WO2024051016A1 - 一种用于模型小动物的肌肉超声图像分析方法 - Google Patents

一种用于模型小动物的肌肉超声图像分析方法 Download PDF

Info

Publication number
WO2024051016A1
WO2024051016A1 PCT/CN2022/138045 CN2022138045W WO2024051016A1 WO 2024051016 A1 WO2024051016 A1 WO 2024051016A1 CN 2022138045 W CN2022138045 W CN 2022138045W WO 2024051016 A1 WO2024051016 A1 WO 2024051016A1
Authority
WO
WIPO (PCT)
Prior art keywords
muscle
features
image
analysis
small
Prior art date
Application number
PCT/CN2022/138045
Other languages
English (en)
French (fr)
Inventor
周永进
邓妙琴
田静
Original Assignee
深圳大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳大学 filed Critical 深圳大学
Publication of WO2024051016A1 publication Critical patent/WO2024051016A1/zh

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/52—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/52—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves
    • A61B8/5215—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving processing of medical diagnostic data
    • A61B8/5223—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving processing of medical diagnostic data for extracting a diagnostic or physiological parameter from medical diagnostic data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/54—Extraction of image or video features relating to texture
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B2503/00—Evaluating a particular growth phase or type of persons or animals
    • A61B2503/40—Animals
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10132—Ultrasound image

