A method and system for extracting features from CT image data of gouty arthritis

Abstract

The present invention relates to the field of data analysis technology, and in particular to a method and system for extracting features from CT image data for gouty arthritis. The method comprises the following steps: obtaining a CT image of gouty arthritis; performing joint edge detection on the CT image of gouty arthritis to generate CT image joint edge detection data; performing CT image core area segmentation through the CT image joint edge detection data to generate a joint CT core image; performing low-light area extraction on the joint CT core image to obtain a joint CT image low-light area; performing super-resolution reconstruction on the joint CT image low-light area to generate a joint CT brightness enhancement area; replacing the joint CT brightness enhancement area with the joint CT core image with a low-light area to generate a joint CT core optimized image. The present invention improves the accuracy of CT image feature extraction through deep learning and multimodal feature analysis, as well as reconstruction and replacement of low-light areas.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a method and system for extracting CT image data features for gouty arthritis. Background Art

[0002] In the development of the background technology of CT image data feature extraction methods for gouty arthritis, they initially relied mainly on traditional image processing and feature engineering methods. These methods usually include preprocessing steps based on filters and morphological processing, as well as manually designed feature extractors, such as edge detectors and texture descriptors. However, these methods often rely on the prior knowledge of domain experts and are difficult to capture complex arthritis lesion features. With the rise of deep learning technology, especially convolutional neural networks (CNNs), CT image data feature extraction for gouty arthritis has gradually moved towards automation and higher efficiency. CNNs can extract key information from images by learning features from a large number of labeled data sets, thereby avoiding the limitations of manually designed feature extractors. This method can capture richer and more complex lesion patterns when processing CT image data for gouty arthritis, improving the accuracy and efficiency of diagnosis. However, the current traditional methods are often difficult to accurately extract and reconstruct the weak light areas of CT images, resulting in the loss of important details and information in joint images. At the same time, there is less multimodal feature analysis of abnormal joint areas, and it is impossible to fully evaluate the complexity and diversity of joint lesions, resulting in low accuracy in CT image feature extraction. Summary of the invention

[0003] Based on this, it is necessary to provide a method and system for extracting features from CT image data of gouty arthritis to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for extracting features from CT image data of gouty arthritis is provided, the method comprising the following steps:

[0005] Step S1: acquiring a gouty arthritis CT image; performing joint edge detection on the gouty arthritis CT image to generate CT image joint edge detection data; performing CT image core area segmentation using the CT image joint edge detection data to generate a joint CT core image;

[0006] Step S2: extracting the weak light area of ​​the joint CT core image to obtain the weak light area of ​​the joint CT image; performing super-resolution reconstruction on the weak light area of ​​the joint CT image to generate the joint CT brightness enhancement area; replacing the weak light area of ​​the joint CT brightness enhancement area with the joint CT core image to generate the joint CT core optimized image;

[0007] Step S3: marking the joint structure of the joint CT core optimized image to generate joint structure marking data; performing structure detection on the joint structure marking data to generate joint normal structure detection data and joint abnormal structure detection data; performing abnormal joint structure region segmentation on the joint CT core optimized image according to the joint abnormal structure detection data to generate an abnormal joint structure part segmentation image;

[0008] Step S4: Acquire joint abnormality information data; perform multimodal feature analysis of the joint abnormality region on the abnormal joint structure part segmentation image according to the joint abnormality information data to generate multimodal feature analysis data of the joint abnormality region;

[0009] Step S5: Perform feature visualization on the joint normal structure detection data and the joint abnormal area multimodal feature analysis data to generate a joint abnormal area feature map and a joint normal area feature map; perform structural labeling on the gouty arthritis CT image through the joint abnormal area feature map and the joint normal area feature map to generate a joint CT feature image.

