Gynecological osteoporosis image-oriented analysis environment construction method and system
By using cosine similarity and gray-level co-occurrence matrix techniques to refine the classification of osteoporosis images, this method solves the problem of analysis process deviating from key areas in existing technologies, and achieves fast and accurate osteoporosis image analysis, supporting automatic switching of multiple analysis parameters with different focuses.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANDONG UNIV OF TRADITIONAL CHINESE MEDICINE
- Filing Date
- 2026-04-02
- Publication Date
- 2026-06-19
Smart Images
Figure CN121962151B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data analysis technology, and in particular to a method and system for constructing an analysis environment for gynecological osteoporosis images. Background Technology
[0002] In recent years, with the simultaneous maturation of multimodal imaging, deep learning and low-dose hardware, the technology for constructing analysis environments for osteoporosis images has shown a development status of full-link AI and platform-based deployment. Hybrid models of 2D / 3D U-Net, YOLO and Transformer integrate vertebral segmentation, BMD regression and fracture recognition into the same deep learning flow, and have high T-value accuracy in cross-center external validation.
[0003] Currently, Chinese invention patent CN119784752A discloses an osteoporosis detection system based on nuclear magnetic resonance imaging. This method obtains intervertebral disc images by training an image segmentation model, segments the preprocessed images to obtain regions of interest (ROI) images, determines the first osteoporosis level of the ROI, inputs the ROI image to an expert platform to obtain a second osteoporosis level, and then performs a second operation. This invention improves the accuracy and efficiency of detection through the collaborative work of a pre-operation module, an image processing module, an analysis module, and a precision module. However, related technologies do not provide detailed classifications of location and severity, and cannot easily switch between different analytical parameters focusing on calcified areas, cancellous areas, and cortical areas. This may cause the analysis process to deviate from the key areas, resulting in invalid or redundant analysis, which is detrimental to the speed of analysis and the specificity of the analysis environment. Summary of the Invention
[0004] The technical problem solved by this invention is that related technologies do not classify the location and severity in detail, and cannot switch between different analytical parameters such as calcified area, cancellous area, and cortical area with one click. As a result, the analysis process may deviate from the key area, leading to invalid and redundant analysis, which is not conducive to the speed of analysis and the specificity of the analysis environment.
[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution. In the first aspect, a method for constructing an analysis environment for gynecological osteoporosis images includes the following steps: Step S100, determining whether the retrieved image has been desensitized; if so, proceeding to step S200; if not, performing a desensitization operation; after the desensitization operation, proceeding to step S200.
[0006] Step S200: Based on cosine similarity, the images are classified and the elastic features of the classified images are obtained. Then, a gray-level co-occurrence matrix is constructed, and evolutionary features are extracted based on the gray-level co-occurrence matrix.
[0007] Step S300: The classified images are reclassified based on evolutionary features, and image labels are configured. The evolutionary features are the contrast of the gray-level co-occurrence matrix and the variance of the gray-level co-occurrence matrix.
[0008] As a preferred embodiment of the method for constructing an analysis environment for gynecological osteoporosis images according to the present invention, the image is represented as an X-ray photoacoustic image;
[0009] The desensitization operation is performed by eliminating X-ray photoacoustic images containing sensitive characters. After the elimination operation is completed, the relevant characters in the X-ray photoacoustic image are deleted. When the elimination operation is not required, the process jumps directly to the step of determining whether the relevant characters in the X-ray photoacoustic image need to be deleted.
[0010] As a preferred embodiment of the method for constructing an analysis environment for gynecological osteoporosis images according to the present invention, the shape features of the standard image are extracted and denoted as the first shape feature, and the shape features of the X-ray photoacoustic image are extracted and denoted as the second shape feature. Based on the cosine similarity formula, the similarity between the first shape feature and the second shape feature is calculated. A preset similarity threshold is set, and the similarity and the similarity threshold are compared. Based on the comparison results of the similarity and the similarity threshold, the X-ray photoacoustic images are classified.
