of the human body's muscle strength, endurance and flexibility, and lack detailed muscle and body assessment functions.
By collecting biceps ultrasound images, physiological information and muscle weight data, texture features are extracted, their correlation with muscle strength is determined, clustered labeling is performed, and training is performed based on neural network models to achieve measurement and evaluation of muscle mass.
Accurate evaluation and analysis of muscle mass is achieved, and labels of good performance and poor performance are provided, which improves the accuracy and comprehensiveness of muscle evaluation.
Smart Images

Figure CN118924240A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of muscle mass assessment, and in particular to a muscle mass measurement method, device, equipment, medium and product. Background Art
[0002] Ultrasonic muscle texture analysis technology can accurately analyze muscles through high-definition ultrasonic images and obtain various data such as the fibers and elasticity of each individual muscle.
[0003] Prior art 1 is a muscle strength comprehensive evaluation instrument, which can accurately measure and record the strength and balance state of each muscle group in the whole body by isometric contraction of muscles in various fixed postures, and make an overall evaluation of the muscle function of the whole body, especially the functional state of the waist and back muscles. The muscle strength comprehensive test and evaluation system is controlled by a microprocessor, and the liquid crystal display can display the measured value of strength. The analysis software has powerful data storage, management and analysis functions, and intuitively reflects the result report in the form of numbers and graphics.
[0004] Prior art 2 is an isokinetic muscle strength assessment system PrimaISO. The isokinetic muscle strength assessment training system manufactured by Italian Iteco is a compact, economical and practical isokinetic assessment training system that integrates rehabilitation, training and assessment. The electronically controlled hydraulic brake device is used to ensure the smoothness of movement and the safety of the patient during use. The data recording and training functions of this product are faster, more effective and more accurate than before. Biofeedback-guided training is not only an effective treatment method for correcting muscle defects; it is also a successful method for strengthening coordination ability, which is difficult to achieve with previous training methods. The test data can be displayed and saved in real time, and the final report can be printed (a computer and printer are required). The user software is based on the Windows platform and is easy to use. Games and instant feedback help therapists mobilize the enthusiasm of patients to cooperate. This instrument can be used for: testing and strengthening muscle strength, improving muscle function, improving stability and coordination, conducting endurance training, strengthening joint mobility, continuous functional assessment, and the objectivity of the results is determined and can be used for research. These features make this device a universal device that can be used in the fields of functional assessment, rehabilitation and training. Practice has shown that the device can also be used by extremely frail patients.
[0005] Prior art 3 provides a waist and abdominal muscle strength assessment trainer, which can effectively relieve back pain and lower back pain, increase waist and abdominal strength, control the patient's incontinence, and strengthen and balance the back muscles on both sides of the trunk. This product is designed for rehabilitation training of the deep muscles of the waist and back, lumbar spine, and lumbar sacral muscles. The product is equipped with a tilting seat to promote the joint movement of the pelvic and anterior lumbar sacral muscles, and can train the range of motion of large joints (superficial muscles) and small joints (deep muscles), thereby restoring the normal function and stability of the patient's pelvis and lumbar spine.
[0006] Prior art 4 provides a comprehensive muscle strength test trainer, which provides muscle strength test data for professional physicians to evaluate muscle strength, and provides four modes (passive, isometric, isokinetic, isotonic) for muscle strength rehabilitation training to improve the flexibility, stability and control ability of each joint. At the same time, according to the distribution of human joints and tendons, corresponding accessories can be used to perform exercise training and muscle strength testing of shoulder joints, elbow joints, wrist joints, hip joints, knee joints, ankle joints and hip joints.
[0007] However, most of the above-mentioned fitness platforms and muscle assessment platforms only provide evaluation of a small amount of data. Traditional assessment methods cannot fully and accurately reflect the muscles and physical condition of the human body. For example, observing appearance, measuring weight, measuring BMI, etc. cannot accurately reflect more detailed information such as muscle strength, endurance, flexibility, etc. There is a lack of comprehensive and detailed muscle and body assessment functions, and it cannot accurately analyze and evaluate the user's physical condition. Summary of the invention
[0008] The purpose of this application is to provide a muscle mass measurement method, device, equipment, medium and product that can perform comprehensive muscle and body detection, analysis and evaluation.
