Rheumatism immune disease joint injury assessment system based on image recognition

Abstract

The invention relates to the technical field of image data processing, in particular to a rheumatism immune disease joint injury assessment system based on image recognition. The system comprises a joint image preprocessing module, an image segmentation and injury characteristic analysis module, a bone erosion and synovial thickening evaluation module and a joint injury evaluation and injury visualization module, can obtain joint image data of the rheumatism immune disease, perform size scaling and image denoising processing, and perform joint part image segmentation at the same time. Image data of the joint part region of the rheumatism immune disease is obtained; and performing joint injury structural feature analysis and joint image injury quantitative evaluation on the joint partial region image data of the rheumatism immune disease to obtain a joint image injury score corresponding to the rheumatism immune disease patient, and performing evaluation to obtain a joint image injury type to perform joint injury visualization so as to obtain a joint injury visual image result. The joint injury degree corresponding to the joint image of the rheumatism immune disease can be more accurately detected.

Description

Technical Field

[0001] The present invention relates to the technical field of image data processing, and in particular to a rheumatic immune disease joint injury assessment system based on image recognition. Background Art

[0002] In recent years, the application of image recognition technology in medical imaging has gradually emerged, especially in the diagnosis of rheumatic and immune diseases and the assessment of joint injuries. Image recognition technology can automatically and accurately analyze joint injuries through deep learning, artificial intelligence algorithms, etc., with high efficiency and accuracy. At the same time, X-ray imaging, as the most common means of joint injury examination, can provide information about joint space narrowing, bone injury, joint deformity, etc., but it lacks sensitivity to soft tissue injuries and has limited ability to detect early minor injuries. In addition, although MRI can more accurately display soft tissue injuries (such as synovitis, ligament injury, etc.), its equipment is expensive and the examination process is relatively complicated, so it is not suitable for routine screening of all patients. However, the existing image recognition methods based on artificial intelligence are mainly used for image classification, detection and segmentation, which can help doctors to assist to a certain extent. However, since the traditional model relies on manual settings to extract image features, and the quality and quantity of training data sets are relatively limited, its accuracy and generalization ability when dealing with complex cases and diverse imaging manifestations still need to be improved. Summary of the invention

[0003] Based on this, it is necessary for the present invention to provide a rheumatic autoimmune disease joint injury assessment system based on image recognition to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a rheumatic immune disease joint damage assessment system based on image recognition includes the following modules:

[0005] The joint image preprocessing module is used to obtain the joint image data of rheumatism and immune disease, and to perform size scaling and image denoising on the joint image data of rheumatism and immune disease to obtain the standard image data of rheumatism and immune disease joints;

[0006] The image segmentation and injury feature analysis module is used to perform joint segmentation on the standard image data of rheumatoid arthritis joints using the U-Net neural network to obtain image data of the partial area of ​​the joints of rheumatoid arthritis disease; perform joint injury structural feature analysis on the image data of the partial area of ​​the joints of rheumatoid arthritis disease to obtain the structural features of joint injury of rheumatoid arthritis disease, including the width of the joint space of rheumatoid arthritis disease, the surface morphology of the joints of rheumatoid arthritis disease, the bone changes of the joints of rheumatoid arthritis disease, and the abnormalities of the soft tissue of the joints of rheumatoid arthritis disease;

[0007] The bone erosion and synovial thickening assessment module is used to obtain the corresponding trabecular orientation and synovial thickness of the rheumatoid articular joint region through the image data of the rheumatoid articular joint region, and to perform bone erosion and thickening assessment and analysis on the image data of the rheumatoid articular joint region based on the trabecular orientation and synovial thickness of the rheumatoid articular joint region, so as to obtain the size of the bone erosion area and the degree of synovial thickening of the rheumatoid articular joint;

[0008] The joint injury assessment and damage visualization module is used to perform quantitative joint imaging damage assessment on the structural characteristics of rheumatic immune disease joint injuries based on the size of the rheumatic immune joint bone erosion area and the degree of synovial thickening of the rheumatic immune joints, so as to obtain the joint imaging damage score corresponding to the patients with rheumatic immune diseases; to assess the joint injury type of the imaging data of partial areas of the rheumatic immune disease joints, and to obtain the joint imaging damage type corresponding to the patients with rheumatic immune diseases; to visualize the joint damage of the corresponding rheumatic immune disease joint images according to the joint imaging damage score and joint imaging damage type corresponding to the patients with rheumatic immune diseases, so as to obtain the joint damage visualization image results corresponding to the patients with rheumatic immune diseases.

[0009] Furthermore, the joint image preprocessing module includes the following functions:

[0010] Obtain joint imaging data of rheumatism and immune diseases;

[0011] Performing size scaling processing on rheumatism-immunity disease joint image data to obtain corresponding rheumatism-immunity joint image data within the same size range;

[0012] Grayscale processing is performed on the rheumatism immune joint image data corresponding to the same size range to obtain the rheumatism immune joint grayscale image data;

[0013] Calculate the pixel fuzziness of the grayscale image data of the rheumatoid arthritis joint to obtain the pixel fuzziness value of the rheumatoid arthritis joint image;

[0014] Based on the pixel fuzziness value of the rheumatoid arthritis joint image, the image noise of the grayscale image data of the rheumatoid arthritis joint is removed and standardized to obtain the standard image data of the rheumatoid arthritis joint.

[0015] Furthermore, the image segmentation and damage feature analysis module includes the following functions:

[0016] The U-Net neural network is used to segment the joint images of the standard image data of rheumatism and immune joints, and the images of the non-joint tissue regions are eliminated to obtain the image data of the joint regions of rheumatism and immune diseases, including the images of the joint bones, cartilage, synovium and muscle tissue regions.

[0017] Perform three-dimensional reconstruction of the joint structure of the image data of the partial area of ​​the rheumatic immune disease joint to generate a three-dimensional model of the joint structure area of ​​the rheumatic immune disease; obtain the dynamic change law of the joints corresponding to the flexion, extension and rotation movements of the patients with rheumatic immune disease, and quantify the joint space width of the three-dimensional model of the joint structure area of ​​the rheumatic immune disease based on the dynamic change law of the joints corresponding to the flexion, extension and rotation movements of the patients with rheumatic immune disease to obtain the joint space width of the rheumatic immune disease;

[0018] Perform resolution scanning and surface morphology analysis on the articular cartilage surface corresponding to the three-dimensional model of the joint structure area of ​​rheumatic immune disease to obtain the surface morphology of the rheumatic immune disease joint, including the roughness, undulation and collagen fiber orientation of the articular cartilage surface;

[0019] Perform statistical analysis on the changes in joint bone mass in the three-dimensional model of the joint structure area of ​​rheumatic immune diseases to obtain the changes in joint bone mass in rheumatic immune diseases;

[0020] The soft tissue echo and soft tissue texture complexity corresponding to the joint structure of rheumatoid immune disease are obtained through the three-dimensional model of the joint structure area of ​​rheumatoid immune disease, and the soft tissue abnormality is evaluated and analyzed based on the soft tissue echo and soft tissue texture complexity to obtain the soft tissue abnormality of the joint of rheumatoid immune disease.

[0021] Furthermore, the method of using the U-Net neural network to perform joint image segmentation on the rheumatism immune joint standard image data includes:

[0022] Performing image contrast enhancement processing on the standard image data of rheumatism and immune joints to highlight the contrast difference between the joint part and the non-joint tissue part, and obtaining the rheumatism and immune joint contrast enhanced image data;

[0023] Perform fast Fourier frequency domain transformation on the contrast-enhanced image data of rheumatoid immune joints to obtain the frequency domain maps of rheumatoid immune joint tissue characteristics corresponding to different frequency components;

[0024] The distribution law of morphological characteristics of the corresponding joint tissue at different frequency scales was determined by the frequency domain atlas of rheumatoid immune joint tissue characteristics corresponding to different frequency components, and the image segmentation network architecture was constructed using the U-Net neural network based on the distribution law of morphological characteristics of joint tissue at different frequency scales. The U-Net neural network architecture for rheumatoid immune joint image segmentation was obtained by using a 3x3 convolution kernel at the shallow layer of the network front end and adjusting it to a 5x5 convolution kernel after 3 layers, with 6 layers and introducing an attention fusion mechanism as a skip connection mode.

[0025] The rheumatology and immunity joint image segmentation U-Net neural network architecture is pre-trained using the general image data corresponding to the joint structure in the public large-scale medical image database to learn the basic human joint tissue segmentation knowledge and feature extraction mode, and obtain the rheumatology and immunity joint image rough segmentation model;

[0026] A small amount of annotated standard rheumatoid arthritis joint image data is used to fine-tune the coarse segmentation model of rheumatoid arthritis joint images. The manual annotation standards generated by the region growing algorithm are used as supervision information to optimize the network model weights. The rheumatoid arthritis joint contrast-enhanced image data are then finely segmented into joint parts, and images of non-joint tissue areas are eliminated to obtain image data of partial joint areas of rheumatoid arthritis diseases, including images of joint bone, cartilage, synovium, and partial muscle tissue areas.

[0027] Furthermore, the quantification of the joint gap width of the three-dimensional model of the joint structure area of ​​the rheumatic immune disease based on the dynamic change law of the joints corresponding to the flexion, extension and rotation movements of the rheumatic immune disease patients includes:

[0028] Based on the dynamic change rules of joints corresponding to flexion, extension and rotation of patients with rheumatic immune diseases, the key nodes of joint movement of the three-dimensional model of the joint structure area of ​​rheumatic immune diseases are determined, and the displacement extreme points and velocity turning points corresponding to the joints of rheumatic immune diseases in the flexion, extension and rotation cycle are obtained;

[0029] According to the displacement extreme points and velocity turning points corresponding to the rheumatoid arthritis joints in the flexion and extension and rotation cycles, node trajectory generation processing is performed to obtain the key node trajectory of the joint motion corresponding to the flexion and extension and rotation movements;

[0030] By taking the corresponding joint motion key node trajectory under flexion, extension and rotation as a guide, the corresponding three-dimensional model of the joint structure area of ​​rheumatic immune disease is imported into the finite element analysis software to simulate the corresponding force deformation of the joint under flexion, extension and rotation, so as to obtain the dynamic structural deformation model of the joint of rheumatic immune disease;

[0031] Through the dynamic structural deformation model of rheumatic immune disease joints, virtual light is emitted to the joint gap from different perspectives at each operation moment of the joint, and the coordinates of the joint gap boundary points are determined according to the reflection and refraction characteristics corresponding to the virtual light, so as to obtain the dynamic range point set of the joint gap boundary of rheumatic immune disease;

[0032] The joint space width of rheumatic immune disease joint space boundary dynamic range point set is quantified to obtain the joint space width of rheumatic immune disease.

[0033] Furthermore, the statistical analysis of joint bone changes in the three-dimensional model of the joint structure area of ​​rheumatic immune diseases includes:

[0034] By selecting the corresponding rheumatoid arthritis joint bone region in the three-dimensional model of the rheumatoid arthritis joint structure region;

[0035] Determine the mineral content of the rheumatic immune disease joint bone area corresponding to the rheumatic immune disease patient to obtain the mineral content corresponding to the rheumatic immune disease joint bone area;

[0036] The trabecular structure of the rheumatic immune disease joint bone area corresponding to the rheumatic immune disease patient is quantified to obtain the trabecular structure parameters corresponding to the rheumatic immune disease joint bone area, including the trabecular volume fraction, trabecular thickness and trabecular spacing;

[0037] A bone index database of healthy people is obtained, and based on the bone index database of healthy people, a statistical analysis of the mineral content and trabecular structure parameters corresponding to the joint bone area of ​​rheumatic immune disease is performed to obtain the joint bone changes of rheumatic immune disease.