[0011] As a preferred embodiment of the method for constructing an analysis environment for gynecological osteoporosis images according to the present invention, step S200 includes the following sub-steps, including step S201, selecting any group of classified X-ray photoacoustic images, constructing a tissue region according to the relative position, and modeling the particles in the tissue region.
[0012] Step S202: Based on the pulse energy calculation formula and the relationship expression between single pulse energy and displacement rise time, the maximum displacement duration is obtained, and the elastic characteristics are calculated by combining the definite integral formula.
[0013] Step S203: Construct a gray-level co-occurrence matrix based on the elasticity features, and extract evolutionary features based on the gray-level co-occurrence matrix.
[0014] As a preferred embodiment of the method for constructing an analysis environment for gynecological osteoporosis images according to the present invention, particles are represented as atoms, and modeling particles in the tissue region is represented by constructing a correspondence between particle layer temperature rise and positional modulus, including... ;
[0015] in, For particle density, The average temperature rise caused by particle absorption of light energy is a function of the irradiation duration. The horizontal axis represents the position modulus, and the vertical axis represents the irradiation duration. It is expressed as the average temperature rise produced by any layer of particles at a vertical distance r from the X-ray laser pulse emitter over time t. For the specific heat capacity of blood, This represents the average temperature rise caused by a particle absorbing light energy per unit time, where r is the position modulus. It is expressed as the partial derivative of the average temperature rise caused by particle absorption of light energy with respect to the irradiation time as a function of the average temperature rise with respect to the x-axis.
[0016] As a preferred embodiment of the method for constructing an analysis environment for gynecological osteoporosis images according to the present invention, the repetition frequency and average power are obtained, the single pulse energy is obtained according to the pulse energy calculation formula, and the maximum displacement duration is obtained according to the relationship expression between the single pulse energy and the displacement rise time.
[0017] The formula for calculating pulse energy is as follows: ;
[0018] The expression relating pulse energy and displacement rise time is as follows: ;
[0019] Where E is the energy of a single pulse. Average power, The repetition frequency, The calibration parameter is K, which is a constant. Let be the radius of the incident light spot. This represents the duration of the maximum displacement.
[0020] As a preferred embodiment of the method for constructing an analysis environment for gynecological osteoporosis images according to the present invention, the duration of the maximum expansion deformation of the tissue region corresponding to the emission parameters is obtained by using the maximum displacement duration.
[0021] The elastic characteristics are calculated by combining the maximum displacement time with the definite integral formula. The expression for the calculation of the elastic characteristics is as follows: ;
[0022] Where U represents the elastic characteristic. Let j be a particle in the j-th layer, where j is distributed from 1 to N, and N is a constant. The vertical distance between the Nth layer particles and the X-ray laser pulse emitter. This represents the vertical distance between the first layer of particles and the X-ray laser pulse emitter.
[0023] As a preferred embodiment of the method for constructing an analysis environment for gynecological osteoporosis images according to the present invention, the method involves obtaining alkaline phosphatase content, osteocalcin concentration, and serum type I procollagen unfolding peptide content. Using alkaline phosphatase content as the first dependent variable and the elastic characteristics of each particle layer as the independent variable, a first multiple regression analysis is performed to obtain a first regression equation. Using osteocalcin concentration as the second dependent variable and the elastic characteristics of each particle layer as the independent variable, a second multiple regression analysis is performed to obtain a second regression equation. Using serum type I procollagen unfolding peptide content as the third dependent variable and the elastic characteristics of each particle layer as the independent variable, a third multiple regression analysis is performed to obtain a third regression equation.
[0024] As a preferred embodiment of the method for constructing an analysis environment for gynecological osteoporosis images according to the present invention, the method involves calculating a first ratio of the regression coefficients of the second regression equation to the independent variables corresponding to the first regression equation, and calculating a second ratio of the regression coefficients of the third regression equation to the independent variables corresponding to the first regression equation. Since the number of particle layers is N, an N*N matrix is constructed based on the first ratio and the second ratio.
[0025] Secondly, a system for constructing an analysis environment for gynecological osteoporosis images is provided, including a desensitization module, a computation module, and a classification module.
[0026] The desensitization module desensitizes images that have not been desensitized.