Definitions

  • the invention relates to the field of medical image analysis, and in particular to a muscle ultrasound image analysis method for model small animals.
  • the technical problem to be solved by the present invention is to provide an image feature extraction and classification method based on muscle ultrasound in view of the above-mentioned defects of the existing technology, aiming to solve the problem in the existing technology that the precise detection of the muscle structure of small animals in living models is not possible. and quantitative assessment issues.
  • the present invention provides a muscle ultrasound image analysis method for small animal models, wherein the method includes:
  • the muscle ultrasound image of the small model animal to be analyzed is input into the analysis model of the muscle ultrasound image for analysis, and the detection and quantitative evaluation results of the muscle structure of the small model animal to be analyzed are obtained.
  • the acquisition of muscle ultrasound images of small model animals includes:
  • test model animal The hind limbs of the test model animal were depilated. Under the anesthesia state of the test model animal, the test model animal was placed in the center of the experimental board in a supine position, and its mouth and nose were placed in the anesthetized state. in the pipeline of the instrument;
  • the long axis of the ultrasonic probe is parallel to the Achilles tendon of the small animal under test, and the ultrasonic probe is placed at the first detection position on the hind limb of the small animal under test by setting a mark;
  • an ultrasound gel coupling agent to ensure the acoustic coupling between the ultrasound probe and the skin, adjust the ultrasound probe to optimize the contrast of the muscle bundles in the ultrasound image, based on the detection mode and the first detection position, using real-time B-mode ultrasonic imaging equipment to obtain muscle ultrasound images of the test model small animal.
  • the ROI division is performed on the muscle ultrasound image to obtain the divided muscle ultrasound image, including:
  • the muscle ultrasound image is grayscaled, cropped, and ROI is divided according to the manually outlined or automatically segmented muscle areas to obtain the divided muscle ultrasound image.
  • the extraction of muscle morphological features and image frequency features from the divided muscle ultrasound images includes:
  • the average frequency analysis feature is extracted from the divided muscle ultrasound image; wherein, the calculation formula of the average frequency analysis feature is: n, I, and f are the length, power, and frequency of the power density spectrum, respectively.
  • extracting image texture analysis features from the divided muscle ultrasound images includes:
  • first-order statistical features are extracted from the divided muscle ultrasound images; wherein the first-order statistical features include integrated optical density, mean, standard deviation, variance, skewness, peak degree and energy;
  • Haralick features are extracted from the divided muscle ultrasound images; wherein the Haralick features include contrast, correlation, energy, entropy, homogeneity and symmetry;
  • Galloway features are extracted from the divided muscle ultrasound images; wherein, the Galloway features include short run advantage, long run advantage, grayscale non-uniformity, long run non-uniformity, run length percentage;
  • local binary pattern features are extracted from the divided muscle ultrasound image; wherein the local binary pattern features include energy and entropy, so
  • the image texture analysis feature is obtained according to the first-order statistical feature, the Haralick feature, the Galloway feature and the local binary pattern feature.
  • the statistical analysis of the muscle morphological features, image frequency features and image texture features is performed to obtain statistically different features, including:
  • test result If the first test result obeys normality, perform a homogeneity of variance test on the first test result to obtain a second test result;
  • the first test result does not obey normality, then perform a non-parametric test on the muscle morphological features, image frequency features and image texture features to obtain the statistically different features;
  • the second test result satisfies homogeneity of variances, then conduct an independent sample T test on the muscle morphological features, image frequency features and image texture features to obtain the statistically different features;
  • the second test result satisfies the non-homogeneity of variances, then perform a modified variance T test on the muscle morphological features, image frequency features and image texture features to obtain the statistically different features.
  • the repeatability analysis based on test-retest reliability is performed to obtain reliable features, including:
  • the intra-class correlation coefficient ICC(1,1) is used to evaluate the reliability of the several test results; where, the correlation coefficient is Among them, BMS and WMS are the root mean squares between groups and within groups respectively obtained by Kruskal-Wallis one-way analysis of variance, and k is the number of repeated measurements;
  • the reliability characteristics are obtained based on the reliability.
  • embodiments of the present invention also provide a muscle ultrasound image analysis device for model small animals, wherein the device includes:
  • the ROI division module is used to obtain muscle ultrasound images of small model animals, and divide the muscle ultrasound images into ROIs to obtain divided muscle ultrasound images;
  • a feature extraction module used to extract muscle morphological features, image frequency features and image texture features from the divided muscle ultrasound images
  • the feature analysis module is used to conduct statistical analysis on the muscle morphological features, image frequency features and image texture features to obtain statistically different features, and perform repeatability analysis based on test-retest credibility to obtain reliable features;
  • An analysis model acquisition module is used to obtain an analysis model of muscle ultrasound images based on the statistically different features and the reliable features;
  • the analysis module is used to input the muscle ultrasound image of the model animal to be analyzed into the analysis model of the muscle ultrasound image for analysis, and obtain the detection and quantitative evaluation results of the muscle structure of the model animal to be analyzed.
  • embodiments of the present invention further provide an intelligent terminal, wherein the intelligent terminal includes a memory, a processor, and a muscle model for small animals stored in the memory and operable on the processor.
  • Ultrasound image analysis program when the processor executes the muscle ultrasound image analysis program for model small animals, the steps of the muscle ultrasound image analysis method for model small animals as described in any one of the above are implemented.
  • embodiments of the present invention further provide a computer-readable storage medium, wherein a muscle ultrasound image analysis program for small model animals is stored on the computer-readable storage medium, and the muscle ultrasound image analysis program for small model animals is stored therein.
  • the ultrasound image analysis program is executed by the processor, the steps of the muscle ultrasound image analysis method for small model animals as described in any one of the above are implemented.
  • the present invention provides a muscle ultrasound image analysis method for small model animals.
  • the present invention first utilizes currently widely used ultrasonic equipment to collect images of muscle parts of small model animals in vivo.
  • Ultrasound images have the advantages of being non-invasive, simple and easy to use, and can achieve precise detection and visualization of the muscle structure of small animals in living models.
  • image processing and other means are used to extract muscle morphological features, image frequency analysis features, image texture features, etc.
  • the analysis system finally inputs the muscle ultrasound image of the model animal to be analyzed, and can extract the corresponding features and obtain the analysis results, thereby achieving precise detection and quantitative evaluation of the muscle structure of the living model animal.
  • Figure 1 is a schematic flow chart of a muscle ultrasound image analysis method for small animal models provided by an embodiment of the present invention.
  • Figure 2 is an example diagram of ultrasound images of the hindlimb muscles of a small model animal provided by an embodiment of the present invention.
  • Figure 3 is a statistical analysis flow chart provided by an embodiment of the present invention.
  • Figure 4 is a functional block diagram of a muscle ultrasound image analysis device for model small animals provided by an embodiment of the present invention.
  • Figure 5 is a functional block diagram of the internal structure of an intelligent terminal provided by an embodiment of the present invention.
  • model small animals are considered to be crucial experimental methods and means in modern life science research, and in vivo model small animal imaging technology plays an irreplaceable role in medical research.
  • In vivo model small animal imaging technology refers to the non-destructive detection of the structure, function and physiological information of the animal without damaging it. It can realize long-term dynamic observation of small animals, and for the establishment of various human disease models Small animal imaging, through long-term continuous monitoring, can study the occurrence, development and various therapeutic effects of diseases; in addition, research on living small animals is conducted while maintaining the real in-body environment of the animals. During the research process Integrating all interacting physiological information can make experimental results more realistic and reliable. Testing the muscles of small model animals can provide valuable information for muscle function assessment and medical diagnosis of various diseases. It can also provide valuable information for monitoring disease progression and muscle response during treatment. It can also provide valuable information for various types of diseases. It is of great significance to evaluate the muscle function of the disease, diagnose the disease, and formulate the rehabilitation plan.
  • the first is histological testing based on physiological and biochemical indicators of muscle tissue; the second is testing of skeletal muscle contraction characteristics based on muscle contractility.
  • the seventh is an open field experiment based on the total path of autonomous crawling.
  • Histological testing refers to using the muscles of small model animals to measure tissue biochemical indicators, evaluate histopathological indicators and measure molecular biology indicators.
  • the measurement of tissue biochemical indicators mainly measures muscle protein content, muscle glycogen, muscle creatine kinase, etc.; the evaluation of histopathological indicators refers to the measurement of conventional histology (such as hematoxylin-eosin staining) and special histology (such as Masson staining).
  • myosin ATPase histochemical staining myosin ATPase histochemical staining, modified Gomori staining, etc.
  • ultrahistology transmission electron microscopy
  • the detection of skeletal muscle contraction characteristics refers to perfusing isolated muscles such as soleus and extensor digitorum longus at 30 degrees Celsius and giving a certain voltage stimulus to measure their single contraction force and tetanic contraction force.
  • the maximum contraction force at the optimal initial length of the muscle that is, the single contraction force
  • the muscle is given a tetanic stimulation at the optimal voltage, and the muscle's response to each stimulus is recorded in real time.
  • the tetanic contraction force at the time point was determined, and the corresponding contraction force curve was obtained.
  • Claw gripping strength measurement of limbs refers to measuring the gripping strength of small animals through a claw gripping tester. It can evaluate the strength of the front and rear limbs of small animals, thereby evaluating the degree of nerve damage and limb muscle damage of small animals. Place the small animal in the metal grid measurement area of the claw grip tester. The small animal will automatically grasp the metal grid. After it has a firm grip, pull its tail back and the small animal will instinctively grasp the metal grid. At this time, the tester will generate a reading. After pulling it away from the metal grid, record the maximum reading of the tester during this period, which is the strength of the small animal's front and rear limbs. Usually each small animal is measured three times and the average value is taken. Although this detection method can measure the grasping strength of the limbs of model animals, it cannot achieve precise detection and quantitative evaluation of the muscle structure of living model animals.
  • the pole climbing test refers to testing the crawling time and status of the small animal to evaluate the crawling speed, limb coordination and limb support strength of the small animal.
  • the rotarod test refers to evaluating the balance ability, coordination ability and muscle strength of small animals by testing the time that a model animal runs on a rotating rotarod. Before the experiment, the small animals were subjected to rotarod adaptation and learning training for five consecutive days. At the beginning of the rotarod experiment, adjust the rotating speed of the rotarod instrument to a suitable speed. The specific speed can be selected according to the size of each small model animal. Record the time it spends on the rotarod instrument, and use a certain time as the dividing line.