[0009] To achieve the above objectives, this application provides the following solutions:
[0010] In a first aspect, the present application provides a muscle mass measurement method, the muscle mass measurement method comprising:
[0011] Collecting multiple data indicators; the data indicators include: biceps brachii ultrasound image, physiological information, muscle weight and muscle strength;
[0012] Extracting texture features of the ultrasound image;
[0013] Determining the correlation between the texture feature and the muscle strength and muscle weight;
[0014] Cluster the features with significant correlation to obtain clustering results;
[0015] Labeling the muscles based on the clustering results; the labels include: good performance and poor performance;
[0016] Based on the texture features and labels, a neural network model is trained;
[0017] The collected biceps brachii ultrasound images are input into the trained neural network model to obtain the measurement results of muscle mass.
[0018] Optionally, the muscle mass measurement method further comprises:
[0019] Save the trained neural network model as a pb model;
[0020] Transplant the pb model to Android Studio and configure the environment;
[0021] Design the graphical user interface, embed it into Android Studio, and configure the corresponding interface;
[0022] The trained neural network model is called and encapsulated as a mobile phone applet.
[0023] Optionally, extracting the texture features of the ultrasound image specifically includes the following steps:
[0024] Performing data cleaning on the biceps brachii ultrasound image set;
[0025] Firstly, a binarization operation is performed on the biceps brachii ultrasound image, and then an enhancement process is performed on the biceps brachii ultrasound image, and a contrast enhancement is performed using a histogram;
[0026] The enhanced biceps brachii ultrasound image is filtered using median filtering, Gaussian filtering and mean filtering;
[0027] The filtered biceps brachii ultrasound image is sharpened by Laplace sharpening method to extract the edge;
[0028] Perform threshold segmentation on the biceps brachii ultrasound image after edge extraction to remove redundant background;
[0029] The entropy value of the two-dimensional image after threshold segmentation is calculated using the Shannon entropy formula;
[0030] The biceps brachii ultrasound image was segmented by morphological operation after removing the redundant background, and the muscle structure features were extracted.
[0031] The canny detection operator is used to extract the contour of the muscle area after image segmentation. The pixel values within the contour are calculated first, and then the ultrasound image is calibrated to convert the pixel area into the actual physical area.
[0032] Use local binary pattern LBP to extract the energy value lbp_energy and LBP entropy lbp_entropy in the LBP feature vector;
[0033] The image gray level co-occurrence matrix GLCM is used to extract the contrast mean, correlation mean and energy mean.
[0034] Optionally, the Shannon entropy formula is expressed as follows:
[0035]
[0036] Among them, H represents the two-dimensional image entropy value, L represents the total number of gray levels, i represents, and p(i) represents the probability of image gray level i.
[0037] Optionally, the expression of the energy value lbp_energy in the LBP feature vector is as follows:
[0038]
[0039] Among them, E represents the energy value in the LBP feature vector, N represents the total number of LBP histogram bins, q(t) represents the value of the tth bin in the LBP histogram, and t is the histogram bin value index value.
[0040] Optionally, the expression of the contrast mean value contrast_mean is as follows:
[0041] contrast_mean=∑ i ∑ j (ij) 2 P(i,j)
[0042] Among them, i and j both represent gray levels, and P(i,j) represents the elements of gray levels i and j in the GLCM.
[0043] Optionally, the expression of the correlation mean correlation_mean is as follows:
[0044]
[0045] Where i and j both represent gray levels, P(i,j) represents the element of gray level i and j in GLCM, μ i and μ j denote the mean of the GLCM rows and columns, respectively;
[0046] The expression of the energy average value energy_mean is as follows:
[0047] energy_mean=∑ i ∑ j P(i,j) 2
[0048] In a second aspect, the present invention provides a muscle mass measurement device, the muscle mass measurement device comprising:
[0049] A data index collection module, used to collect multiple data indexes; the data indexes include: biceps brachii ultrasound image, physiological information, muscle weight and muscle strength;
[0050] A texture feature extraction module, used to extract texture features of the ultrasound image;
[0051] A correlation determination module, used to determine the correlation between the texture feature and the muscle strength and muscle weight;
[0052] The clustering module is used to cluster the features with significant correlation to obtain clustering results;
[0053] A label marking module, used to label the muscles based on the clustering results; the labels include: good performance and poor performance;
[0054] A training module, used for training a neural network model based on the texture features and labels;
[0055] The muscle mass prediction module is used to input the collected biceps brachii ultrasound image into the trained neural network model to obtain the measurement result of muscle mass.