[0010] The present invention collects and prepares CT scan images of gouty arthritis for analysis. Edge detection is performed on the CT images to accurately extract the contour and edge information of the joint. Based on the edge detection results, the core area of ​​the joint is segmented for subsequent analysis and processing. Clear joint structure edge information is provided, which helps to accurately locate and analyze the morphology and position of the joint. The generated core area image can be used as the basis for further analysis and processing. Low-light areas are extracted from the joint CT core image, and these areas contain important lesion information. Super-resolution reconstruction is performed on the low-light area to enhance the brightness and clarity of its image. The brightness enhancement area is synthesized with the core image to generate an optimized joint CT image. The visual quality of the low-light area is improved, and the lesion site of gouty arthritis can be more accurately identified and analyzed. The optimized image is more information-rich and helps further structural labeling and analysis. Structural labeling is performed on the optimized joint CT image to identify the various components of the joint. The labeling data is analyzed to detect normal structures and abnormal structures, such as lesion areas. According to the abnormal structure detection data, the abnormal structure site in the joint is segmented to generate an abnormal joint structure image. Detailed structural markers and abnormal structural information are provided to help identify and understand the specific pathological conditions of gouty arthritis. The generation of abnormal structural images is helpful for quantitative analysis and treatment decisions, and improves the accuracy and efficiency of diagnosis. Based on the joint abnormality information data, the multimodal features of the abnormal area, such as shape, density, grayscale, etc., are analyzed. The feature data of normal structures and abnormal areas are visualized to generate feature maps so as to intuitively understand and evaluate the characteristics of the lesions. Intuitive and quantitative abnormal area feature analysis results are provided to help make more accurate decisions during diagnosis and treatment. Therefore, the present invention improves the accuracy of CT image feature extraction through deep learning and multimodal feature analysis, as well as reconstruction and replacement of low-light areas.

[0011] Preferably, step S1 comprises the following steps:

[0012] Step S11: obtaining CT images of gouty arthritis;

[0013] Step S12: performing image preprocessing on the gouty arthritis CT image to generate a standard CT image, wherein the image preprocessing includes image filtering, image denoising and contrast enhancement;

[0014] Step S13: Perform joint edge detection on the standard CT image to generate CT image joint edge detection data;

[0015] Step S14: Segment the core area of ​​the standard CT image using the CT image joint edge detection data to generate a joint CT core image.

[0016] The present invention obtains the joint CT image of the patient from the CT scanning device. A high-quality original CT image is obtained to provide basic data for subsequent processing. A filtering algorithm is applied to remove high-frequency noise in the image, a denoising technique is used to reduce random noise in the CT image, and the image contrast is adjusted to improve the visibility of the joint structure in the image. Through filtering, denoising and contrast enhancement, the clarity and contrast of the image are improved, so that the joint structure in the image is clearer and convenient for subsequent analysis. An edge detection algorithm (such as Canny edge detection) is applied to identify the edges of the joints in the standard CT image. The edges of the joints are accurately identified, so that the outline of the joint structure is more obvious, which is convenient for further image processing and analysis. The core area of ​​the joint in the standard CT image is segmented out using the edge detection data, and the core area of ​​the joint is accurately segmented. The core area segmentation is performed using the edge detection data to ensure the accuracy of the core image of the joint.

[0017] Preferably, step S2 comprises the following steps:

[0018] Step S21: performing brightness detection on the joint CT core image to generate pixel brightness data of the joint CT core image;

[0019] Step S22: extracting the weak light area of ​​the joint CT core image based on the pixel brightness data of the joint CT core image to obtain the weak light area of ​​the joint CT image;

[0020] Step S23: super-resolution reconstruction is performed on the weak-light area of ​​the joint CT image to generate a joint CT brightness enhancement area;

[0021] Step S24: Replace the joint CT brightness enhancement area and the joint CT core image with the weak light area to generate a joint CT core optimized image.