  • this detection method can evaluate the balance ability, coordination ability and muscle strength of small model animals, it cannot achieve precise detection and quantitative evaluation of the muscle structure of small living model animals.
  • the suspension experiment refers to measuring the limb endurance of each small model animal through a suspension net. Place the small animal upside down on the hanging grid. After it has a firm grip, start timing. When it falls from the hanging grid due to exhaustion of strength, it stops timing. A certain time is used as the dividing value. If it exceeds this time, the timing will be as follows. The time is recorded. If the time is less than this time, the actual time is recorded. Repeat three times, with a certain interval between each time to allow the small animal to recover its physical strength. Although this detection method can measure the limb endurance of small model animals, it cannot achieve precise detection and quantitative evaluation of the muscle structure of small living model animals.
  • the open field experiment refers to evaluating the movement autonomy and ability of small animals by counting the total path of independent crawling of small animals in an open box for a certain period of time. Place the small animal in a light and dark box, use a high-definition camera to track and record its movement trajectory within a certain period of time in the box, use open field experiment video analysis software to analyze its activity trajectory map, and make statistics on the total crawling path.
  • this detection method can evaluate the movement autonomy and ability of small model animals, it cannot achieve precise detection and quantitative evaluation of the muscle structure of small living model animals.
  • the present invention provides a muscle ultrasound image analysis method for small model animals.
  • the present invention first utilizes currently widely used ultrasound equipment to collect ultrasound images of muscle parts of small model animals in vivo. It is non-invasive, simple and easy to perform, and can achieve precise detection and visualization of the muscle structure of small animals in living models. Then, image processing and other means are used to extract muscle morphological features, image frequency analysis features, image texture features, etc. from the obtained muscle ultrasound images of the model animals, so as to achieve quantitative evaluation of the muscle structure of the model animals, and use statistics and repeatability analysis methods to analyze the extracted image features, thereby screening out important features that are highly related to the disease and stable and reliable features with high test-retest reliability, and construct muscle ultrasound images for small model animals.
  • the analysis system finally inputs the muscle ultrasound image of the model animal to be analyzed, and can extract the corresponding features and obtain the analysis results, thereby achieving precise detection and quantitative evaluation of the muscle structure of the living model animal.
  • This embodiment provides a muscle ultrasound image analysis method for small model animals. As shown in Figure 1, the method includes the following steps:
  • Step S100 Obtain the muscle ultrasound image of the small model animal, and divide the muscle ultrasound image into ROIs to obtain the divided muscle ultrasound image;
  • ultrasound imaging equipment has the potential to detect the fine structure of muscles and allows the visualization and quantification of muscle structures, with the advantages of non-invasiveness, no radiation, and real-time dynamic monitoring.
  • the muscle ultrasound image needs to be divided into ROIs to obtain the divided muscle ultrasound image.
  • ROI region of interest
  • region of interest ROI is the region of interest.
  • machine vision and image processing the area to be processed is outlined from the processed image in the form of boxes, circles, ellipses, irregular polygons, etc., which is called the region of interest ROI.
  • Various operators and functions are commonly used in machine vision software such as Halcon, OpenCV, and Matlab to obtain the ROI of the area of interest and perform the next step of image processing.
  • step S100 in this embodiment includes the following steps:
  • Step S101 Depilate the hind limbs of the small animal under test. Under the anesthesia state of the small animal under test, place the small animal under test in a supine position in the center of the experimental board and place its mouth and nose. In the pipeline connected to the anesthesia machine;
  • Step S102 Set the detection mode of the ultrasound imaging system to the musculoskeletal detection mode
  • Step S103 Place the long axis of the ultrasonic probe parallel to the Achilles tendon of the small animal under test, and set a mark to keep the ultrasonic probe at the first detection position on the hind limb of the small animal under test;
  • Step S104 Under the anesthesia state of the test model animal, apply ultrasound gel coupling agent to ensure the acoustic coupling between the ultrasound probe and the skin, adjust the ultrasound probe to optimize the contrast of the muscle bundles in the ultrasound image, based on the In the detection mode and the first detection position, use real-time B-mode ultrasonic imaging equipment to obtain muscle ultrasound images of the small model animal being tested;
  • Step S105 Grayscale and crop the muscle ultrasound image, and divide the ROI according to the manually outlined or automatically segmented muscle areas to obtain the divided muscle ultrasound image.
  • the specific steps for collecting ultrasound images of the hindlimb muscles of the small model animal are as follows: first, before collecting the ultrasound images, the hindlimbs of the small model animal are depilated to avoid the impact of their fur on the collection and quality of the muscle ultrasound images; Secondly, while the model animal is anesthetized, place it supine in the center of the experimental board, and place its mouth and nose in the pipe connected to the anesthesia machine so that it remains under anesthesia during the experiment to facilitate the collection of ultrasound images. ; Then use real-time ultrasound imaging equipment to obtain muscle ultrasound images of its hind limbs.
  • the ultrasound imaging system selects the musculoskeletal detection mode, places the long axis of the ultrasound probe parallel to the Achilles tendon, and places it on the surface of the hind limbs or other specific locations; apply an appropriate amount Ultrasound gel coupling agent to ensure acoustic coupling between the probe and the skin; the ultrasound probe can be adjusted to optimize the contrast of the muscle bundles in the ultrasound image, and the position can be marked to ensure the probe is placed in the same position every time.
  • An example of ultrasound image of the hindlimb muscles of a small model animal is shown in Figure 2. And before extracting features, the image is preprocessed, including grayscale, cropping, and dividing ROI based on manually outlined or automatically segmented muscle areas.
  • muscle ultrasound image collection in addition to collecting static muscle ultrasound images of small model animals, it is also possible to collect muscle ultrasound images of small model animals during the dynamic structural changes caused by muscle stretching; in addition to B-type acquisition equipment
  • shear wave elastography equipment or other ultrasound imaging methods can also be used for acquisition.
  • Step S200 Extract muscle morphological features, image frequency features and image texture features from the divided muscle ultrasound images
  • step S200 in this embodiment includes the following steps:
  • Step S201 Perform normalized Redon transformation on the divided muscle ultrasound image to obtain a Redon transformation matrix
  • Step S202 Calculate the gradient of the Redon transformation matrix and perform edge enhancement to obtain the Redon transformation gradient matrix
  • Step S203 Obtain muscle thickness, muscle fiber length and pennation angle characteristics according to the Leyden transform gradient matrix
  • Step S204 Obtain the muscle morphological characteristics according to the muscle thickness, muscle fiber length and pennation angle characteristics
  • Step S205 Extract the average frequency analysis feature from the divided muscle ultrasound image; wherein the calculation formula of the average frequency analysis feature is: n, I, and f are the length, power, and frequency of the power density spectrum respectively;
  • Step S206 Extract first-order statistical features from the divided muscle ultrasound image based on pixel grayscale distribution calculation; wherein the first-order statistical features include integrated optical density, average value, standard deviation, variance, and bias. degree, kurtosis and energy;
  • Step S207 Extract Haralick features from the divided muscle ultrasound image based on gray level co-occurrence matrix calculation; wherein the Haralick features include contrast, correlation, energy, entropy, homogeneity and symmetry;
  • Step S208 Extract Galloway features from the divided muscle ultrasound image based on grayscale run length matrix calculation; wherein the Galloway features include short run advantages, long run advantages, grayscale unevenness, and long run unevenness. sex, run percentage;
  • Step S210 Obtain the image texture analysis feature according to the first-order statistical feature, the Haralick feature, the Galloway feature and the local binary pattern feature.
  • the morphological characteristics of muscles include muscle thickness, muscle fiber length, pennation angle, etc. Muscle fiber length and pennation angle are calculated based on the muscle fiber.
  • the morphological characteristics of the hindlimb muscles are mainly the area of the muscle area that is manually outlined or automatically segmented by an algorithm. If the muscle fibers in the collected muscle area are relatively clear, some feature detection methods can also be used to estimate the morphological parameters of the muscle, such as feature detection methods based on the Leiden transform gradient matrix.
  • MFAF mean frequency analysis feature
  • n, I, and f are the length, power and frequency of the power density spectrum respectively.
  • Image texture analysis features mainly include first-order statistical features and high-order texture features.
  • first-order statistical features can effectively and quantitatively describe the ultrasonic echo intensity of skeletal muscles; in addition, other studies have shown that there are differences in the ultrasonic echo intensity information of skeletal muscles of different ages or groups, and, These features can also provide some structural information related to muscle status, thereby providing effective information for muscle damage assessment.
  • high-order texture features such as Haralick features, Galloway features and Local Binary Pattern (LBP) features, can perform better in fine tasks such as muscle gender recognition than first-order statistical features. better.
  • LBP Local Binary Pattern
  • first-order statistical features are features calculated directly based on the pixel grayscale distribution of the original image, including features such as integrated optical density, mean, standard deviation, variance, skewness, kurtosis, and energy.
  • the Haralick feature is calculated from the gray level co-occurrence matrix.
  • the gray level co-occurrence matrix is a matrix function of the pixel distance and direction. Calculating the correlation between the gray level values of two points at a given spatial distance d and direction ⁇ is a method through A common method to describe image texture by studying the spatial correlation characteristics of grayscale.
  • Typical Haralick features include contrast, correlation, energy, entropy, homogeneity and symmetry, etc.
  • each Haralick feature contains four directions: 0°, 45°, 90° and 135°.
  • the Galloway feature is calculated based on the grayscale run length matrix.
  • the grayscale run length matrix represents the regularity of texture changes of an image. Its size is determined by the gray level and image size of the image, including short run length advantages and long run length advantages. , grayscale non-uniformity, long run non-uniformity, run percentage and other texture statistical features, and each Galloway feature also contains four directions of 0°, 45°, 90° and 135°.
  • LBP features are obtained by comparing the central pixel of a local area of the image with its neighborhood. It mainly describes the local texture features of the image. It has significant advantages such as rotation invariance and grayscale invariance, including energy and entropy. , the specific calculation formula is shown below:
  • muscle morphological features such as pennation angle, muscle thickness, muscle fiber angle, muscle bundle length, muscle physiological cross-sectional area, etc.
  • features based on the Redden transform gradient matrix In addition to detection methods, methods based on Hough transform, deep learning, etc. can also be used to locate various structural elements of muscle tissue, thereby achieving automatic measurement of morphological parameters.
  • Step S300 Perform statistical analysis on the muscle morphological features, image frequency features and image texture features to obtain statistically different features, and perform repeatability analysis based on test-retest credibility to obtain reliable features;
  • Statistical analysis is to collect a large amount of data/numbers, use statistical methods to find regular features from a large amount of data, and realize the visualization and visualization of a large amount of data in order to understand the statistical regularities contained in the data, and then use the regular features of the data to explain problems.