[0027] The calculation module classifies images based on cosine similarity and obtains the elastic features of the classified images, thereby constructing a gray-level co-occurrence matrix and extracting evolutionary features based on the gray-level co-occurrence matrix.
[0028] The classification module reclassifies the categorized images based on evolutionary features and assigns image labels.
[0029] The beneficial effects of this invention are as follows: By pre-classifying images using cosine similarity, tens of thousands of images can be compressed into hundreds of templates, reducing repetitive computational effort and enabling rapid incremental image matching. Through three-level extraction of elastic features, gray-level co-occurrence matrix, and evolutionary features, it can construct mechanical and textural dual-mode verification, self-correcting false positives and false negatives caused by single-mode noise, scientifically classifying images, and constructing a refined analysis environment for subsequent image analysis. Different parts and different degrees of severity require different focused analysis areas. Through the scientific classification of this application, the parts and severity can be automatically mapped to dedicated ROI templates. It can switch between different focused analysis parameters such as calcification area, cancellous area, and cortical area with one click, achieving refined scanning of different areas in the same image and different sides of the same area. It establishes a hierarchical prior for subsequent ALP / OC / PIPIN biochemical regression, locking high, medium, and low-risk bone metabolism intervals within the corresponding gray-level and elastic intervals, improving the goodness of fit of the regression equation. Attached Figure Description
[0030] Figure 1 This is a basic flowchart illustrating a method for constructing an analysis environment for gynecological osteoporosis images, as provided in one embodiment of the present invention. Detailed Implementation
[0031] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.
[0032] It should be understood that the step numbers used herein are for ease of description only and are not intended to limit the order in which the steps are performed. It should also be understood that the terminology used in this specification is for the purpose of describing specific embodiments only and is not intended to limit the invention.
[0033] Unless the context clearly indicates otherwise, the singular forms “one,” “a,” and “the” are intended to include the plural forms.
[0034] The terms “comprising” and “including” indicate the presence of the described feature, whole, step, operation, element and / or component, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.
[0035] The term “and / or” refers to any combination of one or more of the associated listed items, as well as all possible combinations, and includes these combinations.
[0036] With the simultaneous maturation of multimodal imaging, deep learning, and low-dose hardware, the technology for constructing analysis environments for osteoporosis images is showing a development status of full-link AI and platform-based deployment. Hybrid models of 2D / 3D U-Net, YOLO, and Transformer integrate vertebral segmentation, BMD regression, and fracture recognition into the same deep learning flow, achieving high T-value accuracy in cross-center external validation.
[0037] Based on this, embodiments of this application provide a method and system for constructing an analysis environment for gynecological osteoporosis images.
[0038] The method for constructing an analysis environment for gynecological osteoporosis images provided in this application is specifically illustrated through the following embodiments. First, the method for constructing an analysis environment for gynecological osteoporosis images in this application is described.
[0039] The method for constructing an analysis environment for gynecological osteoporosis images provided in this application relates to the field of data analysis. The method for constructing an analysis environment for osteoporosis images provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the method for constructing the analysis environment for osteoporosis images, but is not limited to the above forms.
[0040] This application can also be used in numerous general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.
[0041] Example, refer to Figure 1 As an embodiment of the present invention, a method for constructing an analysis environment for gynecological osteoporosis images is provided, including the following steps: Step S100, determining whether the retrieved image has been desensitized; if yes, proceeding to step S200; if no, performing a desensitization operation; after the desensitization operation, proceeding to step S200.
[0042] Step S200: Based on cosine similarity, the images are classified and the elastic features of the classified images are obtained. Then, a gray-level co-occurrence matrix is constructed, and evolutionary features are extracted based on the gray-level co-occurrence matrix.
[0043] Step S300: The classified images are reclassified based on evolutionary features, and image labels are configured. The evolutionary features are the contrast of the gray-level co-occurrence matrix and the variance of the gray-level co-occurrence matrix.