[0022] The present invention performs brightness detection on each pixel in the joint CT core image and records the brightness value of each pixel. The brightness distribution data of the joint CT core image is obtained to provide basic data for subsequent weak light area extraction and processing. The pixel brightness data is analyzed to identify and extract weak light areas with low brightness. The weak light areas in the joint CT core image are accurately located and extracted so that subsequent processing can be concentrated in these areas, thereby improving processing efficiency and effect. The super-resolution reconstruction algorithm is applied to process the extracted weak light areas to improve the resolution and brightness of these areas. Through super-resolution reconstruction, the image details and brightness of the weak light areas are enhanced to improve the image quality of these areas. The brightness enhancement areas reconstructed by super-resolution are replaced back to the corresponding weak light areas in the original joint CT core image. Through replacement processing, a joint CT core optimized image with uniform brightness and rich image details is generated to improve the overall image quality. Through brightness detection, detailed brightness data is obtained to provide a basis for subsequent processing. The weak light areas in the joint CT core image are accurately extracted to ensure the pertinence and effectiveness of the processing. Through super-resolution reconstruction, the image quality of the weak light areas is significantly improved, making the details of these areas clearer.

[0023] Preferably, step S23 includes the following steps:

[0024] Step S231: performing low-light region threshold segmentation on the low-light region of the joint CT image to generate low-light region mask data;

[0025] Step S232: performing multi-view image acquisition on the weak-light region of the joint CT image based on the weak-light region mask data to obtain a multi-view CT image of the weak-light region;

[0026] Step S233: spatially registering the multi-view CT images in the weak-light region to generate a multi-view CT registered image in the weak-light region; and reconstructing a super-resolution regional image on the multi-view CT registered image in the weak-light region to generate a super-resolution reconstructed image.

[0027] Step S234: pyramid decompose the super-resolution reconstructed image and the weak-light area of ​​the joint CT image to generate a reconstructed image decomposition image and a joint CT image weak-light area decomposition image; fuse the reconstructed image decomposition image and the joint CT image weak-light area decomposition image to generate a joint CT brightness enhancement area.

[0028] The present invention segments the weak light area in the joint CT image by using a threshold segmentation algorithm to generate corresponding mask data. The weak light area in the joint CT image is accurately identified and marked, and the generated mask data provides a clear reference for subsequent processing. The marked weak light area is CT scanned at different angles and viewing angles to obtain multi-view CT image data. Through multi-view acquisition, image information from more angles is obtained, image details and depth are increased, and the accuracy of image processing is improved. CT images of different viewing angles are spatially registered so that they are aligned in the same coordinate system. Super-resolution reconstruction is performed on the registered multi-view CT images to improve the resolution and details of the images. Through spatial registration, the consistency of images from different viewing angles is ensured; through super-resolution reconstruction, the clarity and details of the images are significantly improved, laying a solid foundation for further processing. The super-resolution reconstructed image and the original weak light area image are pyramid decomposed to generate decomposed images of different levels. The decomposed images of different levels are fused to synthesize the final brightness enhancement area. Through pyramid decomposition and image fusion, the advantages of reconstructed images and original images are effectively combined to generate joint CT brightness enhancement areas with rich details and uniform brightness, thereby improving the overall image quality.

[0029] Preferably, step S3 comprises the following steps:

[0030] Step S31: Calculate the image gray value of the joint CT core optimized image to obtain image gray value data;

[0031] Step S32: performing image texture feature analysis on the joint CT core optimized image to generate image texture feature data;

[0032] Step S33: marking the joint structure of the joint CT core optimized image by using the image gray value data and the image texture feature data to generate joint structure marking data;

[0033] Step S34: perform structural detection on the joint structure marking data to generate normal joint structure detection data and abnormal joint structure detection data; perform abnormal joint structure area segmentation on the joint CT core optimized image according to the abnormal joint structure detection data to generate an abnormal joint structure part segmentation image.