  • Repeatability analysis is a test-retest reliability level that evaluates image features through multiple repeated measurements to screen out stable and reliable features.
  • this embodiment uses statistics and repeatability analysis methods to analyze the extracted image features, thereby screening out important features that are highly related to the disease as well as stable and reliable features with high test-retest reliability.
  • step S300 in this embodiment includes the following steps:
  • Step S301 Perform a normality test on the muscle morphological features, image frequency features and image texture features to obtain the first test result;
  • Step S302 If the first test result obeys normality, perform a homogeneity of variance test on the first test result to obtain a second test result;
  • Step S303 If the first test result does not obey normality, perform a non-parametric test on the muscle morphological features, image frequency features and image texture features to obtain the statistically different features;
  • Step S304 If the second test result satisfies homogeneity of variances, perform an independent sample T test on the muscle morphological features, image frequency features and image texture features to obtain the statistically different features;
  • Step S305 If the second test result satisfies variance non-homogeneity, perform a modified variance T test on the muscle morphological features, image frequency features and image texture features to obtain the statistically different features;
  • Step S306 Perform repetitive tests on the muscle morphological features, image frequency features and image texture features to obtain several test results;
  • Step S307 Use the intra-class correlation coefficient ICC(1,1) to evaluate the reliability of the several test results; wherein, the correlation coefficient is Among them, BMS and WMS are the root mean squares between groups and within groups respectively obtained by Kruskal-Wallis one-way analysis of variance, and k is the number of repeated measurements;
  • Step S308 Obtain the reliable feature according to the reliability.
  • BMS and WMS respectively represent the root mean square between groups and within groups obtained by Kruskal-Wallis one-way analysis of variance, and k is the number of repeated measurements.
  • Step S400 Obtain an analysis model of the muscle ultrasound image based on the statistically different features and the reliable features;
  • Step S500 Input the muscle ultrasound image of the small model animal to be analyzed into the analysis model of the muscle ultrasound image for analysis, and obtain detection and quantitative evaluation results of the muscle structure of the small model animal to be analyzed.
  • a muscle ultrasound image analysis system for small model animals is constructed.
  • the corresponding features can be extracted and the analysis results obtained, thereby achieving precise analysis of the muscle structure of the small living model animal. Detection and quantitative assessment.
  • This embodiment also provides a muscle ultrasound image analysis device for model small animals, the device includes:
  • the ROI dividing module 10 is used to obtain muscle ultrasound images of small model animals, and divide the muscle ultrasound images into ROIs to obtain divided muscle ultrasound images;
  • the feature extraction module 20 is used to extract muscle morphological features, image frequency features and image texture features from the divided muscle ultrasound images;
  • the feature analysis module 30 is used to perform statistical analysis on the muscle morphological features, image frequency features and image texture features to obtain statistically different features, and perform repeatability analysis based on test-retest credibility to obtain reliable features;
  • the analysis model acquisition module 40 is used to obtain an analysis model of muscle ultrasound images based on the statistically different features and the reliable features;
  • the analysis module 50 is used to input the muscle ultrasound image of the small model animal to be analyzed into the analysis model of the muscle ultrasound image for analysis, and obtain the detection and quantitative evaluation results of the muscle structure of the small model animal to be analyzed.
  • the ROI dividing module 10 includes:
  • the pretreatment unit is used to depilate the hind limbs of the small model animal. Under the anesthesia state of the small animal, the small animal is placed in the center of the experimental board in a supine position. The mouth and nose are placed in the tube connected to the anesthesia machine;
  • a detection mode setting unit used to set the detection mode of the ultrasonic imaging system to the musculoskeletal detection mode
  • An ultrasonic probe placement unit is used to position the long axis of the ultrasonic probe parallel to the Achilles tendon of the small animal being tested, and to maintain the ultrasonic probe at the first detection position on the hind limb of the small animal being tested by setting a mark;
  • a muscle ultrasound image acquisition unit is used to apply ultrasound gel coupling agent to ensure the acoustic coupling between the ultrasound probe and the skin under the anesthesia state of the test model small animal, and adjust the ultrasound probe to optimize the muscle bundles in the ultrasound image. Contrast, based on the detection mode and the first detection position, use real-time B-mode ultrasonic imaging equipment to obtain the muscle ultrasound image of the test model small animal;
  • the ROI dividing unit is used to grayscale and crop the muscle ultrasound image, divide the ROI according to the manually outlined or automatically segmented muscle areas, and obtain the divided muscle ultrasound image.
  • the feature extraction module 20 includes:
  • a Redon transformation matrix acquisition unit configured to perform normalized Redon transformation on the divided muscle ultrasound image to obtain a Redon transformation matrix
  • a Redden transform gradient matrix acquisition unit used to obtain the gradient of the Redden transform matrix and perform edge enhancement to obtain the Redden transform gradient matrix
  • a first feature acquisition unit configured to obtain muscle thickness, muscle fiber length and pennation angle features based on the Leyden transform gradient matrix
  • a muscle morphological feature acquisition unit is used to obtain the muscle morphological features based on the muscle thickness, muscle fiber length and pennation angle features;
  • An average frequency analysis feature acquisition unit is used to extract the average frequency analysis feature from the divided muscle ultrasound image; wherein the calculation formula of the average frequency analysis feature is: n, I, and f are the length, power, and frequency of the power density spectrum respectively;
  • a first-order statistical feature extraction unit is used to extract first-order statistical features from the divided muscle ultrasound image based on pixel grayscale distribution calculation; wherein the first-order statistical features include integrated optical density, average value , standard deviation, variance, skewness, kurtosis and energy;
  • a Haralick feature extraction unit is used to extract Haralick features from the divided muscle ultrasound images based on gray level co-occurrence matrix calculation; wherein the Haralick features include contrast, correlation, energy, entropy, homogeneity and symmetry. ;
  • the Galloway feature extraction unit is used to extract Galloway features from the divided muscle ultrasound images based on grayscale run length matrix calculation; wherein the Galloway features include short run advantages, long run advantages, grayscale non-uniformity, Long run non-uniformity, run percentage;
  • a local binary pattern feature extraction unit configured to extract local binary pattern features from the divided muscle ultrasound image based on a comparison result between the central pixel of the local area of the image and the central pixel neighborhood;
  • the local Binary mode features include energy and entropy
  • fi represents the i-th block in the image Corresponding frequencies in local areas
  • An image texture analysis feature extraction unit is used to obtain the image texture analysis features based on the first-order statistical features, the Haralick features, the Galloway features and the local binary pattern features.
  • the feature analysis module 30 includes:
  • a first test result acquisition unit is used to perform a normality test on the muscle morphological features, image frequency features and image texture features to obtain the first test result;
  • a second test result acquisition unit is configured to perform a homogeneity of variance test on the first test result to obtain a second test result if the first test result obeys normality;
  • the first statistically different feature acquisition unit is used to perform a non-parametric test on the muscle morphological features, image frequency features and image texture features if the first test result does not obey normality to obtain the statistically significant feature. learning difference characteristics;
  • the second statistically different feature acquisition unit is used to perform an independent sample T test on the muscle morphological features, image frequency features and image texture features if the second test result satisfies the homogeneity of variances to obtain the statistically significant features.
  • the third one has a statistical difference feature acquisition unit, which is used to perform a modified variance T test on the muscle morphological features, image frequency features and image texture features if the second test result satisfies variance non-homogeneity to obtain the above There are statistically different characteristics;
  • a repeatability testing unit used to conduct repeatability testing on the muscle morphological characteristics, image frequency characteristics and image texture characteristics, and obtain several test results;
  • the reliability acquisition unit is used to evaluate the reliability of the several test results using the intra-class correlation coefficient ICC(1,1); wherein the correlation coefficient is Among them, BMS and WMS are the root mean squares between groups and within groups respectively obtained by Kruskal-Wallis one-way analysis of variance, and k is the number of repeated measurements;
  • a reliable feature acquisition unit is used to obtain the reliable features according to the reliability.
  • the present invention also provides an intelligent terminal, the functional block diagram of which can be shown in Figure 5 .
  • the intelligent terminal includes a processor, memory, network interface, display screen, and temperature sensor connected through a system bus.
  • the processor of the smart terminal is used to provide computing and control capabilities.
  • the memory of the smart terminal includes non-volatile storage media and internal memory.
  • the non-volatile storage medium stores operating systems and computer programs. This internal memory provides an environment for the execution of operating systems and computer programs in non-volatile storage media.
  • the network interface of the smart terminal is used to communicate with external terminals through network connections. When the computer program is executed by the processor, it implements an image feature extraction and classification method based on muscle ultrasound.
  • the display screen of the smart terminal may be a liquid crystal display or an electronic ink display.
  • the temperature sensor of the smart terminal is pre-set inside the smart terminal for detecting the operating temperature of the internal device.
  • Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • Volatile memory may include random access memory (RAM) or external cache memory.
  • RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Synchlink DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
  • SRAM static RAM
  • DRAM dynamic RAM
  • SDRAM synchronous DRAM
  • DDRSDRAM double data rate SDRAM
  • ESDRAM enhanced SDRAM
  • SLDRAM synchronous chain Synchlink DRAM
  • Rambus direct RAM
  • DRAM direct memory bus dynamic RAM
  • RDRAM memory bus dynamic RAM
  • the present invention discloses a muscle ultrasound image analysis method for small model animals.
  • the method includes: dividing the muscle ultrasound image of the model animal into ROIs, and extracting muscle morphological features from the divided muscle ultrasound images. , image frequency features and image texture features; perform statistical analysis on the features to obtain statistically different features, perform repeatability analysis based on test-retest credibility to obtain reliable features; based on statistically different features and reliable features , obtain an analysis model of the muscle ultrasound image; input the muscle ultrasound image of the model animal to be analyzed into the analysis model of the muscle ultrasound image for analysis, and obtain the detection and quantitative evaluation results of the muscle structure of the model animal to be analyzed.
  • the present invention can collect ultrasonic images of muscle parts of small model animals in vivo, and has the advantages of being non-invasive, simple and easy to implement, and enabling precise detection and visualization of the muscle structure of small animal models in vivo.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Theoretical Computer Science (AREA)
  • Medical Informatics (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • General Physics & Mathematics (AREA)
  • Biophysics (AREA)
  • Multimedia (AREA)
  • Pathology (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Quality & Reliability (AREA)
  • Physiology (AREA)
  • Ultra Sonic Daignosis Equipment (AREA)