[0044] More preferably, this application, through cosine similarity pre-classification, can compress tens of thousands of images into hundreds of templates, reducing redundant computing power and enabling rapid incremental image matching. Through three-level extraction of elastic features, gray-level co-occurrence matrix, and evolutionary features, it can construct mechanical and textural dual-mode verification, self-correct false positives and false negatives caused by single-mode noise, and scientifically classify images. It can construct a refined analysis environment for subsequent image analysis, with different emphasis on analysis areas for different parts and different severity levels. Through the scientific classification of this application, the parts and severity can be automatically mapped to dedicated ROI templates. It can switch between different emphasis analysis parameters such as calcification area, cancellous area, and cortical area with one click, achieving refined scanning of different areas in the same image and different sides of the same area. It establishes a hierarchical prior for subsequent ALP / OC / PIPIN biochemical regression, and can lock high, medium, and low risk bone metabolism intervals into corresponding gray-level and elastic intervals, improving the goodness of fit of the regression equation.
[0045] The image is represented as an X-ray photoacoustic map.
[0046] More preferably, in this application, the X-ray photoacoustic images in the default image database are all X-ray photoacoustic images of the ROI region obtained after machine vision processing steps, that is, the part of the X-ray photoacoustic image that is greater than or equal to the first ratio is the joint image to be analyzed, or the part that is greater than or equal to the first ratio is the vertebral image to be analyzed.
[0047] In this application, it is assumed that the X-ray photoacoustic image has already undergone noise reduction processing. For example, Morlet wavelet is selected as the mother wavelet, and broadband pass processing is performed by combining multiple wavelets (i.e., combined wavelets). The lowest spectral distribution range in the combined wavelets is selected according to the imaging resolution, and the highest spectral distribution range in the combined wavelets is selected according to the imaging sensitivity.
[0048] More preferably, the images in the image database are represented as X-ray photoacoustic images of joints prone to osteoporosis symptoms, or X-ray photoacoustic images of vertebrae prone to osteoporosis symptoms.
[0049] More preferably, the vertebrae prone to osteoporosis symptoms include thoracic T7 to T12 vertebrae and lumbar L1 to L4 vertebrae;
[0050] Joints prone to osteoporosis include the hip joint, wrist joint, proximal scapular joint, sacral joint, and hip joint.
[0051] The desensitization operation is performed by eliminating X-ray photoacoustic images containing sensitive characters. After the elimination operation is completed, the relevant characters in the X-ray photoacoustic image are deleted. When the elimination operation is not required, the process jumps directly to the step of determining whether the relevant characters in the X-ray photoacoustic image need to be deleted.
[0052] More preferably, the X-ray photoacoustic image is represented as a light absorption distribution map of human tissue, which can extract the elastic characteristics of the human tissue from the X-ray photoacoustic image through a preset algorithm.
[0053] More preferably, an X-ray free electron laser (XFEL) is emitted by an X-ray pulse laser emitter to generate light energy. The human tissue absorbs the generated light energy, and based on the principle of thermal expansion and contraction, the local temperature rise of the human tissue causes the tissue to expand. The expansion of the tissue causes the surrounding material to be compressed, and the surrounding material forms a pressure wave (i.e., ultrasound). The pressure wave (i.e., ultrasound) is received by an ultrasound transducer to obtain the light absorption distribution map of the tissue.
[0054] More preferably, the X-ray photoacoustic imager is equipped with a DICOM (Distributed Information Module), which includes categories such as age information, name information, gender information, detection data, and emission data. The detection data includes bone density, alkaline phosphatase content, osteocalcin concentration, and serum type I procollagen unfolding peptide content. The emission data includes the relative position to the joint or vertebral body, repetition frequency, average power, and incident light plate radius.
[0055] More preferably, the relative position with respect to the joint or vertebral body includes a position vector and a position modulus;
[0056] The method for calculating the position vector and position modulus includes: selecting the geometric center point of the joint or vertebra, denoted as the first key point; obtaining the geometric center point of the X-pulse laser emitter, denoted as the second key point; constructing a coordinate system with the first key point as the origin, the axis passing through the origin and having the first direction as the x-axis, the axis passing through the origin and having the second direction as the y-axis, and the direction perpendicular to the plane formed by the x-axis and y-axis as the z-axis; obtaining the coordinates of the second key point in the coordinate system; denoteing the vector connecting the origin to the coordinates of the second key point in the coordinate system as the position vector; and denoteing the magnitude of the position vector as the position modulus.