[0038] Furthermore, the bone erosion and synovial thickening assessment module includes the following functions:

[0039] Obtaining corresponding rheumatoid arthritis joint trabecular region images and rheumatoid arthritis joint synovial region images through partial region image data of rheumatoid arthritis joints;

[0040] The trabecular topological feature analysis is performed on the rheumatoid immune joint trabecular region images to analyze the connectivity and branching structure of the trabecular region, and the corresponding branch point number and backbone length topological features are extracted to obtain the rheumatoid immune trabecular topological feature set; a trabecular three-dimensional spatial geometric model is constructed based on the rheumatoid immune trabecular topological feature set, and the long axis direction of each trabecular unit corresponding to the trabecular three-dimensional spatial geometric model is determined to obtain the long axis direction vector of each rheumatoid immune trabecular;

[0041] Based on the long axis direction vector of each rheumatoid trabecula, the trabecular orientation of the rheumatoid trabecula region image is analyzed to obtain the trabecular orientation of the rheumatoid trabecula region of the rheumatoid trabecula; the synovial thickness of the rheumatoid trabecula region image is measured to obtain the synovial thickness of the rheumatoid trabecula region of the rheumatoid trabecula;

[0042] According to the direction of trabeculae in the rheumatic immune joint area, the corresponding direction of bone erosion expansion of the rheumatic immune disease is simulated and tracked, and based on the direction of bone erosion expansion of the rheumatic immune disease, the bone erosion of the imaging data of part of the rheumatic immune disease joint area is quantified to obtain the size of the bone erosion area of ​​the rheumatic immune joint;

[0043] The corresponding rheumatoid arthritis joint synovial thickness was used to obtain the corresponding rheumatoid arthritis joint synovial thickening rate, and based on the rheumatoid arthritis joint synovial thickening rate, the synovial thickening degree calculation formula was used to perform thickening evaluation on the imaging data of some areas of the joints of rheumatoid arthritis joints to obtain the degree of synovial thickening of the rheumatoid arthritis joints.

[0044] Furthermore, the calculation formula for the degree of synovial thickening is specifically:

[0045]

[0046] Where R is the degree of synovial thickening in rheumatoid arthritis joints, A is the area of ​​the rheumatoid arthritis joint region, r is the spatial position parameter corresponding to the rheumatoid arthritis joint region, t is the time variable parameter, H(r) is the synovial thickness of the rheumatoid arthritis joint region at the regional spatial position r, C(r,t) is the synovial thickening rate of the rheumatoid arthritis joint at the regional spatial position r and time t, α is the synovial thickening rate influence coefficient, γ is the position attenuation coefficient, β is the synovial thickening rate time attenuation factor, and η is the correction coefficient of the degree of synovial thickening in rheumatoid arthritis joints.

[0047] Furthermore, the joint injury assessment and damage visualization module includes the following functions:

[0048] Based on the size of the bone erosion area of ​​rheumatoid arthritis joints and the degree of synovial thickening of rheumatoid arthritis joints, the joint imaging injury quantitative assessment formula was used to quantitatively assess the structural characteristics of joint injury in rheumatoid arthritis diseases, so as to obtain the corresponding joint imaging injury score for patients with rheumatoid arthritis diseases.

[0049] The imaging data of some areas of joints with rheumatic autoimmune diseases are analyzed for joint regional imaging features to obtain the morphological and textural features corresponding to the images of joint areas with rheumatic autoimmune diseases. The morphological and textural features reveal the tissue structure changes corresponding to the damaged areas of the joints, including the imaging manifestations of cartilage damage and osteoporosis.

[0050] Based on the morphological and textural features of the joint area images of rheumatic immune diseases, the unsupervised clustering K-means method is used to evaluate the joint injury type of the image data of the joint area of ​​rheumatic immune diseases, and the corresponding joint image injury type of patients with rheumatic immune diseases is obtained;

[0051] The corresponding joint damage images of rheumatic immune disease joints are visualized according to the corresponding joint image damage scores and joint image damage types of the rheumatic immune disease patients, so as to obtain the corresponding joint damage visualization image results of the rheumatic immune disease patients.

[0052] Furthermore, the calculation formula for quantitative assessment of joint damage is specifically:

[0053]

[0054] Where J is the joint imaging injury score, τ is the observation period of patients with rheumatic autoimmune diseases, t′ is the patient observation time variable, and C p (t′) is the size of the bone erosion area of ​​the rheumatoid arthritis joint at the observation time t′, R is the degree of synovial thickening of the rheumatoid arthritis joint, θ is the weight coefficient of bone erosion, K is the width of the joint space, D c is the joint surface roughness, Q f is the undulation of the joint surface, B is the degree of change in the joint bone, ε is the degree of abnormality of the joint soft tissue, λ is the injury attenuation coefficient, and ξ is the correction coefficient of the joint imaging injury score.

[0055] Beneficial effects of the present invention:

[0056] The image recognition-based joint injury assessment system for rheumatic immune diseases proposed in the present invention is generally composed of a joint image preprocessing module, an image segmentation and injury feature analysis module, a bone erosion and synovial thickening assessment module, and a joint injury assessment and damage visualization module. Compared with the prior art, the beneficial effect of the present application lies in that by acquiring joint image data of patients with rheumatic immune diseases and performing size scaling and image denoising, the goal is to standardize the image data so that it is suitable for subsequent processing and analysis. The size scaling makes the size of the image data uniform, which facilitates the training and reasoning of subsequent models and ensures that the image resolution does not affect the model results. Image denoising can effectively remove the noise interference in the image, which is crucial to improving the accuracy of subsequent image analysis. The noise in the image comes from changes in the patient's body position, interference from the imaging equipment or other external factors. Through denoising, the key structures and details in the image can be retained, irrelevant information can be removed, and the clarity and contrast of the image can be improved, so that further analysis can be more accurate and reliable. Secondly, the U-Net neural network is used to segment the joint images of the standard image data of rheumatism and immune joints. The U-Net neural network is used to segment the joint images of rheumatism and immune diseases and extract the joint area. As a deep learning model widely used in medical image segmentation, U-Net has good feature extraction capabilities. It can achieve efficient segmentation while ensuring image details. Through segmentation, the joint area can be accurately extracted, the interference of non-joint parts can be removed, and the error in the analysis process can be reduced. Subsequently, the extracted joint area image data is used to analyze the structural characteristics of joint injury. This analysis includes key features such as joint gap width, joint surface morphology, joint bone changes, and soft tissue abnormalities. These features are important bases for evaluating joint injuries in patients with rheumatism and immune diseases. For example, changes in joint gap width often reflect the degree of damage to the articular cartilage, changes in joint surface morphology mean degeneration of the joint structure, and bone changes and soft tissue abnormalities can reveal functional damage to the joint. The extraction of these structural features provides a scientific basis for subsequent pathological analysis and image evaluation. Then, through image data analysis, the orientation of trabeculae and the thickness of synovium are extracted, which can accurately capture the subtle changes in the internal structure of joints in patients with rheumatic autoimmune diseases. The analysis of trabecular orientation can reveal the damage of bone tissue in the joint and help evaluate the degree of bone erosion. Synovial thickening reflects the inflammatory response of the joint. Quantification of the degree of thickening helps to judge the activity of the disease and the severity of inflammation. Through this step of analysis, the area of ​​bone erosion and the degree of synovial thickening in the joint can be effectively evaluated, thereby providing a quantitative basis for subsequent risk assessment.Finally, through the comprehensive evaluation of bone erosion area and synovial thickening, the joint damage of rheumatic immune diseases can be further quantified. By quantifying joint damage, accurate analysis of risk immune disease types can be achieved, and joint damage of corresponding rheumatic immune disease joint images can be visualized according to the corresponding joint imaging damage scores and joint imaging damage types of patients with rheumatic immune diseases. By combining the patient's joint imaging damage scores and types, the joint injury results of patients with rheumatic immune diseases can be visualized. Through this visualization technology, the patient's joint damage status can be intuitively displayed, which is convenient for doctors and patients to understand the severity and development trend of the disease. The visualized images present various characteristics of joint injuries such as bone damage, joint space changes, soft tissue abnormalities, etc. in a graphical form, helping doctors to make more accurate judgments and decisions in actual operations. This can provide clear visual effects, enhance patients' participation and understanding of disease management, and thus improve the accuracy and generalization ability of joint damage assessment in rheumatic immune diseases. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0058] Figure 1 It is a module schematic diagram of the image recognition-based rheumatic immune disease joint injury assessment system of the present invention;

[0059] Figure 2 for Figure 1 Schematic diagram of the functional flow of the mid-joint image preprocessing module;

[0060] Figure 3 for Figure 1 Schematic diagram of the functional flow of the image segmentation and damage feature analysis module. DETAILED DESCRIPTION

[0061] The technical system of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0062] To achieve this, please refer to Figures 1 to 3 The present invention provides a rheumatic immune disease joint injury assessment system based on image recognition, the system comprising the following modules:

[0063] The joint image preprocessing module is used to obtain the joint image data of rheumatism and immune disease, and to perform size scaling and image denoising on the joint image data of rheumatism and immune disease to obtain the standard image data of rheumatism and immune disease joints;

[0064] The image segmentation and injury feature analysis module is used to perform joint segmentation on the standard image data of rheumatoid arthritis joints using the U-Net neural network to obtain image data of the partial area of ​​the joints of rheumatoid arthritis disease; perform joint injury structural feature analysis on the image data of the partial area of ​​the joints of rheumatoid arthritis disease to obtain the structural features of joint injury of rheumatoid arthritis disease, including the width of the joint space of rheumatoid arthritis disease, the surface morphology of the joints of rheumatoid arthritis disease, the bone changes of the joints of rheumatoid arthritis disease, and the abnormalities of the soft tissue of the joints of rheumatoid arthritis disease;

[0065] The bone erosion and synovial thickening assessment module is used to obtain the corresponding trabecular orientation and synovial thickness of the rheumatoid articular joint region through the image data of the rheumatoid articular joint region, and to perform bone erosion and thickening assessment and analysis on the image data of the rheumatoid articular joint region based on the trabecular orientation and synovial thickness of the rheumatoid articular joint region, so as to obtain the size of the bone erosion area and the degree of synovial thickening of the rheumatoid articular joint;

[0066] The joint injury assessment and damage visualization module is used to perform quantitative joint imaging damage assessment on the structural characteristics of rheumatic immune disease joint injuries based on the size of the rheumatic immune joint bone erosion area and the degree of synovial thickening of the rheumatic immune joints, so as to obtain the joint imaging damage score corresponding to the patients with rheumatic immune diseases; to assess the joint injury type of the imaging data of partial areas of the rheumatic immune disease joints, and to obtain the joint imaging damage type corresponding to the patients with rheumatic immune diseases; to visualize the joint damage of the corresponding rheumatic immune disease joint images according to the joint imaging damage score and joint imaging damage type corresponding to the patients with rheumatic immune diseases, so as to obtain the joint damage visualization image results corresponding to the patients with rheumatic immune diseases.