Abstract

一种用于模型小动物的肌肉超声图像分析方法，包括：对模型小动物的肌肉超声图像进行ROI划分，得到划分后的肌肉超声图像（S100）；从划分后的肌肉超声图像中提取肌肉形态学特征、图像频率特征以及图像纹理特征（S200）；对肌肉形态学特征、图像频率特征以及图像纹理特征进行统计学分析得到有统计学差异特征，进行基于测试-重测可信度的重复性分析得到可靠特征（S300）；根据有统计学差异特征和可靠特征，得到肌肉超声图像的分析模型（S400）；将待分析的模型小动物的肌肉超声图像输入到肌肉超声图像的分析模型进行分析，得到待分析的模型小动物肌肉结构的检测和量化评估指标（S500）。通过采集活体下模型小动物的肌肉部位的超声图像，具有无创、简便易行且可实现对活体模型小动物肌肉结构的精细检测与可视化的优势。

Description

一种用于模型小动物的肌肉超声图像分析方法 技术领域
本发明涉及医学影像分析领域，具体涉及一种用于模型小动物的肌肉超声图像分析方法。
背景技术
研究发现各种模型小动物的基因序列与人类的基因具有高度同源性，其机能反应与人类相似，因此通过小动物模型能够更加有效地探究人类疾病的发生和发展规律；其次，通过利用模型小动物能够进行大量的探究实验，获取更多的实验数据，有助于使评价结果更为准确、科学、可靠。其中，对模型小动物进行肌肉方面的检测，可以为各类疾病的肌肉功能评估和医学分析提供有价值的信息。
目前常用的针对模型小动物的肌肉方面的检测方法有组织学检测、骨骼肌收缩特性检测、四肢爪抓力测定、爬杆测试、转棒实验、悬挂实验、旷场实验等，以上检测方式或有创，或仅能对模型小动物的肌肉力量和功能进行检测，并不能实现对活体模型小动物肌肉结构的精细检测和量化评估。
因此，现有技术还有待于改进和发展。
发明内容
本发明要解决的技术问题在于针对现有技术的上述缺陷，提供一种基于肌肉超声的影像特征提取及分类方法，旨在解决现有技术中的不能实现对活体模型小动物肌肉结构的精细检测和量化评估的问题。
本发明解决技术问题所采用的技术方案如下：
第一方面，本发明提供一种用于模型小动物的肌肉超声图像分析方法，其中，所述方法包括：
获取模型小动物的肌肉超声图像，并对所述肌肉超声图像进行ROI划分，得到划分后的肌肉超声图像；
从所述划分后的肌肉超声图像中提取肌肉形态学特征、图像频率特征以及图像纹理特征；
对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行统计学分析得到有统计学差异特征，进行基于测试-重测可信度的重复性分析得到可靠特征；
根据所述有统计学差异特征和所述可靠特征，得到肌肉超声图像的分析模型；
将待分析的模型小动物的肌肉超声图像输入到所述肌肉超声图像的分析模型进行分析，得到所述待分析的模型小动物肌肉结构的检测和量化评估结果。
在一种实现方式中，所述获取模型小动物的肌肉超声图像，包括：
对受试模型小动物的后肢进行脱毛处理，在所述受试模型小动物的麻醉状态下，将所述受试模型小动物仰卧式置于实验板中央，并将其口鼻置于连接麻醉仪的管道中；
设置超声成像系统的检测模式为肌骨检测模式；
将超声探头的长轴与受试模型小动物的跟腱平行，通过设置标记保持所述超声探头放置在所述受试模型小动物后肢上的第一检测位置；
在所述受试模型小动物的麻醉状态下，涂以超声凝胶耦合剂以确保超声探头与皮肤之间的声学耦合，调整超声探头以优化超声图像中肌束的对比度，基于所述检测模式和所述第一检测位置，使用实时B型超声成像设备获取所述受试模型小动物的肌肉超声图像。
在一种实现方式中，所述对所述肌肉超声图像进行ROI划分，得到划分后的肌肉超声图像，包括：
对所述肌肉超声图像进行灰度化、裁剪、根据人工勾画或自动分割的肌肉区域划分ROI，得到所述划分后的肌肉超声图像。
在一种实现方式中，所述从所述划分后的肌肉超声图像中提取肌肉形态学特征、图像频率特征，包括：
对所述划分后的肌肉超声图像进行归一化雷登变换，得到雷登变换矩阵；
对所述雷登变换矩阵求梯度并进行边缘增强，得到雷登变换梯度矩阵；
根据所述雷登变换梯度矩阵，得到肌肉厚度、肌纤维长度和羽状角特征；
根据所述肌肉厚度、肌纤维长度和羽状角特征，得到所述肌肉形态学特征；
从所述划分后的肌肉超声图像中提取所述平均频率分析特征；其中，所述平均频率分析特征的计算公式为
Figure PCTCN2022138045-appb-000001
n,I,和f分别为功率密度谱的长度、功率和频率。
在一种实现方式中，所述从所述划分后的肌肉超声图像中提取图像纹理分析特征，包括：
基于像素灰度分布计算，从所述划分后的肌肉超声图像中提取一阶统计学特征；其中，所述一阶统计学特征包括积分光密度、平均值、标准差、方差、偏度、峰度和能量；
基于灰度共生矩阵计算，从所述划分后的肌肉超声图像中提取Haralick特征；其中，所述Haralick特征包括对比度、相关性、能量、熵、同质性和对称性；
基于灰度游程长度矩阵计算，从所述划分后的肌肉超声图像中提取Galloway特征；其中，所述Galloway特征包括短游程优势、长游程优势、灰度不均匀性、长游程不均匀性、游程百分比；
基于图像局部区域的中心像素与所述中心像素邻域的比较结果，从所述划分后的肌肉超声图像中提取局部二值模式特征；其中，所述局部二值模式特征包括能量和熵，所述能量为LBP energy＝∑ if i 2，所述熵为LBP entropy＝-∑ if i 2log 2(f i)，f i表示第i块在图像局部区域的相应频率；
根据所述一阶统计学特征、所述Haralick特征、所述Galloway特征和所述局部二值模式特征得到所述图像纹理分析特征。
在一种实现方式中，所述对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行统计学分析得到有统计学差异特征，包括：
对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行正态性检验，得到第一检验结果；
若所述第一检验结果服从正态，则对所述第一检验结果进行方差齐性检验，得到第二检验结果；
若所述第一检验结果不服从正态，则对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行非参数检验，得到所述有统计学差异特征；
若所述第二检验结果满足方差齐性，则对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行独立样本T检验，得到所述有统计学差异特征；
若所述第二检验结果满足方差非齐性，则对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行修正方差T检验，得到所述有统计学差异特征。
在一种实现方式中，所述进行基于测试-重测可信度的重复性分析得到可靠特征，包括：
对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行重复性测试，得到若干测试结果；
对所述若干测试结果采用类内相关系数ICC(1,1)评估可靠度；其中，所述相关系数为
Figure PCTCN2022138045-appb-000002
其中，BMS和WMS分别为由Kruskal-Wallis单因素方差分析得到的组间和组内的均方根，k为重复测量次数；
根据所述可靠度得到所述可靠特征。
第二方面，本发明实施例还提供一种用于模型小动物的肌肉超声图像分析装置，其中，所述装置包括：
ROI划分模块，用于获取模型小动物的肌肉超声图像，并对所述肌肉超声图像进行ROI划分，得到划分后的肌肉超声图像；
特征提取模块，用于从所述划分后的肌肉超声图像中提取肌肉形态学特征、图像频率特征以及图像纹理特征；
特征分析模块，用于对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行统计学分析得到有统计学差异特征，进行基于测试-重测可信度的重复性分析得到可靠特征；
分析模型获取模块，用于根据所述有统计学差异特征和所述可靠特征，得到肌肉超声图像的分析模型；
分析模块，用于将待分析的模型小动物的肌肉超声图像输入到所述肌肉超声图像的分析模型进行分析，得到所述待分析的模型小动物肌肉结构的检测和量化评估结果。
第三方面，本发明实施例还提供一种智能终端，其中，所述智能终端包括存储器、处理器及存储在所述存储器中并可在所述处理器上运行的用于模型小动物的肌肉超声图像分析程序，所述处理器执行所述用于模型小动物的肌肉超声图像分析程序时，实现如以上任一项所述的用于模型小动物的肌肉超声图像分析方法的步骤。
第四方面，本发明实施例还提供一种计算机可读存储介质，其中，所述计算机可读存储介质上存储有用于模型小动物的肌肉超声图像分析程序，所述用于模型小动物的肌肉超声图像分析程序被处理器执行时，实现如以上任一项所述的用于模型小动物的肌肉超声图像分析方法的步骤。
有益效果：与现有技术相比，本发明提供了一种用于模型小动物的肌肉超声图像分析方法，本发明首先利用目前广泛使用的超声设备，可采集活体下模型小动物的肌肉部位的超声图像，具有无创、简便易行且可实现对活体模型小动物肌肉结构的精细检测与可视化等优势。然后，利用图像处理等手段对获取的模型小动物的肌肉超声图像提取肌肉形态学特征、图像频率分析特征、图像纹理特征等，从而可实现对模型小动物肌肉结构的量化评估，并且运用统计学和重复性分析方法对提取的图像特征进行分析，从而筛选出与疾病高度相关的重要特征以及具有较高测试-重测可靠度的稳定、可靠的特征，构建出针对模型小动物的肌肉超声图像分析系统，最后，输入待分析模型小动物的肌肉超声图像，即可提取出相应特征，并得到分析结果，从而实现对活体模型小动物肌肉结构的精细检测和量化评估。
附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明中记载的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1是本发明实施例提供的用于模型小动物的肌肉超声图像分析方法的流程示意图。
图2是本发明实施例提供的模型小动物后肢肌肉超声图像示例图。
图3是本发明实施例提供的统计学分析流程图。
图4本发明实施例提供的用于模型小动物的肌肉超声图像分析装置的原理框图。
图5是本发明实施例提供的智能终端的内部结构原理框图。
具体实施方式
为使本发明的目的、技术方案及效果更加清楚、明确，以下参照附图并举实施例对本发明进一步详细说明。应当理解，此处所描述的具体实施例仅用以解释本发明，并不用于限定本发明。
研究发现各种模型小动物的基因序列与人类的基因具有高度同源性，其机能反应与人类相似，因此通过小动物模型能够更加有效地探究人类疾病的发生和发展规律；其次，通过利用模型小动物能够进行大量的探究实验，获取更多的实验数据，有助于使评价结果更为准确、科学、可靠，且有利于发现生物体早期病变，提高疾病的早期诊断能力。模型小动物的应用对于人类疾病的发病机理研究、疾病发展过程分析、新药物和新治疗方法研究、治疗效果评价、治疗仪器的研发、疾病的防治等都具有重大的促进作用。因此模型小动物被认为是现代生命科学研究中至关重要的试验方法和手段，而在体模型小动物成像技术在医学研究中起到了不可代替的作用。在体模型小动物成像技术是指在不损伤动物的前提下实现对其结构、功能与生理信息的无损检测，可实现对小动物长时间的动态观测，而对于建立有人类各类疾病模型的小动物成像，通过长时间的连续监测，可对疾病的发生、发展以及各种治疗效果进行研究；此外，对活体小动物的研究是在保持动物真实在体环境下的研究，在研究过程中综合了所有相互作用的生理信息，可使得实验结果更为真实可靠。对模型小动物进行肌肉方面的检测，可以为各类疾病的肌肉功能评估和医学诊断提供有价值的信息，而且还可为监测疾病进展和治疗过程中的肌肉反应提供宝贵的信息，对各类疾病的肌肉功能评估、疾病诊断、康复计划制定等均具有重要意义。
目前，现有的针对模型小动物的肌肉方面的检测方法主要有七种：第一种是基于肌肉组织生理生化指标的组织学检测；第二种是基于肌肉收缩力的骨骼肌收缩特性检测，第三种是基于抓握力量的四肢爪抓力测定；第四种是基于爬行时间和状态的爬杆测试；第五种是基于转棒跑步时间的转棒实验；第六种是基于四肢耐力的悬挂实验；第七种是基于自主爬行总路径的旷场实验。
组织学检测是指利用对模型小动物的肌肉进行取材，对其进行组织生化指标测定、组织病理学指标评定和分子生物学指标测定。组织生化指标测定主要是对肌肉蛋白含量、肌糖原、肌肉肌酸激酶等进行测定；组织病理学指标评定是指通过常规组织学(如苏木 素-伊红染色)、特殊组织学(如Masson染色、肌球蛋白ATP酶组织化学染色、改良Gomori染色等)以及超微组织学(透射电镜)对肌肉的组织学特征进行观测评估；分子生物学指标测定是指通过内对照半定量反转录聚合酶链式反应方法对骨骼肌机械生长因子、肌动蛋白和肌球蛋白重链进行检测分析。虽然组织学检测可以对肌肉的各生理生化指标进行细致检测，且可以对肌肉的形态结构进行量化分析与评估，但是这种检测方法是有创的，需要将小动物处死取材，这导致需要大量的小动物个体，也意味着无法对同一小动物个体进行长时间的监测和跟踪观察。