[0057] More preferably, step S100 includes the following steps: step S101, obtaining any X-ray photoacoustic image from the image database;
[0058] Step S102: Detect whether there is text on the X-ray photoacoustic image. If there is text, remove the text and proceed to step S103. If there is no text, proceed to step S103.
[0059] Step S103: Extract the DICOM of the X-ray photoacoustic image, and identify whether the age information, name information, and gender information categories in the DICOM include characters. If characters are included, perform the desensitization operation; if characters are not included, proceed to the next X-ray photoacoustic image.
[0060] The desensitization operation means deleting all characters from the categories of age information, name information, and gender information.
[0061] Step S104: Repeat steps S101 to S103 until all X-ray photoacoustic images are displayed, then stop repeating steps S101 to S103.
[0062] The shape features of the standard image are extracted and denoted as the first shape feature, and the shape features of the X-ray photoacoustic image are extracted and denoted as the second shape feature. Based on the cosine similarity formula, the similarity between the first shape feature and the second shape feature is calculated. A preset similarity threshold is set, and the similarity is compared with the similarity threshold. Based on the comparison results of the similarity and the similarity threshold, the X-ray photoacoustic image is classified.
[0063] More preferably, the first shape feature quantity and the second shape feature quantity are extracted by machine vision algorithms, including contour feature comparison algorithms, region feature comparison algorithms, texture feature comparison algorithms, shape context feature comparison algorithms, etc.
[0064] The standard images include standard joint images and standard vertebral images, represented as standard medical images corresponding to joints prone to osteoporosis symptoms and standard medical images corresponding to vertebral bodies prone to osteoporosis symptoms in the influence database.
[0065] More preferably, the first value is set as the similarity threshold.
[0066] More preferably, the method for classifying X-ray photoacoustic images includes comparing the similarity with a first value;
[0067] When the similarity is greater than or equal to the first value, the joint name or vertebral body name corresponding to the standard image is set to the joint name or vertebral body name corresponding to the current X-ray photoacoustic image;
[0068] When the similarity is less than the first value, jump to the next standard image, repeat the similarity calculation steps, and repeat the similarity comparison steps until the similarity is greater than or equal to the first value. Then stop jumping to the next standard image and set the joint name or vertebral body name corresponding to the current standard image as the joint name or vertebral body name corresponding to the current X-ray photoacoustic image.
[0069] Number the joint or vertebral body name; the joint or vertebral body name is numbered X. i Where i is a natural number, representing the i-th name, the X-ray photoacoustic images are numbered according to the joint name or vertebral body name and in the order of calculation. The X-ray photoacoustic images are numbered as X ijLet j be a natural number, representing the j-th X-ray photoacoustic image under the i-th name. Complete the classification of X-ray photoacoustic images. The classification of X-ray photoacoustic images is represented as Xi={Xi1, Xi2, ..., Xij}.
[0070] Step S200 includes the following sub-steps, including step S201, selecting any group of classified X-ray photoacoustic images, constructing tissue regions according to their relative positions, and modeling the particles in the tissue regions;
[0071] Step S202: Based on the pulse energy calculation formula and the relationship expression between single pulse energy and displacement rise time, the maximum displacement duration is obtained, and the elastic characteristics are calculated by combining the definite integral formula.
[0072] Step S203: Construct a gray-level co-occurrence matrix based on the elasticity features, and extract evolutionary features based on the gray-level co-occurrence matrix.
[0073] More preferably, the method for constructing a tissue region based on relative position includes: obtaining a position vector and a position vector; enlarging the position vector by a second ratio in the opposite direction of the position vector; translating the position vector along the enlarged position vector until the starting point of the position vector coincides with the endpoint of the position vector near the joint or vertebral body; stopping the translation of the position vector to obtain a new starting point of the position vector; drawing a sphere with the new starting point of the position vector as the center and a first length as the radius; removing the portion of the sphere belonging to the joint or vertebral body; and setting the remaining portion of the sphere as the tissue region.