[0056] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any one of the above-mentioned muscle mass measurement methods.
[0057] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any one of the above-mentioned muscle mass measurement methods.
[0058] In a fifth aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of any one of the above-mentioned muscle mass measurement methods.
[0059] According to the specific embodiments provided in this application, this application discloses the following technical effects:
[0060] The present application provides a muscle mass measurement method, device, equipment, medium and product, which accurately analyzes the user's muscles and body indicators through ultrasonic equipment, extracts muscle texture features using cloud servers and deep learning technology, and realizes accurate evaluation and analysis of user muscle data. BRIEF DESCRIPTION OF THE DRAWINGS
[0061] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0062] Figure 1 This is a flow chart of a muscle mass measurement method in one embodiment of the present application;
[0063] Figure 2 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION
[0064] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0065] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.
[0066] See also Figure 1 The present invention provides a muscle mass measurement method, the muscle mass measurement method comprising:
[0067] Step 1: Collect multiple data indicators; the data indicators include: biceps brachii ultrasound image, physiological information, muscle weight and muscle strength.
[0068] Specifically, the volunteers sat on a chair and used a retractable bandage to fix the trunk to the back of the chair to keep the posture unchanged during the experiment. The upper arm was horizontal to the ground and perpendicular to the trunk. The forearm and upper arm were kept at the same horizontal plane and were also tied and fixed to a forearm bracket with a bandage. The elbow was bent at about 90° and the hand held a vertical rod welded to the forearm bracket. When the experiment started, the volunteers were asked to hold the vertical rod and flex their elbows. The torque value was transmitted through the rod and recorded by the data acquisition card in the computer. The B-ultrasound probe was fixed by a homemade bracket and placed on the belly of the biceps brachii and perpendicular to the skin surface to obtain a cross-sectional image perpendicular to the direction of the biceps brachii muscle fibers. A certain distance was maintained between the probe and the biceps brachii, and a large amount of coupling agent filled the gap so that it did not contact the skin on the surface of the biceps brachii during the entire experiment to avoid affecting the test results. The B-ultrasound image was transmitted to the video capture card, which digitized it and input it into the computer at a sampling rate of 8MHz. After cleaning the skin with alcohol, the surface electromyography electrodes were attached between the B-ultrasound probe and the elbow joint, with the electrodes parallel to the direction of the biceps muscle fibers and as close to the B-ultrasound probe as possible. The distance between the electrode pairs was about 20 mm, and the reference electrode was attached to the ulnar head. The surface electromyography signals were amplified 10 times and filtered at 10 to 800 Hz, then collected by the data acquisition card in the computer, and the data was amplified 10 times, with a sampling rate of 4 kHz.
[0069] Step 2: Extracting texture features of the ultrasound image.
[0070] The specific steps include:
[0071] Step 2.1: performing data cleaning on the biceps brachii ultrasound image set.
[0072] Data cleaning can be used to remove images of poor quality, scale images, and reduce the amount of calculation.
[0073] Step 2.2: firstly perform a binarization operation on the biceps brachii ultrasound image, and then perform enhancement processing on the biceps brachii ultrasound image, and use a histogram to perform contrast enhancement.
[0074] Step 2.3: Filter the enhanced biceps brachii ultrasound image using median filtering, Gaussian filtering and mean filtering.
[0075] Step 2.4: sharpen the edges of the filtered biceps ultrasound image using the Laplace sharpening method and perform edge extraction.
[0076] Step 2.5: Perform threshold segmentation on the biceps brachii ultrasound image after edge extraction to remove redundant background.
[0077] Step 2.6: Use the Shannon entropy formula to calculate the entropy value of the two-dimensional image after threshold segmentation.
[0078] Among them, the expression of Shannon entropy formula is as follows:
[0079]
[0080] Among them, H represents the two-dimensional image entropy value, L represents the total number of gray levels, and p(i) represents the probability of image gray level i.