骨骼肌收缩特性检测是指对离体肌肉如比目鱼肌和趾长伸肌等进行30摄氏度灌流并给予一定的电压刺激来测量其单次收缩力和强直收缩力。肌肉在孵育过程中，给予肌肉最适的电压刺激后，测得肌肉最适初长度下的最大收缩力，即单收缩力；然后给予肌肉最适电压下的强直刺激，实时记录肌肉在各个刺激时间点的强直收缩力，得到相应的收缩力曲线。虽然骨骼肌收缩特性检测可以对肌肉的收缩特性进行检测，但是这种检测方法同样是有创的，需要将小动物处死取材，这导致需要大量的小动物个体，也意味着无法对同一小动物个体进行长时间的监测和跟踪观察。
四肢爪抓力测定是指通过爪抓力测试仪测定小动物的抓握力量，可以评估小动物前后肢的力量大小，从而评估小动物神经损伤及四肢肌肉损伤程度。将小动物放在爪抓力测试仪的金属网格测量区，小动物会自动抓住金属网格，待其抓稳后，拉住其尾巴向后拉，小动物会本能地抓牢金属网格来抵抗，此时测试仪会产生读数，将其拉离金属网格后，记录这期间测试仪的最大读数，即为该只小动物的前后肢力量大小。通常每只小动物测量三次取平均值。这种检测方法虽可以测量模型小动物的四肢抓握力量，但是并不能实现对活体模型小动物肌肉结构的精细检测和量化评估。
爬杆测试是指通过测试小动物爬行时间和状态来评估小动物爬行速度、肢体协调能力和四肢支撑的力量。通过爬杆装置和计时器来记录小动物的爬杆时间，在正式测试开始前先对小动物进行训练，避免其出现停滞不前或反向往上爬的情况。正式测试时，将小动物放在杆子顶部同时开始计时，待其头部触地即停止计时，所记录的时间即为爬行时间。若小动物出现因肌肉力量不足以支撑其完成整个爬杆实验的情况，则按照程度不同将其划分为单支腿无力、双腿无力和四肢无力。这种检测方法虽可以测量模型小动物 的爬行速度、肢体协调能力和四肢支撑的力量，但是并不能实现对活体模型小动物肌肉结构的精细检测和量化评估。
转棒实验是指通过测试模型小动物在不停转动的转棒上跑步的时间来评估小动物的平衡能力、协调能力和肌肉力量。实验前，先对小动物进行连续五天的转棒适应和学习训练。转棒实验开始时，将转棒仪的转速调至合适的速度，具体速度大小可根据各模型小动物的体型大小等来选择，记录其在转棒仪上的时间，以某一时间为分界值，超过该时间则按照该时间记录，不足该时间的则按照实际时间记录，重复进行三次，每次转棒实验之间间隔休息一定时间。这种检测方法虽可以对模型小动物的平衡能力、协调能力和肌肉力量进行评估，但是并不能实现对活体模型小动物肌肉结构的精细检测和量化评估。
悬挂实验是指通过悬挂网对每只模型小动物的四肢耐力进行测量。将小动物倒置放在悬挂网格上，待其抓稳后，开始计时，当其因力气耗尽从悬挂网格上掉下时停止计时，以某一时间为分界值，超过该时间则按照该时间记录，不足该时间的则按照实际时间记录，重复进行三次，每次间隔一定时间让小动物的体力得到恢复。这种检测方法虽可以测量模型小动物的四肢耐力，但是并不能实现对活体模型小动物肌肉结构的精细检测和量化评估。
旷场实验是指通过统计小动物在敞箱内一定时间的自主爬行总路径来评估小动物的运动自主性和运动能力。将小动物置于明暗箱中，用高清摄像头跟踪并记录其在箱子中一定时间内的运动轨迹，运用旷场实验视频分析软件分析其活动轨迹图，并对爬行总路径进行统计。这种检测方法虽可以对模型小动物的运动自主性和运动能力进行评估，但是并不能实现对活体模型小动物肌肉结构的精细检测和量化评估。
因此，本发明针对上述问题，提供了一种用于模型小动物的肌肉超声图像分析方法，本发明首先利用目前广泛使用的超声设备，可采集活体下模型小动物的肌肉部位的超声图像，具有无创、简便易行且可实现对活体模型小动物肌肉结构的精细检测与可视化等优势。然后，利用图像处理等手段对获取的模型小动物的肌肉超声图像提取肌肉形态学特征、图像频率分析特征、图像纹理特征等，从而可实现对模型小动物肌肉结构的量化评估，并且运用统计学和重复性分析方法对提取的图像特征进行分析，从而筛选出与疾 病高度相关的重要特征以及具有较高测试-重测可靠度的稳定、可靠的特征，构建出针对模型小动物的肌肉超声图像分析系统，最后，输入待分析模型小动物的肌肉超声图像，即可提取出相应特征，并得到分析结果，从而实现对活体模型小动物肌肉结构的精细检测和量化评估。
示例性方法
本实施例提供一种用于模型小动物的肌肉超声图像分析方法。如图1所示，所述方法包括如下步骤：
步骤S100、获取模型小动物的肌肉超声图像，并对所述肌肉超声图像进行ROI划分，得到划分后的肌肉超声图像；
具体地，超声成像设备具有检测肌肉精细结构的潜力，并允许对肌肉结构进行可视化和量化，且具有无创性、无辐射、实时动态监测等优点。本实施例获取模型小动物的肌肉超声图像后，需要对所述肌肉超声图像进行ROI划分，得到划分后的肌肉超声图像。
其中，ROI(region ofinterest)即为感兴趣区域。机器视觉、图像处理中从被处理的图像以方框、圆、椭圆、不规则多边形等方式勾勒出需要处理的区域称为感兴趣区域ROI。在Halcon、OpenCV、Matlab等机器视觉软体上常用到各种运算元(Operator)和函式来求得感兴趣区域ROI并进行图像的下一步处理。
在一种实现方式中，本实施例所述步骤S100包括如下步骤：
步骤S101、对受试模型小动物的后肢进行脱毛处理，在所述受试模型小动物的麻醉状态下，将所述受试模型小动物仰卧式置于实验板中央，并将其口鼻置于连接麻醉仪的管道中；
步骤S102、设置超声成像系统的检测模式为肌骨检测模式；
步骤S103、将超声探头的长轴与受试模型小动物的跟腱平行，通过设置标记保持所述超声探头放置在所述受试模型小动物后肢上的第一检测位置；
步骤S104、在所述受试模型小动物的麻醉状态下，涂以超声凝胶耦合剂以确保超声探头与皮肤之间的声学耦合，调整超声探头以优化超声图像中肌束的对比度，基于所述检测模式和所述第一检测位置，使用实时B型超声成像设备获取所述受试模型小动物的肌肉超声图像；
步骤S105、对所述肌肉超声图像进行灰度化、裁剪、根据人工勾画或自动分割的肌肉区域划分ROI，得到所述划分后的肌肉超声图像。
具体地，关于模型小动物后肢肌肉超声图像的具体采集步骤如下：首先，在超声图像采集前，对模型小动物的后肢进行脱毛处理，以避免其皮毛对肌肉超声图像的采集和质量造成影响；其次，在模型小动物麻醉状态下，将其仰卧式置于实验板中央，并将其口鼻置于连接麻醉仪的管道中，使其在实验过程中一直处于麻醉状态，便于超声图像的采集；然后使用实时超声成像设备获取其后肢的肌肉超声图像，其中超声成像系统选择肌骨检测模式，将超声探头的长轴与跟腱平行，置于后肢的表面或其它特定位置；涂以合适量的超声凝胶耦合剂以确保探头与皮肤之间的声学耦合；可调整超声探头来优化超声图像中肌束的对比度，并可对位置进行标记以确保探头每次都放置在相同的位置。模型小动物后肢肌肉超声图像示例如图2所示。并在提取特征前先对图像进行预处理，包括灰度化、裁剪、以及根据人工勾画或自动分割的肌肉区域划分ROI。
需要注意的是，在肌肉超声图像采集方面，除了采集模型小动物静态下的肌肉超声图像，还可以采集模型小动物肌肉拉伸产生的结构动态变化过程中的肌肉超声图像；采集设备除了B型超声设备外，还可以使用剪切波弹性成像设备或者其他超声成像方式等进行采集。
步骤S200、从所述划分后的肌肉超声图像中提取肌肉形态学特征、图像频率特征以及图像纹理特征；
在一种实现方式中，本实施例所述步骤S200包括如下步骤：
步骤S201、对所述划分后的肌肉超声图像进行归一化雷登变换，得到雷登变换矩阵；
步骤S202、对所述雷登变换矩阵求梯度并进行边缘增强，得到雷登变换梯度矩阵；
步骤S203、根据所述雷登变换梯度矩阵，得到肌肉厚度、肌纤维长度和羽状角特征；
步骤S204、根据所述肌肉厚度、肌纤维长度和羽状角特征，得到所述肌肉形态学特征；
步骤S205、从所述划分后的肌肉超声图像中提取所述平均频率分析特征；其中，所述平均频率分析特征的计算公式为
Figure PCTCN2022138045-appb-000003
n,I,和f分别为功率密度谱的长度、功率和频率；
步骤S206、基于像素灰度分布计算，从所述划分后的肌肉超声图像中提取一阶统计学特征；其中，所述一阶统计学特征包括积分光密度、平均值、标准差、方差、偏度、峰度和能量；
步骤S207、基于灰度共生矩阵计算，从所述划分后的肌肉超声图像中提取Haralick特征；其中，所述Haralick特征包括对比度、相关性、能量、熵、同质性和对称性；
步骤S208、基于灰度游程长度矩阵计算，从所述划分后的肌肉超声图像中提取Galloway特征；其中，所述Galloway特征包括短游程优势、长游程优势、灰度不均匀性、长游程不均匀性、游程百分比；
步骤S209、基于图像局部区域的中心像素与所述中心像素邻域的比较结果，从所述划分后的肌肉超声图像中提取局部二值模式特征；其中，所述局部二值模式特征包括能量和熵，所述能量为LBP energy＝∑ if i 2，所述熵为LBP entropy＝-∑ if i 2log 2(f i)，f i表示第i块在图像局部区域的相应频率；
步骤S210、根据所述一阶统计学特征、所述Haralick特征、所述Galloway特征和所述局部二值模式特征得到所述图像纹理分析特征。
具体地，肌肉的形态学特征包括肌肉厚度、肌纤维长度和羽状角等。肌纤维的长度和羽状角都是基于肌纤维来计算的。但是，由于一些模型小动物如小鼠的后肢肌肉较小，肌肉区域的肌纤维并不能很好的显示出来，因此，其后肢肌肉的形态特征主要为由人工勾勒或算法自动分割的肌肉区域面积。若采集的肌肉区域的肌纤维较为清晰，还可以使用一些特征检测方法来估计肌肉的形态学参数，如基于雷登变换梯度矩阵的特征检测方式等。
平均频率分析特征(mean frequency analysis feature,MFAF)作为一种与肌肉质量相关，并且有望描述骨骼肌结构差异的有效参数，不会显著受不同超声设备的配置所影响。计算公式见下方：
Figure PCTCN2022138045-appb-000004
其中，n,I,和f分别功率密度谱的长度、功率和频率。
图像纹理分析特征主要包括一阶统计学特征、高阶纹理特征。一方面，根据以往研究表明，一阶统计学特征可以有效定量地描述骨骼肌的超声回波强度；此外还有研究表明不同年龄或者组别的骨骼肌的超声回波强度信息存在差异，并且，这些特征还可以提供一些肌肉状态相关的结构信息，从而为肌肉损伤评估提供有效信息。另一方面，高阶纹理特征，如Haralick特征，Galloway特征和局部二值模式(Local Binary Pattern,LBP)特征等，相比于一阶统计学特征，在精细任务如肌肉性别识别中能够表现地更好。具体地，一阶统计学特征是直接基于原始图像的像素灰度分布而计算出来的特征，包括积分光密度、平均值、标准差、方差、偏度、峰度和能量等特征。Haralick特征是由灰度共生矩阵计算而来，灰度共生矩阵是像素距离与方向的矩阵函数，在给定空间距离d和方向θ计算两点灰度值之间的相关性，是一种通过研究灰度的空间相关特性来描述图像纹理的常用方法。典型的Haralick特征包括对比度、相关性、能量、熵、同质性和对称性等。通常，每个Haralick特征包含0°、45°、90°和135°四个方向。Galloway特征是基于灰度游程长度矩阵计算而来，灰度游程长度矩阵表示一张图像的纹理变化的规律性，其大小由图像的灰度水平和图像大小决定，包括短游程优势、长游程优势、灰度不均匀性、长游程不均匀性、游程百分比等纹理统计特征，且每个Galloway特征也包含0°、45°、90°和135°四个方向。LBP特征是通过将图像局部区域的中心像素与其邻域进行比较得到的，主要描述图像的局部纹理特征，具有旋转不变性和灰度不变性等显著优点，包括能量(energy)和熵(entropy)，具体计算公式见下方：
LBP energy＝∑ if i 2
LBP entropy＝-∑ if i 2log 2(f i)
其中f i表示第i块在图像局部区域的相应频率。
需要注意的是，在提取肌肉超声图像的肌肉形态学特征，如羽状角、肌肉厚度、肌纤维角度、肌束长度、肌肉生理横截面积等特征时，除了使用基于雷登变换梯度矩阵的特征检测方式外，还可使用基于霍夫变换、深度学习等方法来定位肌肉组织的各结构要素，从而实现形态学参数的自动测量。
步骤S300、对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行统计学分析得到有统计学差异特征，进行基于测试-重测可信度的重复性分析得到可靠特征；