[0067] In the embodiment of the present invention, please refer to Figure 1 FIG. 1 is a schematic diagram of a module of a rheumatic immune disease joint injury assessment system based on image recognition according to the present invention. In this example, the rheumatic immune disease joint injury assessment system based on image recognition includes the following modules:

[0068] S1: joint image preprocessing module, used to obtain rheumatism and immune disease joint image data, and perform size scaling and image denoising on the rheumatism and immune disease joint image data to obtain rheumatism and immune disease joint standard image data;

[0069] In an embodiment of the present invention, joint imaging data of patients with rheumatic immune diseases are obtained by using medical imaging acquisition equipment (such as X-ray, MRI, CT, etc.). The data source should be high-resolution images to ensure the clarity of details. After the image data is obtained, it is first resized. In order to ensure the consistency of the data dimension when input into the neural network model, all image data can be resized to a uniform size, for example, to a pixel resolution of 256×256 or 512×512, while retaining the key features of the image. The resizing operation can be completed by an interpolation algorithm (such as bilinear interpolation or cubic interpolation) to ensure that the image information does not lose too many details due to scaling. Next, image denoising is performed. Common image denoising methods include median filtering, Gaussian filtering, bilateral filtering, etc. In this process, a fuzziness value is selected to remove noise caused by factors such as shooting equipment or patient movement, thereby smoothing the image, removing noise and retaining the structural features of the joint area. After denoising, standard image data of rheumatic immune joints are finally obtained.

[0070] S2: Image segmentation and injury feature analysis module, used to use U-Net neural network to perform joint segmentation on standard image data of rheumatoid arthritis joints, so as to obtain image data of partial regions of rheumatoid arthritis joints; perform joint injury structural feature analysis on partial regions of rheumatoid arthritis joint image data, so as to obtain structural features of rheumatoid arthritis joint injury, including joint space width of rheumatoid arthritis joints, surface morphology of rheumatoid arthritis joints, bone changes of rheumatoid arthritis joints, and abnormalities of soft tissue of rheumatoid arthritis joints;

[0071] In an embodiment of the present invention, the image segmentation of the joint part is performed by using a U-Net neural network. U-Net is a typical convolutional neural network architecture, which is widely used in medical image segmentation tasks. The network is characterized by its symmetrical U-shaped structure, which can effectively extract features and retain spatial information. First, the standard image data is input into the pre-trained U-Net network. U-Net will accurately segment the joint area through a series of convolution operations, pooling layers and deconvolution operations. Through this process, U-Net can segment out some areas of the joints of patients with rheumatic autoimmune diseases, usually joint-related structures such as joint cavities, articular cartilages and surrounding bones. After completing the image segmentation, the structural characteristics of joint injuries are analyzed. Specifically, the following features can be extracted: the width of the joint gap, The width of the segmented joint cavity area, especially the narrowness of the joint space, is often used as an important indicator of bone and joint damage in rheumatic and immune diseases; the joint surface morphology, analyzes whether there are abnormal morphologies such as bone spurs and synovial hyperplasia on the joint surface. Common features include irregularity of the joint surface or bone hyperplasia; bone changes, including changes in bone density, bone absorption or bone hyperplasia. By analyzing the bone area in the segmentation results, it can be determined whether there is osteoporosis or bone erosion; soft tissue abnormalities, by analyzing the segmented soft tissue area, it can be determined whether there is swelling, inflammation or other abnormal changes, and finally the structural characteristics of joint damage in rheumatic and immune diseases are obtained, including the width of the joint space in rheumatic and immune diseases, the joint surface morphology in rheumatic and immune diseases, the bone changes in the joints in rheumatic and immune diseases, and the abnormalities of the soft tissue in the joints in rheumatic and immune diseases.

[0072] S3: Bone erosion and synovial thickening assessment module, which is used to obtain the corresponding trabecular orientation and synovial thickness of the rheumatoid articular region through the image data of the rheumatoid articular region, and to perform bone erosion and thickening assessment and analysis on the image data of the rheumatoid articular region based on the trabecular orientation and synovial thickness of the rheumatoid articular region, so as to obtain the size of the bone erosion area and the degree of synovial thickening of the rheumatoid articular region;

[0073] In an embodiment of the present invention, the orientation of trabeculae and the thickness of synovium in the image data of a partial area of ​​a rheumatic immune joint are obtained by using image analysis technology. The extraction of the orientation of trabeculae is usually achieved through directional gradient and texture analysis of the image, and the texture information in the image is extracted using classical image processing algorithms (such as Sobel operator or Canny edge detection). The orientation of trabeculae is further identified through techniques such as Hough transform. For the measurement of synovial thickness, the soft tissue area of ​​the joint can be separated through an image segmentation algorithm (such as threshold segmentation, edge detection), and the thickness of the synovium is measured based on the grayscale value or contour change of the image. Through these techniques, the orientation information of trabeculae in the joint area and the thickness data of the synovium can be obtained. At the same time, based on the extraction results of trabecular orientation and synovial thickness, bone erosion and thickening evaluation and analysis are then performed. First, by comparing the trabecular orientation in the rheumatoid arthritis disease joint imaging data with that of normal joints, the existing bone erosion areas are identified. This process uses pattern recognition-based algorithms, such as structural similarity index (SSIM) or difference map calculation, to detect trabecular damage areas in the image and calculate the bone erosion area. At the same time, synovial thickening evaluation can be performed by quantitatively analyzing the thickness of the synovial area. The average thickness value of the synovial area is compared with the normal reference value to obtain the degree of synovial thickening of the joint. These evaluation results include the size of the bone erosion area and the degree of synovial thickening of the rheumatoid arthritis joints, and ultimately the size of the bone erosion area and the degree of synovial thickening of the rheumatoid arthritis joints are obtained.

[0074] S4: Joint injury assessment and damage visualization module, which is used to perform quantitative joint imaging damage assessment on the structural characteristics of rheumatic immune disease joint injuries based on the size of the rheumatic immune joint bone erosion area and the degree of synovial thickening of the rheumatic immune joints, so as to obtain the corresponding joint imaging damage score for patients with rheumatic immune diseases; to evaluate the joint injury type of the imaging data of partial areas of the rheumatic immune disease joints, and to obtain the corresponding joint imaging damage type for patients with rheumatic immune diseases; to visualize the corresponding rheumatic immune disease joint images according to the corresponding joint imaging damage score and joint imaging damage type for patients with rheumatic immune diseases, so as to obtain the corresponding joint damage visualization imaging results for patients with rheumatic immune diseases.

[0075] In an embodiment of the present invention, the structural characteristics of joint injury are quantitatively evaluated based on the rheumatoid arthritis joint bone erosion area and the degree of synovial thickening, so as to convert the bone erosion area and the degree of synovial thickening into a quantitative injury score by using a comprehensive scoring system, such as the joint injury score (JSN scoring system) or other standardized scales. The process involves assigning weights to each feature (bone erosion, synovial thickening, etc.) and calculating a comprehensive injury score based on its contribution to the overall degree of injury. This score can reflect the overall degree of injury to the patient's joint. In addition, by analyzing the joint injury area segmented from the imaging data, the type of injury is evaluated. For example, the degree of cartilage wear, the degree of bone erosion, or the manifestation of inflammation around the joint can be determined by morphological analysis. Based on these characteristics, the type of joint injury is determined, such as early bone injury, inflammatory joint lesions, or late degenerative changes. Finally, joint damage is visualized by combining the joint image injury score and injury type. The damaged area is highlighted using image processing software, and color coding (such as red or blue) is used to indicate the different degrees of injury. This visualized image can help clinicians observe the patient's joint injuries more intuitively, and ultimately obtain the corresponding joint damage visualization image results for patients with rheumatic autoimmune diseases, so as to achieve a comprehensive assessment and quantitative analysis of joint injuries in patients with rheumatic autoimmune diseases.

[0076] Further, as an embodiment of the present invention, refer to Figure 2 As shown, Figure 1 The functional flow diagram of the joint image preprocessing module in this embodiment includes the following functions:

[0077] S11: Obtain joint imaging data for rheumatological and immune diseases;

[0078] In an embodiment of the present invention, by collecting joint imaging data of patients with rheumatic autoimmune diseases, these imaging data can be acquired through medical imaging equipment, such as magnetic resonance imaging (MRI), computed tomography (CT) or X-ray imaging, and through the hospital's imaging system, the joint imaging data of relevant patients at different time points are acquired. The imaging data needs to include a sufficiently high resolution so that subsequent analysis and processing can accurately capture subtle changes in the joints. During the data acquisition process, the consistency of the image data format should be ensured, such as unifying it into a DICOM format, to facilitate subsequent image processing software to read and process it, ensure image quality, and avoid affecting the accuracy of subsequent analysis due to missing data or excessive noise, and ultimately obtain joint imaging data of rheumatic autoimmune diseases.

[0079] S12: performing size scaling processing on the rheumatism and immune disease joint image data to obtain corresponding rheumatism and immune disease joint image data within the same size range;

[0080] In an embodiment of the present invention, image data from different sources or different specifications are adjusted to a uniform size range. To this end, common image processing techniques are used, such as nearest neighbor interpolation, bilinear interpolation or cubic interpolation. Specifically, the target image size (for example, 512×512 pixels) is first determined, and then interpolation processing is performed according to the size of the original image. Taking bilinear interpolation as an example, a new pixel value is generated by calculating the weighted average of the four surrounding pixels, so that the image retains detail information as much as possible when scaling. The purpose of this operation is to unify the image size so that subsequent processing steps can be performed on the basis of data of the same size. This process needs to ensure that the main features of the image content are not lost during the scaling process, and finally the corresponding rheumatoid arthritis joint image data within the same size range are obtained.

[0081] S13: grayscale processing is performed on the rheumatism immune joint image data corresponding to the same size range to obtain rheumatism immune joint grayscale image data;

[0082] In an embodiment of the present invention, the previously scaled rheumatoid arthritis joint image data is grayscaled. The purpose of this step is to convert the color image into a grayscale image, remove the color information and retain the brightness information of the image, so as to facilitate subsequent analysis. The grayscale processing usually adopts a weighted average method to convert the color value of each pixel into a single grayscale value according to the weights of the three RGB channels. Specifically, the grayscale value of each pixel is calculated using the following formula: Gray = 0.2989×R+0.5870×G+0.1140×B, wherein R, G, and B are the values ​​of the red, green, and blue channels of the pixel point, respectively, and 0.2989, 0.5870, and 0.1140 are the weight coefficients of each color channel. Through this weighted average method, the color image can be effectively converted into a grayscale image, and the brightness information of the joint structure can be retained, and finally the grayscale image data of the rheumatoid arthritis joint can be obtained.

[0083] S14: Calculate pixel fuzziness of the grayscale image data of the rheumatoid arthritis joint to obtain a pixel fuzziness value of the rheumatoid arthritis joint image;

[0084] In an embodiment of the present invention, the pixel blurriness of the grayscale image data of the rheumatoid arthritis joint is calculated. The purpose of this step is to evaluate the loss of details or blurring caused by joint injury or disease process in the image. In order to calculate the pixel blurriness value, a common image quality assessment algorithm, such as Laplace transform method, edge detection method or image entropy calculation, can be used. Taking Laplace transform as an example, it can effectively capture the changing area in the image by performing second-order derivative processing on the grayscale image. The change in the blurred area will be less, thereby obtaining the blurriness value. The higher the blurriness value, the blurrier the image, corresponding to the severity of the joint injury, and finally the pixel blurriness value of the rheumatoid arthritis joint image is obtained.

[0085] S15: Based on the pixel fuzziness value of the rheumatoid arthritis joint image, the grayscale image data of the rheumatoid arthritis joint is subjected to image noise removal and standardization to obtain standard image data of the rheumatoid arthritis joint.

[0086] In an embodiment of the present invention, noise removal and standardization are performed based on the pixel fuzziness value of the rheumatoid arthritis joint image. First, the pixel fuzziness value is used as a reference for noise removal to perform denoising on the grayscale image. The denoising method can select Gaussian filtering, median filtering or adaptive filtering, etc. In this example, a Gaussian filter can be used to smooth the image to reduce the impact of noise on subsequent analysis. The specific operation is to construct a filter according to the formula of the Gaussian function, and then convolve the image with the filter through a convolution operation to obtain a denoised image. After denoising, the image will be clearer and the noise component will be reduced. Then, standardization is performed. The purpose of standardization is to unify the grayscale value of the image data into a certain range (for example, 0 to 255) for subsequent machine learning analysis or image comparison. The standardization process can be achieved by subtracting the mean and dividing by the standard deviation, so that the processed image data has the same scale, and finally clear and standardized rheumatoid arthritis joint standard image data is obtained.