[0081] Step 2.7: Use morphological operations to segment the biceps brachii ultrasound image after removing redundant background and extract muscle structure features.
[0082] Step 2.8: Use the canny detection operator to extract the contour of the muscle area after image segmentation. First calculate the pixel value within the contour, then calibrate the ultrasound image and convert the pixel area into the actual physical area.
[0083] Step 2.9: Use the local binary pattern LBP to extract the energy value lbp_energy and LBP entropy lbp_entropy in the LBP feature vector.
[0084] Among them, the expression of energy value lbp_energy is as follows:
[0085]
[0086] Among them, E represents the energy value in the LBP feature vector, N represents the total number of LBP histogram bins, q(t) represents the value of the tth bin in the LBP histogram, and t is the histogram bin value index value.
[0087] The LBP histogram is an array, and bin is an interval into which it is divided.
[0088] Step 3.0: Use the image gray level co-occurrence matrix GLCM to extract the contrast mean, correlation mean and energy mean.
[0089] Among them, the expression of contrast mean contrast_mean is as follows:
[0090] contrast_mean=∑ i ∑ j (ij) 2 P(i,j)
[0091] Among them, i and j both represent gray levels, and P(i,j) represents the elements of gray levels i and j in the GLCM.
[0092] GLCM stands for Gray Level Co-occurence matrix.
[0093] The expression of correlation mean correlation_mean is as follows:
[0094]
[0095] Where i and j both represent gray levels, P(i,j) represents the element of gray level i and j in GLCM, μ i and μ j denote the means of the GLCM rows and columns, respectively.
[0096] The expression of energy mean energy_mean is as follows:
[0097] energy_mean=∑ i ∑ j P(i,j) 2 .
[0098] Step 3: Determine the correlation between the texture feature and the muscle strength and muscle weight.
[0099] The correlation results are shown in the following table:
[0100]
[0101] Table 3-1 Correlation between texture features and muscle strength (** represents P<0.01, significant correlation)
[0102] It can be seen that the above texture parameters have significant correlations with muscle weight and muscle strength. Domestic and foreign researchers have discussed the correlation between muscle cross-sectional area or muscle thickness and muscle weight and muscle strength. The present invention proposes for the first time to use muscle texture characteristics as indicators for evaluating muscle weight and muscle strength, and through practice, it is found that these texture characteristics are significantly correlated with muscle performance.
[0103] Step 4: Cluster the features with significant correlation to obtain clustering results.
[0104] Since the present invention proposes for the first time to use texture features as an indicator for evaluating muscle performance, it is necessary to discuss the range of these features that corresponds to better muscle performance. In order to explore the internal connections between the features, clustering is used to group similar features into one category, thereby distinguishing between good and poor muscle performance features. The results of clustering are used as the basis for the formulation of data set labels for machine learning data set labels. Normally, labels are manually calibrated, but clustering is used in the invention to scientifically and efficiently label the data set, thereby improving the final prediction accuracy.
[0105] The specific results of clustering are as follows:
[0106] Cluster labels: Each volunteer's data points are assigned to different clusters, each cluster representing a different category of muscle performance. Each data point (i.e., texture feature of each subject) is given a cluster label indicating the cluster to which it belongs. For example, an array of length N, where each element is a cluster number (from 0 to K-1), N is the number of subjects, and K is the number of selected clusters.
[0107] Cluster categories: Based on the clustering results, we can analyze which clusters represent better muscle performance and which represent worse muscle performance. These cluster categories will be used to label each data point (such as "good performance" or "poor performance") in preparation for subsequent machine learning.
[0108] Cluster centroid: The center point (centroid) of each cluster, representing the typical features of the cluster. The centroid is a point in the feature space that can be used to understand the feature distribution and importance of each cluster.
[0109] Step 5: Label the muscles based on the clustering results; the labels include: good performance and poor performance.
[0110] Step 6: Based on the texture features and labels, the neural network model is trained.
[0111] Step 7: Input the collected biceps brachii ultrasound image into the trained neural network model to obtain the measurement result of muscle mass.