统计学分析是通过收集大量的数据/数字，运用统计学方法从大量数据中寻找规律特征，实现大量数据直观化、可视化，以便了解数据中蕴含的统计规律，从而用数据的规律特征解释问题。重复性分析是通过多次重复测量来评估图像特征的测试-重测可信度等级，以筛选出稳定、可靠的特征。
具体地，本实施例运用统计学和重复性分析方法对提取的图像特征进行分析，从而筛选出与疾病高度相关的重要特征以及具有较高测试-重测可靠度的稳定、可靠的特征。
在一种实现方式中，本实施例所述步骤S300包括如下步骤：
步骤S301、对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行正态性检验，得到第一检验结果；
步骤S302、若所述第一检验结果服从正态，则对所述第一检验结果进行方差齐性检验，得到第二检验结果；
步骤S303、若所述第一检验结果不服从正态，则对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行非参数检验，得到所述有统计学差异特征；
步骤S304、若所述第二检验结果满足方差齐性，则对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行独立样本T检验，得到所述有统计学差异特征；
步骤S305、若所述第二检验结果满足方差非齐性，则对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行修正方差T检验，得到所述有统计学差异特征；
步骤S306、对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行重复性测试，得到若干测试结果；
步骤S307、对所述若干测试结果采用类内相关系数ICC(1,1)评估可靠度；其中，所述相关系数为
Figure PCTCN2022138045-appb-000005
其中，BMS和WMS分别为由Kruskal-Wallis单因素方差分析得到的组间和组内的均方根，k为重复测量次数；
步骤S308、根据所述可靠度得到所述可靠特征。
具体地，对提取的图像特征进行统计学分析和重复性分析。分析的具体步骤如下：首先，通过统计学分析将各分组之间有显著性差异的特征筛选出来。做统计学分析时， 首先做正态性检验和方差齐性检验，如果数据服从正态分布且方差齐性则执行独立样本T检验；如果数据服从正态分布且方差不齐，则执行修正方差T检验Welch’s t-test；如果数据不服从正态分布，则执行非参数检验Mann-Whitney test。p<0.05被认为是统计学上有显著性差异。统计学分析流程图如图3所示。第二，通过评估图像特征的测试-重测可信度等级来筛选出稳定、可靠的特征，在此，采用类内相关系数ICC(1,1)(由下式给出)来评估测试-重测可靠度。
Figure PCTCN2022138045-appb-000006
其中，BMS和WMS分别表示由Kruskal-Wallis单因素方差分析得到的组间和组内的均方根，k为重复测量次数。
需要注意的是，在对提取的图像特征进行分析时，除了可以进行统计学分析和重复性分析外，还可对特征进行相关性分析、对比分析等。
步骤S400、根据所述有统计学差异特征和所述可靠特征，得到肌肉超声图像的分析模型；
具体地，运用统计学和重复性分析方法对提取的图像特征进行分析，从而筛选出与疾病高度相关的重要特征以及具有较高测试-重测可靠度的稳定、可靠的特征，就可以基于所述有统计学差异特征和所述可靠特征，得到肌肉超声图像的分析模型。
步骤S500、将待分析的模型小动物的肌肉超声图像输入到所述肌肉超声图像的分析模型进行分析，得到所述待分析的模型小动物肌肉结构的检测和量化评估结果。
具体地，构建出针对模型小动物的肌肉超声图像分析系统，输入待分析模型小动物的肌肉超声图像，即可提取出相应特征，并得到分析结果，从而实现对活体模型小动物肌肉结构的精细检测和量化评估。
示例性装置
本实施例还提供一种用于模型小动物的肌肉超声图像分析装置，所述装置包括：
ROI划分模块10，用于获取模型小动物的肌肉超声图像，并对所述肌肉超声图像进行ROI划分，得到划分后的肌肉超声图像；
特征提取模块20，用于从所述划分后的肌肉超声图像中提取肌肉形态学特征、图像频率特征以及图像纹理特征；
特征分析模块30，用于对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行统计学分析得到有统计学差异特征，进行基于测试-重测可信度的重复性分析得到可靠特征；
分析模型获取模块40，用于根据所述有统计学差异特征和所述可靠特征，得到肌肉超声图像的分析模型；
分析模块50，用于将待分析的模型小动物的肌肉超声图像输入到所述肌肉超声图像的分析模型进行分析，得到所述待分析的模型小动物肌肉结构的检测和量化评估结果。
在一种实现方式中，所述ROI划分模块10包括：
预处理单元，用于对受试模型小动物的后肢进行脱毛处理，在所述受试模型小动物的麻醉状态下，将所述受试模型小动物仰卧式置于实验板中央，并将其口鼻置于连接麻醉仪的管道中；
检测模式设置单元，用于设置超声成像系统的检测模式为肌骨检测模式；
超声探头放置单元，用于将超声探头的长轴与受试模型小动物的跟腱平行，通过设置标记保持所述超声探头放置在所述受试模型小动物后肢上的第一检测位置；
肌肉超声图像采集单元，用于在所述受试模型小动物的麻醉状态下，涂以超声凝胶耦合剂以确保超声探头与皮肤之间的声学耦合，调整超声探头以优化超声图像中肌束的对比度，基于所述检测模式和所述第一检测位置，使用实时B型超声成像设备获取所述受试模型小动物的肌肉超声图像；
ROI划分单元，用于对所述肌肉超声图像进行灰度化、裁剪、根据人工勾画或自动分割的肌肉区域划分ROI，得到所述划分后的肌肉超声图像。
在一种实现方式中，所述特征提取模块20包括：
雷登变换矩阵获取单元，用于对所述划分后的肌肉超声图像进行归一化雷登变换，得到雷登变换矩阵；
雷登变换梯度矩阵获取单元，用于对所述雷登变换矩阵求梯度并进行边缘增强，得到雷登变换梯度矩阵；
第一特征获取单元，用于根据所述雷登变换梯度矩阵，得到肌肉厚度、肌纤维长度和羽状角特征；
肌肉形态学特征获取单元，用于根据所述肌肉厚度、肌纤维长度和羽状角特征，得到所述肌肉形态学特征；
平均频率分析特征获取单元，用于从所述划分后的肌肉超声图像中提取所述平均频率分析特征；其中，所述平均频率分析特征的计算公式为
Figure PCTCN2022138045-appb-000007
n,I,和f分别为功率密度谱的长度、功率和频率；
一阶统计学特征提取单元，用于基于像素灰度分布计算，从所述划分后的肌肉超声图像中提取一阶统计学特征；其中，所述一阶统计学特征包括积分光密度、平均值、标准差、方差、偏度、峰度和能量；
Haralick特征提取单元，用于基于灰度共生矩阵计算，从所述划分后的肌肉超声图像中提取Haralick特征；其中，所述Haralick特征包括对比度、相关性、能量、熵、同质性和对称性；
Galloway特征提取单元，用于基于灰度游程长度矩阵计算，从所述划分后的肌肉超声图像中提取Galloway特征；其中，所述Galloway特征包括短游程优势、长游程优势、灰度不均匀性、长游程不均匀性、游程百分比；
局部二值模式特征提取单元，用于基于图像局部区域的中心像素与所述中心像素邻域的比较结果，从所述划分后的肌肉超声图像中提取局部二值模式特征；其中，所述局部二值模式特征包括能量和熵，所述能量为LBP energy＝∑ if i 2，所述熵为LBP entropy＝-∑ if i 2log 2(f i)，f i表示第i块在图像局部区域的相应频率；
图像纹理分析特征提取单元，用于根据所述一阶统计学特征、所述Haralick特征、所述Galloway特征和所述局部二值模式特征得到所述图像纹理分析特征。
在一种实现方式中，所述特征分析模块30包括：
第一检验结果获取单元，用于对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行正态性检验，得到第一检验结果；
第二检验结果获取单元，用于若所述第一检验结果服从正态，则对所述第一检验结果进行方差齐性检验，得到第二检验结果；
第一有统计学差异特征获取单元，用于若所述第一检验结果不服从正态，则对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行非参数检验，得到所述有统计学差异特征；
第二有统计学差异特征获取单元，用于若所述第二检验结果满足方差齐性，则对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行独立样本T检验，得到所述有统计学差异特征；
第三有统计学差异特征获取单元，用于若所述第二检验结果满足方差非齐性，则对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行修正方差T检验，得到所述有统计学差异特征；
重复性测试单元，用于对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行重复性测试，得到若干测试结果；
可靠度获取单元，用于对所述若干测试结果采用类内相关系数ICC(1,1)评估可靠度；其中，所述相关系数为
Figure PCTCN2022138045-appb-000008
其中，BMS和WMS分别为由Kruskal-Wallis单因素方差分析得到的组间和组内的均方根，k为重复测量次数；
可靠特征获取单元，用于根据所述可靠度得到所述可靠特征。
基于上述实施例，本发明还提供了一种智能终端，其原理框图可以如图5所示。该智能终端包括通过系统总线连接的处理器、存储器、网络接口、显示屏、温度传感器。其中，该智能终端的处理器用于提供计算和控制能力。该智能终端的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作系统和计算机程序。该内存储器为非易失性存储介质中的操作系统和计算机程序的运行提供环境。该智能终端的网络接口用于与外部的终端通过网络连接通信。该计算机程序被处理器执行时以实现一种基于肌肉超声的影像特征提取及分类方法。该智能终端的显示屏可以是液晶显示屏或者电子墨水显示屏，该智能终端的温度传感器是预先在智能终端内部设置，用于检测内部设备的运行温度。
本领域技术人员可以理解，图5中示出的原理框图，仅仅是与本发明方案相关的部分结构的框图，并不构成对本发明方案所应用于其上的智能终端的限定，具体的智能终端以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，所述的计算机程序可存储于一非易失性计算机可读取存储介质中，该计算机程序在执行时，可包括如上述各方法的实施例的流程。其中，本发明所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用，均可包括非易失性和/或易失性存储器。非易失性存储器可包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM)或闪存。易失性存储器可包括随机存取存储器(RAM)或者外部高速缓冲存储器。作为说明而非局限，RAM以多种形式可得，诸如静态RAM(SRAM)、动态RAM(DRAM)、同步DRAM(SDRAM)、双数据率SDRAM(DDRSDRAM)、增强型SDRAM(ESDRAM)、同步链路(Synchlink)DRAM(SLDRAM)、存储器总线(Rambus)直接RAM(RDRAM)、直接存储器总线动态RAM(DRDRAM)、以及存储器总线动态RAM(RDRAM)等。
综上，本发明公开了一种用于模型小动物的肌肉超声图像分析方法，方法包括：对模型小动物的肌肉超声图像进行ROI划分，并从划分后的肌肉超声图像中提取肌肉形态学特征、图像频率特征以及图像纹理特征；对所述特征进行统计学分析得到有统计学差异特征，进行基于测试-重测可信度的重复性分析得到可靠特征；根据有统计学差异特征和可靠特征，得到肌肉超声图像的分析模型；将待分析的模型小动物的肌肉超声图像输入到肌肉超声图像的分析模型进行分析，得到待分析的模型小动物肌肉结构的检测和量化评估结果。本发明可采集活体下模型小动物的肌肉部位的超声图像，具有无创、简便易行且可实现对活体模型小动物肌肉结构的精细检测与可视化的优势。
最后应说明的是：以上实施例仅用以说明本发明的技术方案，而非对其限制；尽管参照前述实施例对本发明进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本发明各实施例技术方案的精神和范围。