[0034] The present invention calculates the gray value of each pixel in the joint CT core optimized image and records the gray value of each pixel. The gray value data of each pixel in the image is obtained to provide a basis for subsequent image texture feature analysis and joint structure marking. The texture features of the joint CT core optimized image are extracted by texture analysis algorithms (such as grayscale co-occurrence matrix, local binary pattern, etc.). The texture feature data of the image is generated to reflect the texture details and features of different areas in the image, and provide more information for joint structure marking. Combined with the image gray value data and texture feature data, the joint structure in the joint CT core optimized image is marked. The joint structure is accurately marked and the joint structure marking data is generated, which provides a basis for subsequent structure detection and abnormal area segmentation. The joint structure marking data is analyzed to detect the normal structure and abnormal structure of the joint. According to the test results, the joint CT core optimized image is segmented for abnormal joint structure area to generate a segmented image. The abnormal structure area in the joint is identified and segmented to provide an intuitive reference image for clinical diagnosis.

[0035] Preferably, marking the joint structure of the joint CT core optimized image by using the image gray value data and the image texture feature data includes:

[0036] Performing grayscale weight division on the image grayscale value data based on a preset grayscale threshold range to generate grayscale weight division data, wherein the grayscale weight division data includes a first grayscale weight, a second grayscale weight, and a third grayscale weight;

[0037] Performing texture morphology analysis on the image texture feature data to generate texture morphology analysis data; performing texture regularity analysis on the image texture feature data to generate texture regularity analysis data; performing image texture density division according to the texture morphology analysis data and the texture regularity analysis data to generate image texture density division data, wherein the image texture density data includes texture dense data, texture uniform data and texture sparse data;

[0038] The joint structure is marked on the joint CT core optimized image through grayscale weight division data and image texture density division data to generate joint structure marking data.

[0039] The present invention divides the image gray value data into different gray weight regions by setting the gray threshold range, and generates gray weight division data. The gray weight division data includes a first gray weight, a second gray weight and a third gray weight. Through gray weight division, the regions of different gray levels in the image are accurately identified, and more detailed gray information is provided. Texture morphological analysis algorithms (such as local binary patterns, texture directionality analysis, etc.) are applied to analyze the image texture feature data to generate texture morphological analysis data. The texture morphological features of different regions in the image are obtained to help identify different structures and tissues. Texture regularity analysis methods (such as gray co-occurrence matrix, texture periodicity analysis, etc.) are used to analyze the image texture feature data to generate texture regularity analysis data, which can understand the regularity and consistency of the texture in the image and help distinguish different types of tissues and structures. Combined with the texture morphological analysis data and the texture regularity analysis data, the texture in the image is density-divided to generate texture dense data, texture uniform data and texture sparse data. Through texture density division, the texture features in the image are more accurately described, and richer texture information is provided. Combine the grayscale weight division data and texture density division data to mark the joint structure. The generated joint structure marking data can accurately identify and distinguish the joint structure in the image. By comprehensively utilizing grayscale and texture feature data, the joint structure can be accurately marked, improving the accuracy and reliability of the marking.

[0040] Preferably, step S4 comprises the following steps:

[0041] Step S41: Acquire joint abnormality information data based on the medical information database;

[0042] Step S42: performing abnormal feature association on the abnormal joint structure segmentation image according to the joint abnormality information data to generate joint abnormality structure feature association data;

[0043] Step S43: constructing a model based on the joint abnormality information data to obtain an abnormal joint structure area detection model;

[0044] Step S44: Performing joint abnormality region multimodal feature analysis on the joint abnormality region feature association data through the abnormal joint structure region detection model to generate joint abnormality region multimodal feature analysis data.