[0112] The biggest advantage of deep learning is that there is no need to do feature engineering, that is, there is no need to manually extract texture features. The input data can only provide ultrasound images and labels obtained by clustering. The present invention uses a classic convolutional neural network to establish a model for automatic classification of muscle performance. The initial test accuracy of the model reached 83%. Since the model established by deep learning belongs to a "black box" and no one has manually marked out the focus (features) of learning, deep neural networks require more learning samples, that is, more ultrasound images and labels, compared to traditional machine learning that requires artificial features. Therefore, with the deepening of practical projects and the expansion of data volume, the classification accuracy will continue to improve.
[0113] Step 8: Save the trained neural network model as pb model.
[0114] Step 9: Port the pb model to Android Studio and configure the environment.
[0115] Step 10: Design the graphical user interface, embed it into Android Studio, and configure the corresponding interfaces.
[0116] Step 11: Call the trained neural network model and encapsulate it into a mobile app.
[0117] Specifically, before the experiment began, the volunteers were first familiarized with the isometric contraction movement, and then performed three isometric maximum voluntary contractions (MVC), and recorded the torque value, and the maximum torque value was selected as the MVC torque value of the volunteer. At the beginning of the experiment, the volunteers were required to flex their elbows and exert force to 80% of the MVC torque value, and maintain the torque value for 30 seconds through the visual feedback of the torque value on the display screen. The same experiment was conducted a total of 3 times, with at least 5 minutes of rest between each test. When the volunteer exerted force to 80% of the MVC, B-ultrasound images, torque and surface electromyography signals were immediately collected.
[0118] All collected B-ultrasound images and surface electromyographic signals are processed offline by a self-developed system. Muscle deformation is obtained by continuous analysis of B-ultrasound images. Correlation tracking is a tracking method based on image similarity measurement, which searches for the area closest to the reference image template in the current image. It does not require segmentation of the target and background, has low requirements on image quality, can work normally under low signal-to-noise ratio conditions, is insensitive to all other scenes that are not similar to the selected tracking target image, can track smaller targets and a special part of the target area or targets with poor contrast, and is a better tracking method. The traditional correlation tracking algorithm calculates the mutual correlation coefficient and uses point-to-point multiplication and accumulation as the similarity judgment criterion, and the amount of calculation is very large. Combined with the characteristic that the target area of the image moves less during the experiment, the present invention adopts the mean absolute difference method (MAD) with small amount of calculation, and the expression of its measurement function is:
[0119]
[0120] Among them, f(x,y) is the reference image template, and g(x,y) is the current image. When processing the image, first add two rectangular boxes under the skin and on the cortical surface of the first frame of the B-ultrasound image as the upper and lower boundaries of the biceps brachii. The distance between the center points of the two boxes is used as the muscle thickness of the biceps brachii, and the muscle thickness calculated in the first frame of the image is used as the initial muscle thickness. Then, the images in the two rectangular boxes in the first frame of the image are used as template one and template two respectively, and two-dimensional cross-correlation calculations are performed on both templates in the second frame of the image to find the best matching position. At the same time, the image in the rectangular box at the best matching position in the second frame of the image is updated as a new template, and the next frame of the image is matched, and so on until the last frame of the B-ultrasound image. For the muscle thickness of the biceps brachii in each frame of the image obtained, the muscle thickness change rate is calculated according to the following formula:
[0121]
[0122] Where d0 is the muscle thickness of the first frame image, and d is the muscle thickness of the current frame image. For the surface electromyographic signal, the root mean square (RMS) and median frequency (MDF) of each segment of the electromyographic signal corresponding to each frame image are calculated, and then the change rate of all RMS and MDF is further calculated using linear regression.
[0123] From the above experimental results, it can be found that the biceps brachii has shown fatigue in the isometric contraction experiment, and the muscle thickness is also increasing. The main reason for the increase in muscle thickness is the recruitment of muscle motor units. The motor unit is the smallest functional unit of muscle contraction, which consists of a motor neuron and the muscle fibers it controls. When the muscle force is very small, the motor units of small axons are recruited first. When the muscle force slowly increases, the motor units of large axons are gradually recruited and joined. This is the size principle. This kind of motor unit recruitment sequence can produce smooth contraction. The recruitment of motor units is the main reason for the increase in muscle thickness during muscle training. Continuous force makes the contraction force of the motor unit decrease, and the volunteers are required to maintain 80% of the MVC. In order to make up for the lack of muscle contraction force, more and more motor units will be recruited, so the number of muscle fibers involved will also increase, so the muscle thickness will also increase. Moreover, muscle fibers will become shorter when force is applied, and the cross-sectional area will increase, which will also increase muscle thickness.