Claims (10)

  1. 一种用于模型小动物的肌肉超声图像分析方法，其特征在于，所述方法包括：
    获取模型小动物的肌肉超声图像，并对所述肌肉超声图像进行ROI划分，得到划分后的肌肉超声图像；
    从所述划分后的肌肉超声图像中提取肌肉形态学特征、图像频率特征以及图像纹理特征；
    对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行统计学分析得到有统计学差异特征，进行基于测试-重测可信度的重复性分析得到可靠特征；
    根据所述有统计学差异特征和所述可靠特征，得到肌肉超声图像的分析模型；
    将待分析的模型小动物的肌肉超声图像输入到所述肌肉超声图像的分析模型进行分析，得到所述待分析的模型小动物肌肉结构的检测和量化评估结果。
  2. 根据权利要求1所述的用于模型小动物的肌肉超声图像分析方法，其特征在于，所述获取模型小动物的肌肉超声图像，包括：
    对受试模型小动物的后肢进行脱毛处理，在所述受试模型小动物的麻醉状态下，将所述受试模型小动物仰卧式置于实验板中央，并将其口鼻置于连接麻醉仪的管道中；
    设置超声成像系统的检测模式为肌骨检测模式；
    将超声探头的长轴与受试模型小动物的跟腱平行，通过设置标记保持所述超声探头放置在所述受试模型小动物后肢上的第一检测位置；
    在所述受试模型小动物的麻醉状态下，涂以超声凝胶耦合剂以确保超声探头与皮肤之间的声学耦合，调整超声探头以优化超声图像中肌束的对比度，基于所述检测模式和所述第一检测位置，使用实时B型超声成像设备获取所述受试模型小动物的肌肉超声图像。
  3. 根据权利要求1所述的用于模型小动物的肌肉超声图像分析方法，其特征在于，所述对所述肌肉超声图像进行ROI划分，得到划分后的肌肉超声图像，包括：
    对所述肌肉超声图像进行灰度化、裁剪、根据人工勾画或自动分割的肌肉区域划分ROI，得到所述划分后的肌肉超声图像。
  4. 根据权利要求1所述的用于模型小动物的肌肉超声图像分析方法，其特征在于，所述从所述划分后的肌肉超声图像中提取肌肉形态学特征、图像频率特征，包括：
    对所述划分后的肌肉超声图像进行归一化雷登变换，得到雷登变换矩阵；
    对所述雷登变换矩阵求梯度并进行边缘增强，得到雷登变换梯度矩阵；
    根据所述雷登变换梯度矩阵，得到肌肉厚度、肌纤维长度和羽状角特征；
    根据所述肌肉厚度、肌纤维长度和羽状角特征，得到所述肌肉形态学特征；
    从所述划分后的肌肉超声图像中提取所述平均频率分析特征；其中，所述平均频率分析特征的计算公式为
    Figure PCTCN2022138045-appb-100001
    n,I,和f分别为功率密度谱的长度、功率和频率。
  5. 根据权利要求1所述的用于模型小动物的肌肉超声图像分析方法，其特征在于，所述从所述划分后的肌肉超声图像中提取图像纹理分析特征，包括：
    基于像素灰度分布计算，从所述划分后的肌肉超声图像中提取一阶统计学特征；其中，所述一阶统计学特征包括积分光密度、平均值、标准差、方差、偏度、峰度和能量；
    基于灰度共生矩阵计算，从所述划分后的肌肉超声图像中提取Haralick特征；其中，所述Haralick特征包括对比度、相关性、能量、熵、同质性和对称性；
    基于灰度游程长度矩阵计算，从所述划分后的肌肉超声图像中提取Galloway特征；其中，所述Galloway特征包括短游程优势、长游程优势、灰度不均匀性、长游程不均匀性、游程百分比；
    基于图像局部区域的中心像素与所述中心像素邻域的比较结果，从所述划分后的肌肉超声图像中提取局部二值模式特征；其中，所述局部二值模式特征包括能量和熵，所述能量为LBP energy＝Σ if i 2，所述熵为LBP entropy＝-Σ if i 2log 2(f i)，f i表示第i块在图像局部区域的相应频率；
    根据所述一阶统计学特征、所述Haralick特征、所述Galloway特征和所述局部二值模式特征得到所述图像纹理分析特征。
  6. 根据权利要求5所述的用于模型小动物的肌肉超声图像分析方法，其特征在于，所述对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行统计学分析得到有统计学差异特征，包括：
    对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行正态性检验，得到第一检验结果；
    若所述第一检验结果服从正态，则对所述第一检验结果进行方差齐性检验，得到第二检验结果；
    若所述第一检验结果不服从正态，则对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行非参数检验，得到所述有统计学差异特征；
    若所述第二检验结果满足方差齐性，则对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行独立样本T检验，得到所述有统计学差异特征；
    若所述第二检验结果满足方差非齐性，则对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行修正方差T检验，得到所述有统计学差异特征。
  7. 根据权利要求5所述的用于模型小动物的肌肉超声图像分析方法，其特征在于，所述进行基于测试-重测可信度的重复性分析得到可靠特征，包括：
    对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行重复性测试，得到若干测试结果；
    对所述若干测试结果采用类内相关系数ICC(1,1)评估可靠度；其中，所述相关系数为
    Figure PCTCN2022138045-appb-100002
    其中，BMS和WMS分别为由Kruskal-Wallis单因素方差分析得到的组间和组内的均方根，k为重复测量次数；
    根据所述可靠度得到所述可靠特征。
  8. 一种用于模型小动物的肌肉超声图像分析装置，其特征在于，所述装置包括：
    ROI划分模块，用于获取模型小动物的肌肉超声图像，并对所述肌肉超声图像进行ROI划分，得到划分后的肌肉超声图像；
    特征提取模块，用于从所述划分后的肌肉超声图像中提取肌肉形态学特征、图像频率特征以及图像纹理特征；
    特征分析模块，用于对所述肌肉形态学特征、图像频率特征以及图像纹理特征进行统计学分析得到有统计学差异特征，进行基于测试-重测可信度的重复性分析得到可靠特征；
    分析模型获取模块，用于根据所述有统计学差异特征和所述可靠特征，得到肌肉超声图像的分析模型；
    分析模块，用于将待分析的模型小动物的肌肉超声图像输入到所述肌肉超声图像的分析模型进行分析，得到所述待分析的模型小动物肌肉结构的检测和量化评估结果。
  9. 一种智能终端，其特征在于，所述智能终端包括存储器、处理器及存储在所述存储器中并可在所述处理器上运行的用于模型小动物的肌肉超声图像分析程序，所述处理器执行所述用于模型小动物的肌肉超声图像分析程序时，实现如权利要求1-7任一项所述的用于模型小动物的肌肉超声图像分析方法的步骤。
  10. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质上存储有用于模型小动物的肌肉超声图像分析程序，所述用于模型小动物的肌肉超声图像分析程序被处理器执行时，实现如权利要求1-7任一项所述的用于模型小动物的肌肉超声图像分析方法的步骤。
PCT/CN2022/138045 2022-09-08 2022-12-09 一种用于模型小动物的肌肉超声图像分析方法 WO2024051016A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202211093345.7 2022-09-08
CN202211093345.7A CN115486869A (zh) 2022-09-08 2022-09-08 一种用于模型小动物的肌肉超声图像分析方法

Publications (1)

Publication Number Publication Date
WO2024051016A1 true WO2024051016A1 (zh) 2024-03-14

Family

ID=84468270

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/138045 WO2024051016A1 (zh) 2022-09-08 2022-12-09 一种用于模型小动物的肌肉超声图像分析方法

Country Status (2)

Country Link
CN (1) CN115486869A (zh)
WO (1) WO2024051016A1 (zh)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6099473A (en) * 1999-02-05 2000-08-08 Animal Ultrasound Services, Inc. Method and apparatus for analyzing an ultrasonic image of a carcass
US20150323503A1 (en) * 2014-05-12 2015-11-12 Openshaw Beck LLC Method and system of assessing or analyzing muscle characteristics including strength and tenderness using ultrasound
CN110516762A (zh) * 2019-10-10 2019-11-29 深圳大学 一种肌肉状态量化评定方法、装置、存储介质及智能终端
CN110693526A (zh) * 2019-11-11 2020-01-17 深圳先进技术研究院 一种肌肉疾病评估方法、系统及电子设备
JP2020130596A (ja) * 2019-02-19 2020-08-31 株式会社Ｃｅｓデカルト 超音波診断装置、超音波診断プログラム及び超音波エコー画像の解析方法
CN112288733A (zh) * 2020-11-06 2021-01-29 深圳先进技术研究院 一种肌肉超声图像检测方法、系统、终端以及存储介质
CN114241187A (zh) * 2021-11-25 2022-03-25 深圳高性能医疗器械国家研究院有限公司 基于超声双模态影像的肌肉疾病诊断系统、设备及介质

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6099473A (en) * 1999-02-05 2000-08-08 Animal Ultrasound Services, Inc. Method and apparatus for analyzing an ultrasonic image of a carcass
US20150323503A1 (en) * 2014-05-12 2015-11-12 Openshaw Beck LLC Method and system of assessing or analyzing muscle characteristics including strength and tenderness using ultrasound
JP2020130596A (ja) * 2019-02-19 2020-08-31 株式会社Ｃｅｓデカルト 超音波診断装置、超音波診断プログラム及び超音波エコー画像の解析方法
CN110516762A (zh) * 2019-10-10 2019-11-29 深圳大学 一种肌肉状态量化评定方法、装置、存储介质及智能终端
CN110693526A (zh) * 2019-11-11 2020-01-17 深圳先进技术研究院 一种肌肉疾病评估方法、系统及电子设备
CN112288733A (zh) * 2020-11-06 2021-01-29 深圳先进技术研究院 一种肌肉超声图像检测方法、系统、终端以及存储介质
CN114241187A (zh) * 2021-11-25 2022-03-25 深圳高性能医疗器械国家研究院有限公司 基于超声双模态影像的肌肉疾病诊断系统、设备及介质

Also Published As

Publication number Publication date
CN115486869A (zh) 2022-12-20

Similar Documents

Publication Publication Date Title
US20020016539A1 (en) Method and apparatus for measuring and classifying optically observable changes in skin and mucous membrane
DE102012108121A1 (de) Verfahren und System für ultraschallgestützte automatische Erkennung, Quantifizierung und Nachverfolgung von Pathologien
Fang et al. Anisotropic mechanical properties of the human uterus measured by spherical indentation
WO2021023051A1 (zh) 组织评价方法、装置、设备及计算机可读存储介质
CN109472798A (zh) 生猪脂肪含量检测模型训练方法及生猪脂肪含量检测方法
CN114241187A (zh) 基于超声双模态影像的肌肉疾病诊断系统、设备及介质
US20130316389A1 (en) Method and System for Determining Characteristics of an Embryo and Uses Thereof
Rutkove et al. Electrical impedance myography detects age-related skeletal muscle atrophy in adult zebrafish
Xu et al. Automatic classification of male and female skeletal muscles using ultrasound imaging
WO2024051016A1 (zh) 一种用于模型小动物的肌肉超声图像分析方法
CN112674791B (zh) 肌肉超声弹性成像的优化方法及系统
CN111281426B (zh) 表浅神经的超声成像及图像分析方法
CN111528918A (zh) 消融术后肿瘤体积变化趋势图生成装置、设备及存储介质
Rutkove et al. Surface Electrical Impedance Myography Detects Skeletal Muscle Atrophy in Aged Wildtype Zebrafish and Aged gpr27 Knockout Zebrafish
Li et al. Diagnosis of renal diseases based on machine learning methods using ultrasound images
CN210776809U (zh) 一种望诊图像质量评价系统、设备及存储介质
CN112634218B (zh) 一种基于横断位磁共振图像的腰椎肌肉测量方法
CN116433560A (zh) 基于超声双模态影像的肌肉疾病识别方法、装置及设备
CN110599560A (zh) 磁共振成像方法、装置、存储介质及计算机设备
Fang et al. Equilibrium Mechanical Properties of the Nonhuman Primate Cervix
CN111062945A (zh) 一种早癌医学图像与基因型ai识别系统及其方法和应用
US20240319163A1 (en) Ovum Evaluation Method, Ovum Evaluation Apparatus, and Ovum Evaluation Program
Kerensky et al. Quantifying tethered spinal cord tension: Ultrasound elastography results from computational, cadaveric, and intraoperative first-in-human studies
CN117530659B (zh) 基于光学相干层析成像技术的早期宫颈癌变辅助诊断工具
WO2023287084A1 (ko) 의료 초음파의 정량 정보 추출 방법 및 장치

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22957975

Country of ref document: EP

Kind code of ref document: A1