[0087] Further, as an embodiment of the present invention, refer to Figure 3 As shown, Figure 1 FIG. 1 is a functional flow chart of the image segmentation and damage feature analysis module in FIG. 1 . In this embodiment, the image segmentation and damage feature analysis module includes the following functions:

[0088] S21: Use U-Net neural network to segment the joint images of the standard image data of rheumatism and immune joints, and remove the images of the non-joint tissue areas to obtain the image data of the joint areas of rheumatism and immune diseases, including the images of the joint bones, cartilage, synovium and muscle tissue areas;

[0089] In an embodiment of the present invention, the standard image data of rheumatic immune joints are segmented by using a U-Net neural network. First, standard medical image data of patients with rheumatic immune diseases are collected, usually obtained by magnetic resonance imaging (MRI) or computed tomography (CT). These image data include various tissues around the joints, such as bone, cartilage, synovium, muscle, and vascular tissue. Next, these images are input into a trained U-Net neural network. The U-Net model adopts the architecture of a convolutional neural network (CNN). After multiple layers of convolution and pooling operations, it can accurately identify and segment the joint area in the image. During the training process, U-Net performs supervised learning through image annotation data (such as manual annotation of joint bone, cartilage, and synovium), so that the model can automatically distinguish between joint parts and non-joint tissues in the test data. After the segmentation is completed, the images of the non-joint areas are eliminated to obtain joint area image data containing tissue parts such as joint bone, cartilage, synovium, and muscle, ensuring that only the joint part is subsequently analyzed, and finally obtaining the joint area image data of the rheumatic immune disease.

[0090] S22: Perform three-dimensional reconstruction of the joint structure of the image data of the partial region of the rheumatic immune disease joint to generate a three-dimensional model of the joint structure region of the rheumatic immune disease; obtain the dynamic change law of the joints corresponding to the flexion, extension and rotation movements of the rheumatic immune disease patients, and quantify the joint gap width of the three-dimensional model of the joint structure region of the rheumatic immune disease based on the dynamic change law of the joints corresponding to the flexion, extension and rotation movements of the rheumatic immune disease patients to obtain the joint gap width of the rheumatic immune disease;

[0091] In an embodiment of the present invention, a three-dimensional reconstruction of the joint structure is performed on the segmented joint part image data. First, the original two-dimensional image data is converted into a three-dimensional model through the collected image data by using image reconstruction technology, such as a three-dimensional reconstruction algorithm based on deep learning or a traditional multi-view reconstruction method (for example, voxel reconstruction based on CT data). These three-dimensional reconstruction models can be used to accurately represent the structural morphology of the joint. Next, joint motion simulation is performed based on the dynamic change law of the joint under flexion, extension and rotation movements. Through kinematic analysis, the dynamic changes of the joints of rheumatic immune patients under different movements are simulated. By setting the flexion and extension angles and rotation angles of the joints and combining the image reconstruction results, the changes in the joint gap width during the movement of the joint are evaluated. The quantification of the joint gap width can be achieved by selecting the contact surfaces at both ends of the joint in the three-dimensional model and the shortest distance between the contact surfaces. The changes in the joint gap of patients with rheumatic immune diseases in different postures are obtained, and finally the joint gap width of the rheumatic immune disease is obtained.

[0092] S23: Perform resolution scanning and surface morphology analysis on the articular cartilage surface corresponding to the three-dimensional model of the rheumatic immune disease joint structure area to obtain the surface morphology of the rheumatic immune disease joint, including the roughness, undulation and collagen fiber orientation of the articular cartilage surface;

[0093] In an embodiment of the present invention, the surface area of ​​the articular cartilage is extracted based on the previously obtained three-dimensional joint model, and the surface of the articular cartilage is scanned using high-resolution three-dimensional imaging technology (such as three-dimensional optical scanning or laser scanning) to obtain fine structural data of the articular cartilage. Through fine scanning, detailed information on the surface of the articular cartilage can be obtained, including surface roughness, undulation and the direction of collagen fibers. In order to further analyze these characteristics, a surface morphology analysis algorithm (such as surface analysis based on least squares fitting or a surface reconstruction algorithm based on laser scanning point cloud data) is used to calculate the surface roughness and undulation. These indicators reflect the health status of the articular cartilage. An increase in the roughness or a change in the undulation of the cartilage surface may indicate degeneration or damage of the articular cartilage. On this basis, the direction of the collagen fibers can also be extracted by a surface texture analysis method (such as a texture analysis method based on Gabor wavelet transform), so as to understand the structural integrity of the cartilage and ultimately obtain the surface morphology of the joint of rheumatic immune diseases.

[0094] S24: Statistical analysis of joint bone changes in the three-dimensional model of the joint structure area of ​​rheumatic immune diseases was performed to obtain joint bone changes in rheumatic immune diseases;

[0095] In an embodiment of the present invention, by conducting an in-depth analysis of the bone mass in the joint area, the specific information of the joint bone area is extracted by combining the previously obtained three-dimensional reconstruction model with the actual image data (such as X-ray or CT scan data). The changes in bone mass are usually manifested as changes in bone density, changes in bone surface morphology, and degradation of trabecular structure. In order to quantify bone changes, a bone density analysis method is used, such as Hounsfield unit (HU) analysis based on CT scan data, or the resolution information in the medical image is used to calculate the bone density value of the joint. Furthermore, through statistical analysis of the trabecular structure (for example, using the trabecular index for evaluation), bone loss or other pathological changes can be identified. The statistical analysis results not only help to understand the bone changes in the lesion site, but also ultimately obtain the corresponding situation of bone changes in the joints of rheumatic autoimmune diseases.

[0096] S25: The soft tissue echo and soft tissue texture complexity corresponding to the joint structure of rheumatic immune disease are obtained through the three-dimensional model of the joint structure area of ​​rheumatic immune disease, and the soft tissue abnormality is evaluated and analyzed based on the soft tissue echo and soft tissue texture complexity to obtain the soft tissue abnormality of the joint of rheumatic immune disease.

[0097] In an embodiment of the present invention, the image data of the joint soft tissue is extracted through the three-dimensional model of the joint structure area, so as to obtain the echo data of the soft tissue through ultrasonic imaging technology (such as high-frequency ultrasonic scanning), and the abnormality of the soft tissue is evaluated by using image processing and signal analysis technology according to the intensity, frequency and image texture characteristics of the echo signal. In order to further analyze in depth, combined with the texture complexity characteristics of the soft tissue (such as calculating the gray co-occurrence matrix characteristics of the soft tissue area through a texture analysis algorithm), it can be evaluated whether the soft tissue has abnormal changes, such as symptoms such as edema, inflammation or fibrosis. By comparing and analyzing the echo and texture complexity data of the soft tissue, combined with pathological characteristics, it can be accurately determined whether the joint soft tissue has lesions, such as synovitis or other soft tissue abnormalities related to rheumatic immune diseases. Through these analysis results, it is possible to achieve abnormal evaluation of soft tissues in patients with rheumatic immune diseases, and finally obtain abnormalities in joint soft tissues of rheumatic immune diseases.

[0098] Furthermore, the method of using the U-Net neural network to perform joint image segmentation on the rheumatism immune joint standard image data includes:

[0099] Performing image contrast enhancement processing on the standard image data of rheumatism and immune joints to highlight the contrast difference between the joint part and the non-joint tissue part, and obtaining the rheumatism and immune joint contrast enhanced image data;

[0100] In an embodiment of the present invention, standard image data of rheumatoid arthritis joints are obtained, which are usually X-ray, CT or MRI images. These images are contrast enhanced by image processing tools (such as image enhancement functions in OpenCV or MATLAB) to better distinguish the joint part from the non-joint tissue part. The contrast enhancement can be achieved by histogram equalization or adaptive contrast adjustment method. Specifically, the CLAHE (Contrast Limited Adaptive Histogram Equalization) algorithm can be used. This algorithm avoids the excessive enhancement effect that may be caused by global histogram equalization through local contrast adjustment. After contrast enhancement processing, the image data of the rheumatoid arthritis joints can highlight the details of the joint part (such as bone, cartilage, synovium, etc.) and significantly improve the contrast with the surrounding non-joint tissues, and finally obtain the contrast enhanced image data of the rheumatoid arthritis joints.

[0101] Preferably, the rheumatoid immune joint contrast-enhanced image data is subjected to a fast Fourier frequency domain transform to obtain a frequency domain spectrum of rheumatoid immune joint tissue characteristics corresponding to different frequency components;

[0102] In an embodiment of the present invention, the image data of rheumatoid arthritis contrast enhancement obtained previously is converted from the spatial domain to the frequency domain using the fast Fourier transform (FFT) method. Specifically, each pixel of the image data is converted into a corresponding frequency component using the FFT algorithm. This can be achieved by using the fft2 function in the NumPy library in MATLAB or Python. The transformation can convert local feature information in the image, such as the detail contrast between joint tissue and non-joint tissue, into feature maps at different frequencies. By analyzing the frequency domain map, higher frequency components (usually representing image details and texture information) and lower frequency components (usually representing a wide range of structural features) in the image can be extracted. These frequency domain components reflect the various tissue characteristics of the rheumatoid arthritis joints, and finally the corresponding frequency domain maps of the rheumatoid arthritis joint tissue characteristics at different frequency components are obtained.

[0103] Preferably, the distribution law of morphological characteristics of the corresponding joint tissue at different frequency scales is determined by the frequency domain atlas of rheumatism and immune joint tissue characteristics corresponding to different frequency components, and the image segmentation network architecture is constructed using the U-Net neural network based on the distribution law of morphological characteristics of the joint tissue at different frequency scales, so as to use a 3x3 convolution kernel in the shallow layer of the network front end and adjust it to a 5x5 convolution kernel after 3 layers, the number of layers is 6, and the attention fusion mechanism is introduced as a jump connection mode, so as to obtain the U-Net neural network architecture for rheumatism and immune joint image segmentation;

[0104] In an embodiment of the present invention, the morphological characteristics of joint tissue are analyzed by using the frequency domain spectrum obtained by fast Fourier transform. Different frequency components correspond to the detailed distribution of the image at different scales. By calculating the energy distribution of the frequency domain spectrum, the characteristics of joint tissue at high frequency (such as bone details) and low frequency (such as the overall shape of the joint) can be identified, and by comparing the energy distribution of each frequency band, the distribution law of the morphological characteristics of joint tissue at different frequency scales is determined. For example, the high-frequency part is mainly related to the detailed characteristics of the bone, and the low-frequency part is mainly related to the overall structure of the joint. Based on these characteristic distribution laws, a U-Net neural network architecture is constructed for image segmentation. In the U-Net network design, a 3x3 convolution kernel is used at the front end to extract local features; in the deep part of the network, the convolution kernel size is adjusted to 5x5, which can better capture large-scale structural information. In this process, the attention mechanism is introduced to optimize the jump connection of the network, thereby enhancing the network's attention to important areas (such as joint edges, bones and cartilage), further improving the segmentation accuracy, and finally constructing a U-Net neural network architecture for rheumatoid arthritis joint image segmentation.