[0124] The increase in muscle thickness is a nonlinear process, which can be explained from the following three aspects:
[0125] (1) The sequence of motor unit recruitment is nonlinear;
[0126] (2) The relationship between the force generated by the muscle and the muscle length is nonlinear, which can be explained by the sliding filament theory;
[0127] (3) The relationship between the physiological cross-sectional area (PCSA) of muscle and muscle thickness is nonlinear. PCSA can be calculated by the following formula:
[0128]
[0129] Among them, G is muscle mass, θ is the muscle fiber angle of pennate muscle, ρ is muscle density, and L is muscle fiber length. Since volume = mass / density, the above formula can be expressed as:
[0130]
[0131] Among them, V is the muscle volume. Since it is generally believed that the angle of the biceps brachii is 0°, the PCSA of the biceps brachii is only related to the muscle volume and muscle fiber length. However, most muscles are considered to be fusiform, so the relationship between PCSA and muscle thickness (diameter) cannot be calculated using a linear formula, it is nonlinear. Of course, the current experimental results are not enough to fully describe the muscle state, and more volunteers of different types are needed to join, and further experiments are also in progress. Through the above research: For fitness people, our products will effectively solve the problem of increased muscle thickness during muscle training, and make changes and further guidance for fitness people. During fitness, fitness people often pay attention to muscle growth and physical fitness improvement. Through research, it was found that the increase in muscle thickness is caused by the order of muscle motor unit recruitment, and this increase process is nonlinear. This provides an important theoretical basis and guiding direction for fitness people, helping them to better understand the mechanism of muscle growth and how to optimize the muscle training process. In addition, our research also revealed the characteristics of muscle fibers becoming shorter and increasing cross-sectional area when exerting force, resulting in increased muscle thickness, which is very helpful for fitness people to understand muscle changes and responses more accurately during training. The discovery of these nonlinear characteristics provides guidance for fitness people to develop more scientific muscle training plans, so that they can more effectively increase muscle thickness and improve physical fitness.
[0132] In summary, our research solves the problem of increased muscle thickness during muscle training for fitness people, and by discovering the nonlinear characteristics in muscle training, it provides fitness people with more accurate training guidance and optimization plans, making their fitness effects more significant and promoting fitness people to achieve better results in muscle training.
[0133] The present application also provides an application scenario, which applies the above-mentioned muscle mass measurement method. Specifically: The muscle mass measurement method provided in this embodiment can be applied in the scenario of fitness people. The product in the present invention effectively solves the problem of increased muscle thickness during muscle training, and makes changes and further guidance to fitness people.
[0134] Based on the same inventive concept, the embodiment of the present application also provides a muscle mass measurement device, comprising:
[0135] The data indicator collection module is used to collect multiple data indicators; the data indicators include: biceps brachii ultrasound image, physiological information, muscle weight and muscle strength.
[0136] The texture feature extraction module is used to extract the texture features of the ultrasound image.
[0137] The correlation determination module is used to determine the correlation between the texture feature and the muscle strength and muscle weight.
[0138] The clustering module is used to cluster the features with significant correlation to obtain clustering results.
[0139] The label marking module is used to label the muscles based on the clustering results; the labels include: good performance and poor performance.
[0140] The training module is used to train the neural network model based on the texture features and labels.
[0141] The muscle mass prediction module is used to input the collected biceps brachii ultrasound image into the trained neural network model to obtain the measurement result of muscle mass.
[0142] The solution to the problem provided by the device is similar to the solution described in the above method, so the specific limitations in one or more muscle mass measurement device embodiments provided below can refer to the limitations on the muscle mass measurement method above and will not be repeated here.
[0143] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 2 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store video tag processing data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a muscle mass measurement method is implemented.
[0144] Those skilled in the art will understand that Figure 2 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0145] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.
[0146] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.
[0147] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.
[0148] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.
[0149] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).
[0150] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.
[0151] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0152] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.