[0045] The present invention extracts information data related to joint abnormality from a medical information database, including case records, imaging data, diagnosis results, etc. Utilize rich medical data to obtain detailed information on joint abnormality, providing a basis for subsequent feature association and model construction. Associate the joint abnormality information obtained from the medical information database with the segmented image of the abnormal joint structure part, extract abnormal features and generate joint abnormal structure feature association data. Through feature association, identify and extract detailed feature information of the abnormal joint structure to form high-quality association data. Utilize joint abnormality information data and feature association data to construct an abnormal joint structure region detection model. The model can be trained and optimized using machine learning or deep learning algorithms. By constructing a detection model, automatic recognition and detection of abnormal joint structure regions are achieved, improving the accuracy and efficiency of detection. Apply the detection model, perform multimodal feature analysis on the joint abnormal structure feature association data, extract feature information under different modes, and generate multimodal feature analysis data of joint abnormal regions. Through multimodal feature analysis, the feature information of the abnormal joint region is fully understood, providing a richer basis for diagnosis.

[0046] Preferably, step S43 includes the following steps:

[0047] Step S431: extracting joint abnormality features from the joint abnormality information data to obtain joint abnormality information feature data;

[0048] Step S432: dividing the joint abnormality information feature data into a data set to obtain a model training set and a model test set;

[0049] Step S433: performing model training on the model training set by using a convolutional neural network algorithm to generate an abnormal joint structure area training model;

[0050] Step S434: Use the model test set to perform model optimization iteration on the abnormal joint structure area training model, thereby generating an abnormal joint structure area training model.

[0051] The present invention extracts joint abnormality information from a medical information database, and obtains characteristic data of the joint abnormality information through a feature extraction algorithm (such as feature engineering, statistical analysis, etc.). The key features of the joint abnormality information are obtained to provide a data basis and feature selection basis for building a model. The characteristic data of the joint abnormality information is divided into a model training set and a model test set, usually using a cross-validation or random partitioning method. Ensure the independence and reliability of model training and evaluation, and verify the generalization ability of the model. Use a convolutional neural network (CNN) or other suitable deep learning algorithm to train the model training set to generate a training model specifically for abnormal joint structure area detection. Utilize the powerful learning ability of deep learning to improve the automatic detection and recognition capabilities of abnormal joint structures. Evaluate and optimize the training model through the model test set to ensure that the performance and generalization ability of the model meet the expected requirements. Optimize the parameters and structure of the model to improve the accuracy and stability of abnormal joint structure area detection.

[0052] Preferably, step S5 comprises the following steps:

[0053] Step S51: Performing feature visualization on the joint normal structure detection data and the joint abnormal region multimodal feature analysis data, thereby generating a joint abnormal region feature map and a joint normal region feature map;

[0054] Step S52: Structural marking is performed on the gouty arthritis CT image using the joint abnormal region feature map and the joint normal region feature map to generate a joint CT feature image.

[0055] The present invention generates a joint abnormal region feature map and a joint normal region feature map by visualizing the joint normal structure detection data and the joint abnormal region multimodal feature analysis data, and involves the use of image processing and data visualization technology, such as heat map, feature mapping, etc., to intuitively display the differences and characteristics of normal and abnormal structures in the joint through visualization. The generated joint abnormal region feature map and joint normal region feature map are used to structurally mark the CT image of gouty arthritis. Structural marking includes highlighting normal and abnormal areas by color, marking, contour, etc. By structurally marking the CT image of gouty arthritis, the normal and abnormal areas are highlighted, providing more informative and intuitive image results.

[0056] In this specification, a CT image data feature extraction system for gouty arthritis is provided, which is used to execute the above-mentioned CT image data feature extraction method for gouty arthritis. The CT image data feature extraction system for gouty arthritis includes:

[0057] The region segmentation module is used to obtain CT images of gouty arthritis; perform joint edge detection on the CT images of gouty arthritis to generate CT image joint edge detection data; perform CT image core region segmentation based on the CT image joint edge detection data to generate joint CT core images;

[0058] The low-light reconstruction module is used to extract the low-light area of ​​the joint CT core image to obtain the low-light area of ​​the joint CT image; perform super-resolution reconstruction on the low-light area of ​​the joint CT image to generate the joint CT brightness enhancement area; replace the low-light area of ​​the joint CT brightness enhancement area with the joint CT core image to generate the joint CT core optimized image;