[0105] Preferably, the rheumatology and immunity joint image segmentation U-Net neural network architecture is pre-trained using the general image data corresponding to the joint structure in the public large-scale medical image database to learn basic human joint tissue segmentation knowledge and feature extraction mode, and obtain the rheumatology and immunity joint image rough segmentation model;

[0106] In an embodiment of the present invention, the standard joint structure data contained in a public large-scale medical image database (such as OAI, UK Biobank, etc.) is used to pre-train the U-Net neural network for rheumatoid arthritis joint image segmentation. The purpose of the pre-training is to allow the network to acquire general image segmentation capabilities by learning the joint structure features in large-scale data. At this stage, the U-Net model is trained by a back-propagation algorithm, and the weights are adjusted to adapt to the basic segmentation tasks of joint tissues. The training process uses a cross-entropy loss function, and the Adam optimizer is combined in the optimization process to achieve rapid convergence. The main task of the pre-training is to allow the network to acquire basic segmentation knowledge of human joints, including the ability to recognize bones, cartilage, synovium and other parts, and to provide a preliminary coarse segmentation model for subsequent rheumatoid arthritis joint image segmentation, and finally obtain a coarse segmentation model for rheumatoid arthritis joint images.

[0107] Preferably, a small amount of annotated standard rheumatoid arthritis joint image data is used to fine-tune the rheumatoid arthritis joint image coarse segmentation model, and the manual annotation standard generated based on the region growing algorithm is used as supervision information to optimize the network model weights, and the rheumatoid arthritis joint contrast-enhanced image data is finely segmented for the joint part, and the images of non-joint tissue areas are eliminated to obtain partial joint area image data of rheumatoid arthritis diseases, including partial area images of joint bone, cartilage, synovium and muscle tissue.

[0108] In an embodiment of the present invention, after completing pre-training, a coarse segmentation model is fine-tuned using a small amount of annotated rheumatoid articular standard image data, the annotated data is generated manually or by other image segmentation algorithms, and accurate annotation of joint tissue (including joint bone, cartilage, synovium and muscle areas) is provided. The artificial annotation standard generated by the regional growing algorithm is used as supervision information, which is input into the network as the target label to guide the model to perform fine segmentation of the joint tissue. During the fine-tuning process, the cross entropy loss function is continued to be used for training, and the network weights are updated in combination with the small batch gradient descent method (SGD). In this way, the network can be further optimized on the basis of the original coarse segmentation, thereby realizing the refined segmentation of the joint part, eliminating the non-joint tissue area, and finally obtaining the image data of the joint part area of ​​rheumatic immune disease, including the images of the joint bone, cartilage, synovium and muscle tissue part area.

[0109] Furthermore, the quantification of the joint gap width of the three-dimensional model of the joint structure area of ​​the rheumatic immune disease based on the dynamic change law of the joints corresponding to the flexion, extension and rotation movements of the rheumatic immune disease patients includes:

[0110] Based on the dynamic change rules of joints corresponding to flexion, extension and rotation of patients with rheumatic immune diseases, the key nodes of joint movement of the three-dimensional model of the joint structure area of ​​rheumatic immune diseases are determined, and the displacement extreme points and velocity turning points corresponding to the joints of rheumatic immune diseases in the flexion, extension and rotation cycle are obtained;

[0111] In an embodiment of the present invention, three-dimensional structural data of joints of patients with rheumatic immune diseases are obtained by imaging technology (such as CT, MRI, etc.), and are processed and modeled. The specific structure and tissue characteristics of the joints are obtained by segmentation and three-dimensional reconstruction of the imaging data according to the dynamic change law of joint flexion, extension, rotation and other actions. The kinematic analysis technology is combined with the kinematic model of the joint to determine the displacement extreme points and velocity turning points that appear in the joint movement process during the flexion, extension and rotation cycles. First, the kinematic modeling technology is used to generate the motion trajectory of the joint in different motion stages using the anatomical structure and motion pattern of the joint. The key nodes in the joint movement process are identified by time series analysis method, including the most bent position of the joint during flexion and extension (i.e., the flexion extreme point) and the maximum rotation point of the joint during rotation. The speed and acceleration of each part of the joint are calculated, the change law of the speed during the movement is analyzed, and the velocity turning point is further found. By determining these key nodes, the displacement extreme points and velocity turning points corresponding to the joints of rheumatic immune diseases during the joint flexion, extension and rotation cycles are finally obtained.

[0112] Preferably, node trajectory generation processing is performed according to the displacement extreme value points and velocity turning points corresponding to the rheumatism immune disease joints in the joint flexion, extension and rotation cycle, so as to obtain the key node trajectory of the joint motion corresponding to the flexion, extension and rotation action;

[0113] In an embodiment of the present invention, the trajectories of key nodes in the joint motion process are constructed based on the displacement extreme points and velocity turning points previously determined in the joint flexion, extension and rotation cycles. These trajectories not only reflect the position changes of the joints in the flexion, extension and rotation processes, but also describe the motion paths and velocity characteristics of each node. First, by using kinematic simulation software or a custom simulation tool, the joint displacement extreme points and velocity turning points during the flexion, extension and rotation movements are converted into dynamic node sequences. For each key node, its motion trajectory is further refined, and a smooth joint motion trajectory is generated by an interpolation algorithm (such as spline interpolation or B-spline interpolation). Using this trajectory, accurate simulation of joint motion can be achieved, and the time step and position accuracy of the node can be adjusted according to the characteristics of different motion cycles. Secondly, in the time series of joint motion, the change process of the displacement extreme points and velocity turning points is focused on, and their relative positions and mutual relationships in the entire joint motion are analyzed. The spatial path and dynamic evolution of the joint under different motion states can be obtained, and finally the corresponding joint motion key node trajectory under flexion, extension and rotation movements is generated.

[0114] Preferably, the corresponding joint movement key node trajectory under flexion, extension and rotation is used as a guide, and the corresponding three-dimensional model of the joint structure area of ​​rheumatic immune disease is imported into the finite element analysis software to simulate the corresponding force deformation of the joint under flexion, extension and rotation, so as to obtain the dynamic structural deformation model of the joint of rheumatic immune disease;

[0115] In an embodiment of the present invention, the joint structure of patients with rheumatic immune diseases is imported into finite element analysis (FEA) software for further analysis based on the previously obtained joint motion key node trajectory and three-dimensional structure model. Finite element analysis can simulate the force and deformation of joints during flexion, extension, rotation and other movements. Specifically, firstly, the three-dimensional model of the joint of patients with rheumatic immune diseases (the three-dimensional digital model obtained by reconstructing imaging data) is imported into the finite element analysis software, and then the joint motion trajectory is accurately modeled to make it the initial condition for simulating the joint force deformation process, and the physical properties of each joint part (such as elastic modulus, density, yield strength, etc.) are set in the model, and the deformation characteristics of different tissues such as soft tissue, bone and liquid in the joint are considered. Then, the joint is simulated in the software for flexion, extension and rotation, and the stress and strain distribution of each joint part is obtained by analyzing the stress at different time points during the joint motion, and the corresponding dynamic structure deformation model is generated, which further reveals the influence of the disease on the joint function and structure, and finally obtains the dynamic structure deformation model of the joint of rheumatic immune disease.

[0116] Preferably, a dynamic structural deformation model of a joint of a rheumatic immune disease is used to emit virtual light to the joint gap from different viewing angles at each operation moment of the joint, and the coordinates of the boundary points of the joint gap are determined according to the reflection and refraction characteristics corresponding to the virtual light, so as to obtain a dynamic range point set of the boundary of the joint gap of the rheumatic immune disease;

[0117] In an embodiment of the present invention, by utilizing the previously obtained joint dynamic structure deformation model, the actual morphological changes of the joint during movement are simulated, and the changes in the joint gap and the dynamic characteristics of its boundaries are analyzed through virtual ray tracing technology. The specific steps include first setting a virtual light source for the joint gap based on the joint dynamic deformation model. These light rays are emitted from multiple angles to simulate the lighting conditions under different perspectives during joint movement. The emission direction of the light rays is adjusted according to the angle and position of the joint under flexion, extension, rotation and other actions to ensure that all positions of the joint gap are covered, and the boundary of the joint gap is calculated based on the reflection and refraction characteristics of the light when passing through the joint gap. By tracing the path of the light and analyzing the interaction between the light and the joint surface, the coordinates of the boundary points of the gap can be accurately determined, and the dynamic range of the joint gap during different actions is obtained to form a joint gap boundary point set, and finally a joint gap boundary dynamic range point set for rheumatic immune diseases is obtained.

[0118] Preferably, the joint space width is quantified on the dynamic range point set of the joint space boundary of rheumatic immune diseases to obtain the joint space width of rheumatic immune diseases.

[0119] In an embodiment of the present invention, the joint gap width is quantitatively analyzed based on the joint gap boundary point set obtained in the previous step to further evaluate the impact of rheumatic immune diseases on the joint gap. The specific operations include firstly, analyzing the joint gap boundary point set to calculate the maximum width and minimum width of the joint gap at different time steps. In the finite element model, the gap width is calculated using the point coordinates in three-dimensional space, and the distance between the boundary points is accurately measured using geometric methods. For the dynamic changes of the joint gap, multiple time nodes are selected for comparison to obtain the joint gap width change curve under different actions such as flexion, extension, and rotation, and finally the corresponding joint gap width of the rheumatic immune disease is obtained.

[0120] Furthermore, the statistical analysis of joint bone changes in the three-dimensional model of the joint structure area of ​​rheumatic immune diseases includes:

[0121] By selecting the corresponding rheumatoid arthritis joint bone region in the three-dimensional model of the rheumatoid arthritis joint structure region;

[0122] In an embodiment of the present invention, by acquiring joint imaging data of patients with rheumatic immune diseases, high-resolution three-dimensional medical imaging technology, such as CT (computed tomography) or MRI (magnetic resonance imaging) scanning, is usually used to acquire a three-dimensional image of the joint area, and the image processing software is used to convert the image data into a three-dimensional model, and the joint area is extracted using an image segmentation algorithm (such as threshold segmentation, region growing or deep learning image segmentation model). In this process, the bone area of ​​interest is selected according to the characteristics of the rheumatic immune disease. Specifically, the image segmentation algorithm can divide the bone and soft tissue areas according to the density differences in the joint area. On this basis, the bone damage area of ​​a specific rheumatic immune disease patient is marked through the visualization interface of the model to ensure that the specific location and range of joint bone damage are paid attention to. In this process, it is necessary to mark the characteristic areas of rheumatic immune diseases (such as osteoporosis or trabecular bone damage areas) and perform area selection to finally obtain the corresponding rheumatic immune disease joint bone area.

[0123] Preferably, the mineral content of the rheumatic immune disease joint bone area corresponding to the rheumatic immune disease patient is measured to obtain the mineral content corresponding to the rheumatic immune disease joint bone area;

[0124] In an embodiment of the present invention, the mineral content of the joint bone area is determined by further processing of the imaging data. X-ray absorption method (DXA, dual-energy X-ray absorption method) or CT scanning data combined with quantitative analysis technology can be used to obtain the mineral content of the area. The specific implementation method is to obtain the density data of the area of ​​interest extracted from the three-dimensional model, and use the relationship between the CT value and the mineral density to calculate the mineral content of the area. The measurement accuracy of the mineral content depends on the image resolution and sampling accuracy. This step can use a computer algorithm to compare the CT value (Hounsfield unit) of the joint area with the reference data of the known bone mineral density, so as to obtain the quantitative result of the mineral content, and finally obtain the mineral content corresponding to the joint bone area of ​​the rheumatic immune disease.