[0074] More preferably, particles are represented as atoms, and modeling particles in the tissue region is represented by constructing a correspondence between particle layer temperature rise and positional modulus, including: ;
[0075] in, For particle density, The average temperature rise caused by particle absorption of light energy is a function of the irradiation duration. The horizontal axis represents the position modulus, and the vertical axis represents the irradiation duration. It is expressed as the average temperature rise produced by any layer of particles at a vertical distance r from the X-ray laser pulse emitter over time t. For the specific heat capacity of blood, This represents the average temperature rise caused by a particle absorbing light energy per unit time, where r is the position modulus. It is expressed as the partial derivative of the average temperature rise caused by particle absorption of light energy with respect to the irradiation time, with respect to the horizontal axis.
[0076] More preferably, the particle density is the same as the bone density in the detection data. Through the above formula, the relationship between the temperature rise of the particle layer and the position modulus can be obtained. That is, under any emission parameter, the average temperature rise of the particle layer with different position moduli can be obtained.
[0077] More preferably, the repetition frequency and average power are obtained, the single pulse energy is obtained according to the pulse energy calculation formula, and the maximum displacement duration is obtained according to the relationship expression between the single pulse energy and the displacement rise time.
[0078] The formula for calculating pulse energy is as follows: ;
[0079] The expression relating pulse energy and displacement rise time is as follows: ;
[0080] Where E is the energy of a single pulse. Average power, The repetition frequency, The calibration parameter is K, which is a constant. Let be the radius of the incident light spot. This represents the duration of the maximum displacement.
[0081] More preferably, the duration of the maximum expansion deformation in the tissue region corresponding to the emission parameters is obtained by using the maximum displacement duration;
[0082] More preferably, the elastic characteristics are calculated by combining the maximum displacement time with the definite integral formula. The expression for calculating the elastic characteristics is as follows: ;
[0083] Where U represents the elastic characteristic. Let j be a particle in the j-th layer, where j is distributed from 1 to N, and N is a constant. The vertical distance between the Nth layer particles and the X-ray laser pulse emitter. This represents the vertical distance between the first layer of particles and the X-ray laser pulse emitter.
[0084] More preferably, alkaline phosphatase content, osteocalcin concentration, and serum type I procollagen unfolding peptide content are obtained. Using alkaline phosphatase content as the first dependent variable and the elastic characteristics of each particle layer as the independent variable, a first multiple regression analysis is performed to obtain a first regression equation. Using osteocalcin concentration as the second dependent variable and the elastic characteristics of each particle layer as the independent variable, a second multiple regression analysis is performed to obtain a second regression equation. Using serum type I procollagen unfolding peptide content as the third dependent variable and the elastic characteristics of each particle layer as the independent variable, a third multiple regression analysis is performed to obtain a third regression equation.
[0085] More preferably, the first ratio of the regression coefficients of the second regression equation to the corresponding independent variables of the first regression equation is calculated, and the second ratio of the regression coefficients of the third regression equation to the corresponding independent variables of the first regression equation is calculated. Since the number of particle layers is N, an N*N matrix is constructed based on the first ratio and the second ratio.
[0086] The method for constructing an N*N matrix includes combining the first ratio and the second ratio, and mapping the combined first ratio and the second ratio to gray values. The mapping expression is HD=F(C), where HD represents the gray value, F is the mapping rule, and C is the average of the first ratio and the second ratio.
[0087] When constructing the matrix, the first ratio and the second ratio are represented as x and y to distinguish them. The first ratio and the second ratio are combined to form pixels. The combination methods include: combining the first first ratio with the first second ratio, combining the first first ratio with the second second ratio, ..., combining the first first ratio with the Nth second ratio, combining the second first ratio with the first second ratio, ..., combining the second first ratio with the Nth second ratio, ..., combining the Nth first ratio with the Nth second ratio, resulting in N*N combinations.
[0088] The particles satisfy the constraint condition that the average value of the elastic characteristics of particles in any layer is equal to the elastic characteristics of the corresponding layer.