Claims
1. A method for measuring muscle mass, characterized in that: The muscle mass measurement method comprises: Collecting multiple data indicators; the data indicators include: biceps brachii ultrasound image, physiological information, muscle weight and muscle strength; Extracting texture features of the ultrasound image; Determining the correlation between the texture feature and the muscle strength and muscle weight; Cluster the features with significant correlation to obtain clustering results; Labeling the muscles based on the clustering results; the labels include: good performance and poor performance; Based on the texture features and labels, a neural network model is trained; The collected biceps brachii ultrasound images are input into the trained neural network model to obtain the measurement results of muscle mass.
2. The muscle mass measurement method according to claim 1, characterized in that: The muscle mass measurement method further comprises: Save the trained neural network model as a pb model; Transplant the pb model to Android Studio and configure the environment; Design the graphical user interface, embed it into Android Studio, and configure the corresponding interface; The trained neural network model is called and encapsulated as a mobile phone applet.
3. The muscle mass measurement method according to claim 1, characterized in that: Extracting the texture features of the ultrasound image specifically includes the following steps: Performing data cleaning on the biceps brachii ultrasound image set; Firstly, a binarization operation is performed on the biceps brachii ultrasound image, and then an enhancement process is performed on the biceps brachii ultrasound image, and a contrast enhancement is performed using a histogram; The enhanced biceps brachii ultrasound image is filtered using median filtering, Gaussian filtering and mean filtering; The filtered biceps brachii ultrasound image is sharpened by Laplace sharpening method to extract the edge; Perform threshold segmentation on the biceps brachii ultrasound image after edge extraction to remove redundant background; The entropy value of the two-dimensional image after threshold segmentation is calculated using the Shannon entropy formula; The biceps brachii ultrasound image was segmented by morphological operation after removing the redundant background, and the muscle structure features were extracted. The canny detection operator is used to extract the contour of the muscle area after image segmentation. The pixel values within the contour are calculated first, and then the ultrasound image is calibrated to convert the pixel area into the actual physical area. Use local binary pattern LBP to extract the energy value lbp_energy and LBP entropy lbp_entropy in the LBP feature vector; The image gray level co-occurrence matrix GLCM is used to extract the contrast mean, correlation mean and energy mean.
4. The muscle mass measurement method according to claim 3, characterized in that: The expression of the Shannon entropy formula is as follows: Among them, H represents the two-dimensional image entropy value, L represents the total number of gray levels, and p(i) represents the probability of image gray level i.
5. The muscle mass measurement method according to claim 3, characterized in that: The expression of the energy value lbp_energy in the LBP feature vector is as follows: Among them, E represents the energy value in the LBP feature vector, N represents the total number of LBP histogram bins, q(t) represents the value of the tth bin in the LBP histogram, and t is the histogram bin value index value.
6. The muscle mass measurement method according to claim 3, characterized in that: The expression of the contrast mean contrast_mean is as follows: contrast_mean=∑ i ∑ j (i-j) 2 P(i,j) Among them, i and j both represent gray levels, and P(i,j) represents the elements of gray levels i and j in the GLCM. The expression of the correlation mean correlation_mean is as follows: Where i and j both represent gray levels, P(i,j) represents the element of gray level i and j in GLCM, μ i and μ j denote the mean of the GLCM rows and columns, respectively; The expression of the energy average value energy_mean is as follows: enery_mean=∑ i ∑ j P(i,j) 2 。 7. A muscle mass measuring device, characterized in that: The muscle mass measuring device comprises: A data index collection module, used to collect multiple data indexes; the data indexes include: biceps brachii ultrasound image, physiological information, muscle weight and muscle strength; A texture feature extraction module, used to extract texture features of the ultrasound image; A correlation determination module, used to determine the correlation between the texture feature and the muscle strength and muscle weight; The clustering module is used to cluster the features with significant correlation to obtain clustering results; A label marking module, used to label the muscles based on the clustering results; the labels include: good performance and poor performance; A training module, used for training a neural network model based on the texture features and labels; The muscle mass prediction module is used to input the collected biceps brachii ultrasound image into the trained neural network model to obtain the measurement result of muscle mass.
8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the muscle mass measurement method according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the muscle mass measurement method according to any one of claims 1 to 6 are implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the muscle mass measurement method according to any one of claims 1 to 6 are implemented.