[0125] Preferably, the trabecular structure of the rheumatic immune disease joint bone region corresponding to the rheumatic immune disease patient is quantified to obtain the trabecular structure parameters corresponding to the rheumatic immune disease joint bone region, including the trabecular volume fraction, trabecular thickness and trabecular spacing;

[0126] In an embodiment of the present invention, a quantitative analysis is performed on the trabecular structure of a three-dimensional model to extract the trabecular area of ​​the rheumatic immune disease joint area through segmentation technology, and a trabecular structure analysis tool (such as trabecular analysis software or a customized image processing algorithm) is used to quantify parameters such as the volume fraction, thickness, and spacing of the trabecular structure. First, the fine structure data of the area is obtained through three-dimensional reconstruction technology. Next, the morphological analysis technology is applied to analyze the microstructure of the trabeculae, including measuring the volume fraction of trabeculae, that is, the proportion of trabeculae in the overall bone tissue; the thickness of the trabeculae, that is, the thickness of a single trabecula; and the trabecular spacing, which refers to the average distance between trabeculae. These structural parameters can effectively reflect the health of the bone. Especially in patients with rheumatic immune diseases, trabecular destruction or sparseness is often found. Finally, the trabecular structure parameters corresponding to the bone area of ​​the rheumatic immune disease joint are obtained.

[0127] Preferably, a bone index database of healthy people is obtained, and based on the bone index database of healthy people, a statistical analysis of joint bone changes is performed on the mineral content and trabecular structure parameters corresponding to the joint bone area of ​​rheumatic immune disease to obtain the joint bone changes of rheumatic immune disease.

[0128] In an embodiment of the present invention, by collecting bone index data of healthy people, the bone index of healthy people can be obtained by bone density measurement, trabecular structure quantification and related clinical data collection. Common bone indexes include bone mineral density, trabecular structure parameters (such as volume fraction, thickness, spacing, etc.) and other imaging characteristics. By establishing a bone database of healthy people, it is compared and analyzed with the bone data of patients with rheumatic immune diseases. For this purpose, statistical analysis software, such as SPSS or R, can be used to perform differential analysis on the bone data of patients with rheumatic immune diseases and the data of healthy people. By comparing the differences in mineral content and trabecular structure parameters between patients and healthy people, a quantitative assessment of joint bone changes is performed. This analysis can reveal the specific effects of rheumatic immune diseases on joint bone, such as the degree of bone loss, the manifestation of trabecular damage, etc., and finally obtain the joint bone changes of rheumatic immune diseases.

[0129] Furthermore, the bone erosion and synovial thickening assessment module includes the following functions:

[0130] Obtaining corresponding rheumatoid arthritis joint trabecular region images and rheumatoid arthritis joint synovial region images through image data of partial regions of rheumatoid arthritis joints;

[0131] In an embodiment of the present invention, by acquiring image data of a partial area of ​​a joint of a rheumatic immune disease, medical imaging scanning equipment (such as CT, MRI or X-ray imaging equipment) is used to collect images of the patient's joint parts to ensure that the image resolution is high enough to clearly display the bone structure and soft tissue of the joint. For the joint area in the image, an image segmentation algorithm is used to extract the trabecular area and the synovial area. This process usually uses deep learning-based image recognition technology, such as convolutional neural network (CNN), to perform semantic segmentation on the image, thereby obtaining an accurate boundary between the trabecular area and the synovial area of ​​the joint. After the segmentation is completed, the extracted regional image is saved as a separate image file, and finally the image of the trabecular area of ​​the rheumatic immune joint and the image of the synovial area of ​​the rheumatic immune joint are obtained.

[0132] Preferably, the trabecular topological feature analysis is performed on the rheumatoid immune joint trabecular region image to analyze the connectivity and branching structure of the trabecular region, and the corresponding number of branch points and backbone length topological features are extracted to obtain a rheumatoid immune trabecular topological feature set; a trabecular three-dimensional spatial geometric model is constructed according to the rheumatoid immune trabecular topological feature set, and the long axis direction of each corresponding trabecular unit in the trabecular three-dimensional spatial geometric model is determined to obtain the long axis direction vector of each rheumatoid immune trabecular;

[0133] In an embodiment of the present invention, by performing trabecular topological feature analysis on the segmented rheumatoid immune joint trabecular region image, by using an image processing algorithm, first edge detection is performed on the trabecular region, the specific position of the trabecular is identified, and the structure of the trabecular is analyzed by a topological analysis method (such as an image-based topological data analysis algorithm), during the analysis, the connectivity of the trabecular is determined by a connected domain analysis algorithm, the branch structure is identified, and the number of branch points is counted, in addition, image registration and curve fitting techniques are used to extract the topological features of the trabecular backbone length, and according to these features, a three-dimensional geometric model of the trabecular is constructed. Then, the long axis direction of each trabecular unit in the three-dimensional model is determined, and principal component analysis (PCA) or other direction-based algorithms are usually used to calculate the long axis direction vector of each trabecular, and further obtain the spatial positioning and direction characteristics of each trabecular, and finally obtain the long axis direction vector of each rheumatoid immune trabecular.

[0134] Preferably, the trabecular orientation of the rheumatoid immune joint trabecular region image is analyzed based on the long axis direction vector of each rheumatoid immune trabecula to obtain the trabecular orientation of the rheumatoid immune joint region; the synovial thickness of the rheumatoid immune joint synovial region image is measured to obtain the synovial thickness of the rheumatoid immune joint region;

[0135] In an embodiment of the present invention, the direction of the trabecular area of ​​the rheumatoid immune joint is analyzed based on the long-axis direction vector of the rheumatoid immune trabeculae, and the macroscopic direction law of the trabeculae is analyzed by applying the spatial analysis method through the trabecular direction vector data in three-dimensional space. This analysis usually relies on the local directional analysis technology in computational geometry. The overall direction of the trabecular area is obtained by statistically analyzing the long-axis direction of each trabecular unit. At the same time, the measurement of the synovial area image is carried out using medical image analysis software. For the synovial area, the edge detection algorithm is used to extract the outline of the synovial area. Combined with the thickness measurement algorithm, the specific thickness value of each synovial area is obtained, and finally the synovial thickness of the rheumatoid immune joint area is obtained.

[0136] Preferably, the corresponding rheumatic immune disease bone erosion extension direction is tracked according to the direction of the trabeculae in the rheumatic immune joint area, and the bone erosion is quantified and determined based on the image data of the rheumatic immune disease joint part area, so as to obtain the size of the rheumatic immune joint bone erosion area;

[0137] In an embodiment of the present invention, the extension direction of bone erosion in rheumatic immune disease is simulated according to the orientation characteristics of trabeculae. First, based on the orientation data of trabeculae, an extension model (such as finite element analysis or simulation algorithm) is used to predict the bone erosion extension path of the disease. This process takes into account factors such as the structure and density of trabeculae, and infers the possible extension direction of erosion. On this basis, the bone erosion in the image data is evaluated through a quantitative model of bone erosion, and the area and depth of the bone erosion area are calculated to determine the bone erosion situation of the rheumatic immune joint. The quantification process may rely on image segmentation and measurement techniques, such as region growing method and morphological analysis method, to calculate the area of ​​the eroded area and compare it with the normal joint area, to obtain a quantitative evaluation result of bone erosion, and finally to obtain the size of the bone erosion area of ​​the rheumatic immune joint.

[0138] Preferably, the corresponding rheumatoid immune joint synovial thickening rate is obtained through the synovial thickness of the rheumatoid immune joint area, and based on the rheumatoid immune joint synovial thickening rate, the synovial thickening degree calculation formula is used to perform thickening evaluation calculation on the image data of part of the joint area of ​​the rheumatoid immune disease to obtain the degree of synovial thickening of the rheumatoid immune joint.

[0139] In an embodiment of the present invention, the rate of synovial thickening of the joint is calculated based on the thickness measurement results of the synovial area. The synovial thickening rate is usually achieved by comparing the image data at different time points. The continuous images are accurately aligned using image registration technology to ensure the consistency of the data measured each time. The thickening rate can be calculated by time series analysis, that is, the speed of synovial thickening is deduced by the thickness change of the synovial area in the continuous image data, thereby obtaining the synovial thickening rate of the rheumatoid immune joint. At the same time, by combining the area of ​​the joint part of the rheumatoid immune disease, the spatial position parameter, the time variable parameter, the synovial thickness of the rheumatoid immune joint area, the synovial thickening rate of the rheumatoid immune joint, the synovial thickening rate influence coefficient, the position attenuation coefficient, the synovial thickening rate time attenuation factor and related parameters, a suitable calculation formula for the degree of synovial thickening of the joint is formed to perform thickening evaluation calculation, so as to further quantify the degree of thickening of the image data and finally obtain the degree of synovial thickening of the rheumatoid immune joint.

[0140] Furthermore, the calculation formula for the degree of synovial thickening is specifically:

[0141]

[0142] Where R is the degree of synovial thickening in rheumatoid arthritis joints, A is the area of ​​the rheumatoid arthritis joint region, r is the spatial position parameter corresponding to the rheumatoid arthritis joint region, t is the time variable parameter, H(r) is the synovial thickness of the rheumatoid arthritis joint region at the regional spatial position r, C(r,t) is the synovial thickening rate of the rheumatoid arthritis joint at the regional spatial position r and time t, α is the synovial thickening rate influence coefficient, γ is the position attenuation coefficient, β is the synovial thickening rate time attenuation factor, and η is the correction coefficient of the degree of synovial thickening in rheumatoid arthritis joints.

[0143] The present invention obtains a calculation formula for the degree of synovial thickening of joints by using a specific mathematical model and after verification, which is used to evaluate and calculate the thickening of image data of some areas of joints of rheumatic immune diseases. The calculation formula for the degree of synovial thickening of joints involves multiple variables (such as position, time, synovial thickness, thickening rate, etc.), which can comprehensively reflect the different factors of synovial thickening, rather than relying solely on a single indicator. This comprehensive consideration of multiple factors makes the calculation result more comprehensive and accurate. The r and t variables in the formula represent the spatial position and time, respectively, and can calculate the synovial thickness and thickening rate at different positions, while considering the changing trend of synovial thickening over time, which makes it possible to accurately evaluate the degree of synovial thickening at different time nodes and joint positions. By introducing the synovial thickening rate of rheumatic immune joints, the formula can describe the changes in the synovial thickening rate in different areas and time points. Compared with the traditional fixed thickening rate assumption, this dynamic modeling is more in line with the trend of actual pathological changes, making the thickening evaluation result closer to the real development process of the disease. By introducing the spatial attenuation factor, the attenuation effect of the synovial thickening rate with the change of spatial position can be taken into account. In some joint areas, the thickening rate decreases with the change of distance due to pathological conditions, structural changes or local physiological differences. This attenuation factor enhances the spatial adaptability of the model, enabling it to reflect more accurate local pathological phenomena. Secondly, by introducing the time attenuation factor of the synovial thickening rate, the formula can simulate the trend of the thickening rate changing over time. With the progression of rheumatic autoimmune diseases, the rate of synovial thickening will change. Therefore, considering this factor helps to capture subtle differences in the evolution of the disease. In addition, the introduction of the correction coefficient can adapt to special situations in the actual data, such as individual differences among patients, measurement errors in imaging data, etc. It can make certain adjustments to the output results of the formula to ensure that the evaluation results are more in line with actual observations and clinical needs. In summary, this formula fully considers the degree of synovial thickening R of rheumatoid arthritis joints, the area A of the rheumatoid arthritis joint region, the spatial position parameter r corresponding to the rheumatoid arthritis joint region, the time variable parameter t, the synovial thickness H(r) of the rheumatoid arthritis joint region at the regional spatial position r, the synovial thickening rate C(r,t) of the rheumatoid arthritis joint at the regional spatial position r and time t, the synovial thickening rate influence coefficient α, the position attenuation coefficient γ, the synovial thickening rate time attenuation factor β, and the correction coefficient η of the degree of synovial thickening of the rheumatoid arthritis joints. According to the correlation between the degree of synovial thickening R of the rheumatoid arthritis joints and the above parameters, a functional relationship is formed. This formula can realize the thickening evaluation calculation process of the imaging data of some areas of the joints of rheumatic immune diseases. At the same time, by introducing the correction coefficient η of the synovial thickening degree of rheumatic immune joints, it can be adjusted according to the errors occurring in the calculation process, thereby improving the accuracy and applicability of the calculation formula for the synovial thickening degree of the joints.