[0089] Set a first direction and a first distance. Starting from the grayscale value of the first combination, search according to the first direction and the first distance until all combinations are traversed, obtaining each grayscale value pair. A grayscale value pair is represented as a combination of the grayscale value of the first combination and the searched grayscale value. Convert the grayscale value pairs into a grayscale co-occurrence matrix according to their order of appearance. The expression for the grayscale co-occurrence matrix is as follows: ;
[0090] Where P is the gray-level co-occurrence matrix. Represented as grayscale value pairs ( The number of grayscale value pairs in P is used to normalize P so that the values of each element in P are distributed between 0 and 1.
[0091] Obtain the contrast and variance of the gray-level co-occurrence matrix, and set the contrast and variance as evolutionary features.
[0092] A method for reclassifying categorized images based on evolutionary features includes setting a second value as a contrast threshold, setting a third value as a variance threshold, comparing the evolutionary features with the contrast and variance thresholds, and classifying the categorized image as X when the contrast is less than or equal to the second value and the variance is less than or equal to the third value. iFurthermore, to the first degree, when the contrast is less than or equal to the second value and the variance is greater than the third value, the classified image is classified as X. i Furthermore, in the second stage, when the contrast is greater than the second value and the variance is less than or equal to the third value, the classified image is classified as X. i Furthermore, in the third stage, when the contrast is greater than the second value and the variance is greater than the third value, the classified image is classified as X. i And the fourth degree;
[0093] The first, second, third, and fourth grades indicate increasingly severe osteoporosis.
[0094] Methods for configuring image labels include setting X i And first degree, X i And the second degree, X i And the third degree, X i The fourth degree is set as the label of the corresponding image.
[0095] More preferably, this application, through cosine similarity pre-classification, can compress tens of thousands of images into hundreds of templates, reducing redundant computing power and enabling rapid incremental image matching. Through three-level extraction of elastic features, gray-level co-occurrence matrix, and evolutionary features, it can construct mechanical and textural dual-mode verification, self-correct false positives and false negatives caused by single-mode noise, and scientifically classify images. It can construct a refined analysis environment for subsequent image analysis, with different emphasis on analysis areas for different parts and different severity levels. Through the scientific classification of this application, the parts and severity can be automatically mapped to dedicated ROI templates. It can switch between different emphasis analysis parameters such as calcification area, cancellous area, and cortical area with one click, achieving refined scanning of different areas in the same image and different sides of the same area. It establishes a hierarchical prior for subsequent ALP / OC / PIPIN biochemical regression, and can lock high, medium, and low risk bone metabolism intervals into corresponding gray-level and elastic intervals, improving the goodness of fit of the regression equation.
[0096] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0097] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the protection scope of the present invention.
Claims
1. A method of constructing an analysis environment for gynecological osteoporosis images, characterized by, The process includes the following steps: Step S100: Determine whether the retrieved image has been desensitized. If yes, proceed to step S200; otherwise, perform the desensitization operation. After the desensitization operation, proceed to step S200. Step S200: Based on cosine similarity, the images are classified and the elastic features of the classified images are obtained. Then, a gray-level co-occurrence matrix is constructed, and evolutionary features are extracted based on the gray-level co-occurrence matrix. Step S200 includes the following sub-steps, including step S201, selecting any group of classified X-ray photoacoustic images, constructing tissue regions according to their relative positions, and modeling the particles in the tissue regions; Step S202: Based on the pulse energy calculation formula and the relationship expression between single pulse energy and displacement rise time, the maximum displacement duration is obtained, and the elastic characteristics are calculated by combining the definite integral formula. Step S203: Construct a gray-level co-occurrence matrix based on the elasticity features, and extract evolutionary features based on the gray-level co-occurrence matrix; Particles are represented as atoms. Modeling particles within an organization region involves constructing a relationship between particle layer temperature rise and positional modulus, including... ; in, For particle density, The average temperature rise caused by particle absorption of light energy is a function of the irradiation duration. The horizontal axis represents the position modulus, and the vertical axis represents the irradiation duration. It is expressed as the average temperature rise produced by any layer of particles at a vertical distance r from the X-ray laser pulse emitter