[0144] Furthermore, the joint injury assessment and damage visualization module includes the following functions:

[0145] Based on the size of the bone erosion area of ​​rheumatoid arthritis joints and the degree of synovial thickening of rheumatoid arthritis joints, the joint imaging injury quantitative assessment formula was used to quantitatively assess the structural characteristics of joint injury in rheumatoid arthritis diseases, so as to obtain the corresponding joint imaging injury score for patients with rheumatoid arthritis diseases.

[0146] In an embodiment of the present invention, a suitable joint injury quantitative assessment calculation formula is formed by combining the observation period of patients with rheumatic immune diseases, the patient observation time variable, the size of the rheumatic immune joint bone erosion area, the degree of synovial thickening of the rheumatic immune joints, the bone erosion influence weight coefficient, the joint space width, the joint surface roughness, the joint surface undulation, the degree of joint bone change, the degree of joint soft tissue abnormality, the injury attenuation coefficient and related parameters to perform a quantitative assessment of the degree of injury. The size of the bone erosion area and the degree of synovial thickening are used as input variables and matched with the degree of joint injury (such as bone erosion, narrowing of the joint space, etc.) of joint X-rays, MRI or CT images, so as to obtain the corresponding imaging injury score, and finally obtain the corresponding joint imaging injury score for patients with rheumatic immune diseases.

[0147] Preferably, the image data of the joint region of the rheumatic immune disease are analyzed for the image characteristics of the joint region to obtain the morphological and textural characteristics corresponding to the image of the joint region of the rheumatic immune disease, wherein the morphological and textural characteristics reveal the tissue structure changes corresponding to the damaged joint region, including the imaging manifestations of cartilage damage and osteoporosis;

[0148] In an embodiment of the present invention, feature analysis is performed on images of partial regions of joints of rheumatic immune diseases that have previously undergone image segmentation. The analysis method includes morphological feature extraction and texture feature extraction. Morphological feature extraction mainly uses edge detection, shape analysis and other algorithms (such as Sobel operator, Canny edge detection) to identify the geometric shape and structural changes of the damaged joint region. Texture feature extraction uses algorithms such as gray-level co-occurrence matrix (GLCM) to calculate texture features of the joint region in the image, such as contrast, homogeneity, correlation and other indicators, so as to reflect the degree of cartilage damage and the imaging manifestations of osteoporosis. Through the analysis of these features, it is possible to have an in-depth understanding of the changes in the tissue structure of the damaged joint, such as cartilage wear, bone density changes, etc., and finally obtain the morphological and texture features corresponding to the images of the joint region of the rheumatic immune disease.

[0149] Preferably, the unsupervised clustering K-means method is used to evaluate the joint injury type of the image data of the joint area of ​​the rheumatic immune disease based on the morphological and textural features corresponding to the rheumatic immune disease joint area image, so as to obtain the joint image injury type corresponding to the rheumatic immune disease patient;

[0150] In an embodiment of the present invention, the joint image data of patients with rheumatic immune diseases are classified by using an unsupervised clustering algorithm K-means based on the morphological and textural features extracted in the previous step. The K-means algorithm first performs feature vectorization on all image data to convert the extracted morphological and textural features into vectors in a high-dimensional feature space. Then, a K value (number of clusters) is set, which is optimized by a cross-validation method to ensure that the image data can be reasonably divided into different categories. During the clustering process, each image is assigned to a category, in which the morphological and textural features of similar images are similar and can represent the same type of joint injury, such as mild cartilage injury, severe osteoporosis, etc. The key to this step is the use of unsupervised learning, which can automatically identify the type of joint injury based on the inherent laws in the image data without relying on artificial label data, thereby improving the efficiency and accuracy of the evaluation, and ultimately obtaining the type of joint image injury corresponding to patients with rheumatic immune diseases.

[0151] Preferably, joint damage visualization is performed on the corresponding rheumatic immune disease joint images according to the joint image damage scores and joint image damage types corresponding to the rheumatic immune disease patients, so as to obtain joint damage visualization image results corresponding to the rheumatic immune disease patients.

[0152] In an embodiment of the present invention, visualization processing is performed based on the previously obtained joint injury score and the obtained injury type information in combination with the image data of patients with rheumatic autoimmune diseases. During the visualization process, the joint image data is merged with the injury score and type data using image processing software (such as 3DSlicer or ITK-Snap). Specifically, the software color-codes the damaged area in the image according to the injury score, and the depth of the color is linearly related to the degree of injury. The more seriously damaged area will be marked with a darker color, and the slightly damaged area will be marked with a lighter color. At the same time, based on the information on the injury type, the image data can be further annotated, such as marking cartilage damage areas, osteoporosis areas, etc. In the 3D view, the structural changes and damage conditions of the patient's joints can be clearly presented, and the physician can use the visualized image to make a more intuitive diagnosis and treatment decision support. The realization of visualization greatly improves the doctor's understanding and intervention capabilities of joint injuries in rheumatic autoimmune diseases, and ultimately obtains the corresponding joint damage visualization image results for patients with rheumatic autoimmune diseases.

[0153] Furthermore, the calculation formula for quantitative assessment of joint damage is specifically:

[0154]

[0155] Where J is the joint imaging injury score, τ is the observation period of patients with rheumatic autoimmune diseases, t′ is the patient observation time variable, and C p (t′) is the size of the bone erosion area of ​​the rheumatoid arthritis joint at the observation time t′, R is the degree of synovial thickening of the rheumatoid arthritis joint, θ is the weight coefficient of bone erosion, K is the width of the joint space, D c is the joint surface roughness, Q f is the undulation of the joint surface, B is the degree of change in the joint bone, ε is the degree of abnormality of the joint soft tissue, λ is the injury attenuation coefficient, and ξ is the correction coefficient of the joint imaging injury score.

[0156] The present invention obtains a joint injury quantitative assessment calculation formula by using a specific mathematical model and after verification, which is used for quantitative assessment of joint imaging damage to the structural characteristics of joint damage in rheumatic immune diseases. The joint injury quantitative assessment calculation formula comprehensively considers multiple important factors in joint imaging data, including the bone erosion area of ​​rheumatic immune joints, the degree of synovial thickening, the width of the joint space, the roughness of the joint surface, the degree of soft tissue abnormality, etc., which makes the assessment of joint damage more comprehensive and can reflect the damage status of the joint from multiple dimensions, thereby providing a more accurate basis for the subsequent processing process. By weighted calculation of different factors, the formula can generate a joint imaging damage score, which provides a quantitative standard for clinicians, and is convenient for quantitative comparison of the damage progress of different patients and different time points. This quantitative assessment helps to track the evolution of patients' joint damage and provides a reference for clinical decision-making. By observing the cycle and time variables, the formula can dynamically reflect the time evolution process of joint damage, which makes the assessment not only reflect the current joint damage, but also predict the possible development trend of joint damage in the future, which helps to adjust the decision-making plan in time. Each parameter in the formula is adjusted according to the different conditions of individual patients, so that individualized assessment can be provided for different patients. This personalized assessment method improves the accuracy and pertinence of joint injury assessment and helps to improve the risk assessment effect of patients. This formula is not only based on imaging data, but also combines the specific structural characteristics of rheumatic immune disease joints, such as cartilage damage, osteoporosis, etc., providing a more detailed basis for quantitative assessment, which helps to reveal the specific type of injury and its impact on joint function, and provides more detailed guidance for risk assessment. In addition, the correction coefficient in the formula further enhances the adaptability of the calculation formula, taking into account the influence of different patient groups, disease types, imaging technology differences and other aspects. By adjusting the correction coefficient, the assessment can be made more accurate, avoiding the situation where a single calculation formula cannot adapt to all patients or disease types. This correction coefficient can be adjusted according to the experience accumulated in clinical practice or further research data, making the quantitative assessment results of joint injuries more personalized and accurate. In summary, this formula fully considers the joint imaging injury score J, the observation period τ of patients with rheumatic immune diseases, the patient observation time variable t′, and the size of the rheumatic immune joint bone erosion area C at the observation time t′. p (t'), the degree of synovial thickening of rheumatoid arthritis joints R, the weight coefficient of bone erosion θ, the width of the joint space K, and the roughness of the joint surface D c , joint surface relief Q f , the degree of joint bone change B, the degree of joint soft tissue abnormality ε, the damage attenuation coefficient λ, the correction coefficient ξ of the joint imaging damage score, and a functional relationship is formed according to the mutual correlation between the joint imaging damage score J and the above parameters This formula can realize the quantitative evaluation process of joint imaging damage of the structural characteristics of joint damage in rheumatic autoimmune diseases. At the same time, by introducing the correction coefficient ξ of the joint imaging damage score, it can be adjusted according to the errors occurring in the calculation process, thereby improving the accuracy and applicability of the calculation formula for quantitative evaluation of joint damage.

[0157] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A rheumatic immune disease joint damage assessment system based on image recognition, characterized in that: Includes the following modules: The joint image preprocessing module is used to obtain the joint image data of rheumatism and immune disease, and to perform size scaling and image denoising on the joint image data of rheumatism and immune disease to obtain the standard image data of rheumatism and immune disease joints; The image segmentation and injury feature analysis module is used to perform joint segmentation on the standard image data of rheumatoid arthritis joints using the U-Net neural network to obtain image data of the partial area of ​​the joints of rheumatoid arthritis disease; perform joint injury structural feature analysis on the image data of the partial area of ​​the joints of rheumatoid arthritis disease to obtain the structural features of joint injury of rheumatoid arthritis disease, including the width of the joint space of rheumatoid arthritis disease, the surface morphology of the joints of rheumatoid arthritis disease, the bone changes of the joints of rheumatoid arthritis disease, and the abnormalities of the soft tissue of the joints of rheumatoid arthritis disease; The bone erosion and synovial thickening assessment module is used to obtain the corresponding trabecular orientation and synovial thickness of the rheumatoid articular joint region through the image data of the rheumatoid articular joint region, and to perform bone erosion and thickening assessment and analysis on the image data of the rheumatoid articular joint region based on the trabecular orientation and synovial thickness of the rheumatoid articular joint region, so as to obtain the size of the bone erosion area and the degree of synovial thickening of the rheumatoid articular joint; The joint injury assessment and damage visualization module is used to quantitatively assess the joint imaging damage of rheumatic immune disease joint injury structural characteristics based on the size of the rheumatic immune joint bone erosion area and the degree of synovial thickening of the rheumatic immune joint, so as to obtain the corresponding joint imaging damage score of patients with rheumatic immune diseases; The imaging data of partial areas of rheumatic immune disease joints are evaluated for the type of joint injury to obtain the corresponding joint imaging injury type of the rheumatic immune disease patients; the corresponding rheumatic immune disease joint images are visualized according to the joint imaging injury scores and joint imaging injury types of the rheumatic immune disease patients to obtain the corresponding joint damage visualization image results of the rheumatic immune disease patients.

2. The image recognition-based rheumatic immune disease joint damage assessment system according to claim 1, characterized in that: The joint image preprocessing module includes the following functions: Obtain joint imaging data of rheumatism and immune diseases; Performing size scaling processing on rheumatism-immunity disease joint image data to obtain corresponding rheumatism-immunity joint image data within the same size range; Grayscale processing is performed on the rheumatism immune joint image data corresponding to the same size range to obtain the rheumatism immune joint grayscale image data; Calculate the pixel fuzziness of the grayscale image data of the rheumatoid arthritis joint to obtain the pixel fuzziness value of the rheumatoid arthritis joint image; Based on the pixel fuzziness value of the rheumatoid arthritis joint image, the image noise of the grayscale image data of the rheumatoid arthritis joint is removed and standardized to obtain the standard image data of the rheumatoid arthritis joint.

3. The image recognition-based rheumatic immune disease joint damage assessment system according to claim 1, characterized in that: The image segmentation and damage feature analysis module includes the following functions: The U-Net neural network is used to segment the joint images of the standard image data of rheumatism and immune joints, and the images of the non-joint tissue regions are eliminated to obtain the image data of the joint regions of rheumatism and immune diseases, including the images of the joint bones, cartilage, synovium and muscle tissue regions. Performing three-dimensional reconstruction of the joint structure of the image data of the partial area of ​​the rheumatic immune disease joint to generate a three-dimensional model of the rheumatic immune disease joint structure area; The dynamic change rules of joints corresponding to flexion, extension and rotation of patients with rheumatic immune diseases are obtained, and the joint space width of the three-dimensional model of the joint structure area of ​​rheumatic immune diseases is quantified based on the dynamic change rules of joints corresponding to flexion, extension and rotation of patients with rheumatic immune diseases to obtain the joint space width of rheumatic immune diseases; Perform resolution scanning and surface morphology analysis on the articular cartilage surface corresponding to the three-dimensional model of the joint structure area of ​​rheumatic immune disease to obtain the surface morphology of the rheumatic immune disease joint, including the roughness, undulation and collagen fiber orientation of the articular cartilage surface; Perform statistical analysis on the changes in joint bone mass in the three-dimensional model of the joint structure area of ​​rheumatic immune diseases to obtain the changes in joint bone mass in rheumatic immune diseases; The soft tissue echo and soft tissue texture complexity corresponding to the joint structure of rheumatoid immune disease are obtained through the three-dimensional model of the joint structure area of ​​rheumatoid immune disease, and the soft tissue abnormality is evaluated and analyzed based on the soft tissue echo and soft tissue texture complexity to obtain the soft tissue abnormality of the joint of rheumatoid immune disease.

4. The image recognition-based rheumatic immune disease joint damage assessment system according to claim 3, characterized in that: The method of using the U-Net neural network to perform joint image segmentation on the rheumatism immune joint standard image data includes: Performing image contrast enhancement processing on the standard image data of rheumatism and immune joints to highlight the contrast difference between the joint part and the non-joint tissue part, and obtaining the rheumatism and immune joint contrast enhanced image data; Perform fast Fourier frequency domain transformation on the contrast-enhanced image data of rheumatoid immune joints to obtain the frequency domain maps of rheumatoid immune joint tissue characteristics corresponding to different frequency components; The distribution law of morphological characteristics of the corresponding joint tissue at different frequency scales was determined by the frequency domain atlas of rheumatoid immune joint tissue characteristics corresponding to different frequency components, and the image segmentation network architecture was constructed using the U-Net neural network based on the distribution law of morphological characteristics of joint tissue at different frequency scales. The U-Net neural network architecture for rheumatoid immune joint image segmentation was obtained by using a 3x3 convolution kernel at the shallow layer of the network front end and adjusting it to a 5x5 convolution kernel after 3 layers, with 6 layers and introducing an attention fusion mechanism as a skip connection mode. The rheumatology and immunity joint image segmentation U-Net neural network architecture is pre-trained using the general image data corresponding to the joint structure in the public large-scale medical image database to learn the basic human joint tissue segmentation knowledge and feature extraction mode, and obtain the rheumatology and immunity joint image rough segmentation model; A small amount of annotated standard rheumatoid arthritis joint image data is used to fine-tune the coarse segmentation model of rheumatoid arthritis joint images. The manual annotation standards generated by the region growing algorithm are used as supervision information to optimize the network model weights. The rheumatoid arthritis joint contrast-enhanced image data are then finely segmented into joint parts, and images of non-joint tissue areas are eliminated to obtain image data of partial joint areas of rheumatoid arthritis diseases, including images of joint bone, cartilage, synovium, and partial muscle tissue areas.

5. The image recognition-based rheumatic immune disease joint damage assessment system according to claim 3, characterized in that: The method of quantifying the joint gap width of the three-dimensional model of the joint structure area of ​​rheumatic immune disease based on the dynamic change law of the joints corresponding to the flexion, extension and rotation movements of the rheumatic immune disease patients includes: Based on the dynamic change rules of joints corresponding to flexion, extension and rotation of patients with rheumatic immune diseases, the key nodes of joint movement of the three-dimensional model of the joint structure area of ​​rheumatic immune diseases are determined, and the displacement extreme points and velocity turning points corresponding to the joints of rheumatic immune diseases in the flexion, extension and rotation cycle are obtained; According to the displacement extreme points and velocity turning points corresponding to the rheumatoid arthritis joints in the flexion and extension and rotation cycles, node trajectory generation processing is performed to obtain the key node trajectory of the joint motion corresponding to the flexion and extension and rotation movements; By taking the corresponding joint motion key node trajectory under flexion, extension and rotation as a guide, the corresponding three-dimensional model of the joint structure area of ​​rheumatic immune disease is imported into the finite element analysis software to simulate the corresponding force deformation of the joint under flexion, extension and rotation, so as to obtain the dynamic structural deformation model of the joint of rheumatic immune disease; Through the dynamic structural deformation model of rheumatic immune disease joints, virtual light is emitted to the joint gap from different perspectives at each operation moment of the joint, and the coordinates of the joint gap boundary points are determined according to the reflection and refraction characteristics corresponding to the virtual light, so as to obtain the dynamic range point set of the joint gap boundary of rheumatic immune disease; The joint space width of rheumatic immune disease joint space boundary dynamic range point set is quantified to obtain the joint space width of rheumatic immune disease.

6. The image recognition-based rheumatic immune disease joint damage assessment system according to claim 3, characterized in that: The statistical analysis of joint bone changes on the three-dimensional model of the joint structure area of ​​rheumatic immune diseases includes: By selecting the corresponding rheumatoid arthritis joint bone region in the three-dimensional model of the rheumatoid arthritis joint structure region; Determine the mineral content of the rheumatic immune disease joint bone area corresponding to the rheumatic immune disease patient to obtain the mineral content corresponding to the rheumatic immune disease joint bone area; The trabecular structure of the rheumatic immune disease joint bone area corresponding to the rheumatic immune disease patient is quantified to obtain the trabecular structure parameters corresponding to the rheumatic immune disease joint bone area, including the trabecular volume fraction, trabecular thickness and trabecular spacing; A bone index database of healthy people is obtained, and based on the bone index database of healthy people, a statistical analysis of the mineral content and trabecular structure parameters corresponding to the joint bone area of ​​rheumatic immune disease is performed to obtain the joint bone changes of rheumatic immune disease.

7. The image recognition-based rheumatic immune disease joint damage assessment system according to claim 1, characterized in that: The bone erosion and synovial thickening assessment module includes the following functions: Obtaining corresponding rheumatoid arthritis joint trabecular region images and rheumatoid arthritis joint synovial region images through partial region image data of rheumatoid arthritis joints; The trabecular topological feature analysis is performed on the rheumatoid immune joint trabecular region images to analyze the connectivity and branching structure of the trabecular region, and the corresponding branch point number and backbone length topological features are extracted to obtain the rheumatoid immune trabecular topological feature set; a trabecular three-dimensional spatial geometric model is constructed based on the rheumatoid immune trabecular topological feature set, and the long axis direction of each trabecular unit corresponding to the trabecular three-dimensional spatial geometric model is determined to obtain the long axis direction vector of each rheumatoid immune trabecular; Based on the long axis direction vector of each rheumatoid trabecula, the trabecular orientation of the rheumatoid trabecula region image is analyzed to obtain the trabecular orientation of the rheumatoid trabecula region of the rheumatoid trabecula; the synovial thickness of the rheumatoid trabecula region image is measured to obtain the synovial thickness of the rheumatoid trabecula region of the rheumatoid trabecula; According to the direction of trabeculae in the rheumatic immune joint area, the corresponding direction of bone erosion expansion of the rheumatic immune disease is simulated and tracked, and based on the direction of bone erosion expansion of the rheumatic immune disease, the bone erosion of the imaging data of part of the rheumatic immune disease joint area is quantified to obtain the size of the bone erosion area of ​​the rheumatic immune joint; The corresponding rheumatoid arthritis joint synovial thickness was used to obtain the corresponding rheumatoid arthritis joint synovial thickening rate, and based on the rheumatoid arthritis joint synovial thickening rate, the synovial thickening degree calculation formula was used to perform thickening evaluation on the imaging data of some areas of the joints of rheumatoid arthritis joints to obtain the degree of synovial thickening of the rheumatoid arthritis joints.

8. The image recognition-based rheumatic immune disease joint damage assessment system according to claim 7, characterized in that: The calculation formula for the degree of synovial thickening is specifically: Where R is the degree of synovial thickening in rheumatoid arthritis joints, A is the area of ​​the rheumatoid arthritis joint region, r is the spatial position parameter corresponding to the rheumatoid arthritis joint region, t is the time variable parameter, H(r) is the synovial thickness of the rheumatoid arthritis joint region at the regional spatial position r, C(r,t) is the synovial thickening rate of the rheumatoid arthritis joint at the regional spatial position r and time t, α is the synovial thickening rate influence coefficient, γ is the position attenuation coefficient, β is the synovial thickening rate time attenuation factor, and η is the correction coefficient of the degree of synovial thickening in rheumatoid arthritis joints.

9. The image recognition-based rheumatic immune disease joint damage assessment system according to claim 1, characterized in that: The joint injury assessment and damage visualization module includes the following functions: Based on the size of the bone erosion area of ​​rheumatoid arthritis joints and the degree of synovial thickening of rheumatoid arthritis joints, the joint imaging injury quantitative assessment formula was used to quantitatively assess the structural characteristics of joint injury in rheumatoid arthritis diseases, so as to obtain the corresponding joint imaging injury score for patients with rheumatoid arthritis diseases. The imaging data of some areas of joints with rheumatic autoimmune diseases are analyzed for joint regional imaging features to obtain the morphological and textural features corresponding to the images of joint areas with rheumatic autoimmune diseases. The morphological and textural features reveal the tissue structure changes corresponding to the damaged areas of the joints, including the imaging manifestations of cartilage damage and osteoporosis. Based on the morphological and textural features of the joint area images of rheumatic immune diseases, the unsupervised clustering K-means method is used to evaluate the joint injury type of the image data of the joint area of ​​rheumatic immune diseases, and the corresponding joint image injury type of patients with rheumatic immune diseases is obtained; The corresponding joint damage images of rheumatic immune disease joints are visualized according to the corresponding joint image damage scores and joint image damage types of the rheumatic immune disease patients, so as to obtain the corresponding joint damage visualization image results of the rheumatic immune disease patients.

10. The image recognition-based rheumatic immune disease joint damage assessment system according to claim 9, characterized in that: The quantitative evaluation calculation formula for joint damage is specifically as follows: Where J is the joint imaging injury score, τ is the observation period of patients with rheumatic autoimmune diseases, t′ is the patient observation time variable, and C p (t′) is the size of the bone erosion area of ​​the rheumatoid arthritis joint at the observation time t′, R is the degree of synovial thickening of the rheumatoid arthritis joint, θ is the weight coefficient of bone erosion, K is the width of the joint space, D c is the joint surface roughness, Q f is the undulation of the joint surface, B is the degree of change in the joint bone, ε is the degree of abnormality of the joint soft tissue, λ is the injury attenuation coefficient, and ξ is the correction coefficient of the joint imaging injury score.

Images