over time t. For the specific heat capacity of blood, This represents the average temperature rise caused by a particle absorbing light energy per unit time, where r is the position modulus. It is expressed as the partial derivative of the average temperature rise caused by particle absorption of light energy with respect to the irradiation time, with respect to the horizontal axis. Obtain the repetition frequency and average power, calculate the single pulse energy according to the pulse energy calculation formula, and obtain the maximum displacement duration according to the relationship expression between single pulse energy and displacement rise time. The formula for calculating pulse energy is as follows: ; The expression of the relationship between the pulse energy and the displacement rise time is ; wherein E is a single pulse energy, is an average power, is a repetition frequency, is a calibration parameter, i.e. K is a constant, is an incident light spot radius, is a maximum displacement duration; The duration of maximum expansion deformation in the tissue region corresponding to the emission parameters is obtained by using the maximum displacement duration. The maximum displacement duration is combined with the definite integral formula to calculate the elastic characteristic, and the calculation expression of the elastic characteristic is, ; Where U represents the elastic characteristic. Let j be a particle in the j-th layer, where j is distributed from 1 to N, and N is a constant. The vertical distance between the Nth layer particles and the X-ray laser pulse emitter. The vertical distance between the first layer of particles and the X-ray laser pulse emitter; Step S300: The classified images are reclassified based on evolutionary features, and image labels are configured. The evolutionary features are the contrast of the gray-level co-occurrence matrix and the variance of the gray-level co-occurrence matrix.
2. The method for constructing an analysis environment for gynecological osteoporosis images as described in claim 1, characterized in that, The image is represented as an X-ray photoacoustic image; The desensitization operation is performed by eliminating X-ray photoacoustic images containing sensitive characters. After the elimination operation is completed, the relevant characters in the X-ray photoacoustic image are deleted. When the elimination operation is not required, the process jumps directly to the step of determining whether the relevant characters in the X-ray photoacoustic image need to be deleted.
3. The method for constructing an analysis environment for gynecological osteoporosis images as described in claim 2, characterized in that, The shape features of the standard image are extracted and denoted as the first shape feature, and the shape features of the X-ray photoacoustic image are extracted and denoted as the second shape feature. Based on the cosine similarity formula, the similarity between the first shape feature and the second shape feature is calculated. A preset similarity threshold is set, and the similarity is compared with the similarity threshold. Based on the comparison results of the similarity and the similarity threshold, the X-ray photoacoustic image is classified.
4. The method for constructing an analysis environment for gynecological osteoporosis images as described in claim 1, characterized in that, Alkaline phosphatase content, osteocalcin concentration, and serum type I procollagen unfolding peptide content were obtained. A first multiple regression analysis was performed with alkaline phosphatase content as the first dependent variable and the elastic characteristics of each particle layer as the independent variable, yielding the first regression equation. A second multiple regression analysis was performed with osteocalcin concentration as the second dependent variable and the elastic characteristics of each particle layer as the independent variable, yielding the second regression equation. A third multiple regression analysis was performed with serum type I procollagen unfolding peptide content as the third dependent variable and the elastic characteristics of each particle layer as the independent variable, yielding the third regression equation.
5. The method for constructing an analysis environment for gynecological osteoporosis images as described in claim 4, characterized in that, Calculate the first ratio of the regression coefficients of the independent variables corresponding to the second regression equation and the first regression equation, and calculate the second ratio of the regression coefficients of the independent variables corresponding to the third regression equation and the first regression equation. Since the number of particle layers is N, construct an N*N matrix based on the first and second ratios.
6. A system for constructing an analysis environment for gynecological osteoporosis images, the system being used to execute the method for constructing an analysis environment for gynecological osteoporosis images as described in claim 1, characterized in that, It includes a desensitization module, a calculation module, and a classification module; The desensitization module desensitizes images that have not been desensitized. The calculation module classifies images based on cosine similarity and obtains the elastic features of the classified images, thereby constructing a gray-level co-occurrence matrix and extracting evolutionary features based on the gray-level co-occurrence matrix. The classification module reclassifies the categorized images based on evolutionary features and assigns image labels.