[0059] The structure segmentation module is used to mark the joint structure of the joint CT core optimized image and generate joint structure marking data; perform structure detection on the joint structure marking data to generate joint normal structure detection data and joint abnormal structure detection data; perform abnormal joint structure region segmentation on the joint CT core optimized image according to the joint abnormal structure detection data to generate abnormal joint structure part segmentation image;

[0060] The abnormal feature analysis module is used to obtain joint abnormality information data; perform multimodal feature analysis of the joint abnormality region on the segmented image of the abnormal joint structure part according to the joint abnormality information data, and generate multimodal feature analysis data of the joint abnormality region;

[0061] The feature labeling module is used to perform feature visualization on the normal joint structure detection data and the multimodal feature analysis data of the abnormal joint area, so as to generate a feature map of the abnormal joint area and a feature map of the normal joint area; the gouty arthritis CT image is structurally labeled through the feature map of the abnormal joint area and the feature map of the normal joint area to generate a joint CT feature image.

[0062] The beneficial effect of the present invention is that by performing joint edge detection and core area segmentation on gouty arthritis CT images, the morphological features of the joint area can be extracted, which helps to better understand and analyze the location and range of the lesion. There are some weak light areas in the joint CT images, and the information of these areas is not clear enough. By extracting the weak light areas and performing super-resolution reconstruction, the brightness and details of these areas can be enhanced, and the visualization effect of the image can be improved. Structural marking of the joint CT core optimized image can mark different tissue structures of the joint, such as bones, soft tissues, etc., to provide a basis for subsequent analysis and diagnosis. Through structural detection, the normal structure and abnormal structure of the joint can be further identified and segmented, which is helpful to find lesions or abnormal conditions. According to the joint abnormal information data, the abnormal joint structure part segmentation image is segmented into joint abnormal area, which can accurately locate and extract the abnormal part of the joint. Through multimodal feature analysis, these abnormal areas can be more deeply extracted and analyzed, which helps to understand the nature and extent of the lesion. The joint normal structure detection data and the joint abnormal area multimodal feature analysis data are visualized to generate joint abnormal area feature map and joint normal area feature map. By using these feature maps to structurally mark CT images of gouty arthritis, joint CT feature images can be generated, providing more intuitive and comprehensive information for reference. Therefore, the present invention improves the accuracy of CT image feature extraction through deep learning and multimodal feature analysis, as well as reconstruction and replacement of weak light areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 A schematic diagram of the steps of a method for extracting features from CT image data of gouty arthritis;

[0064] Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart;

[0065] Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG.

[0066] Figure 4 for Figure 1 Detailed implementation steps of step S4 in FIG.

[0067] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0068] The technical method 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.

[0069] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0070] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0071] To achieve this, please refer to Figures 1 to 4 , a method for extracting features from CT image data of gouty arthritis, the method comprising the following steps:

[0072] Step S1: acquiring a gouty arthritis CT image; performing joint edge detection on the gouty arthritis CT image to generate CT image joint edge detection data; performing CT image core area segmentation using the CT image joint edge detection data to generate a joint CT core image;

[0073] Step S2: extracting the weak light area of ​​the joint CT core image to obtain the weak light area of ​​the joint CT image; performing super-resolution reconstruction on the weak light area of ​​the joint CT image to generate the joint CT brightness enhancement area; replacing the weak light area of ​​the joint CT brightness enhancement area with the joint CT core image to generate the joint CT core optimized image;

[0074] Step S3: marking the joint structure of the joint CT core optimized image to generate joint structure marking data; performing structure detection on the joint structure marking data to generate joint normal structure detection data and joint abnormal structure detection data; performing abnormal joint structure region segmentation on the joint CT core optimized image according to the joint abnormal structure detection data to generate an abnormal joint structure part segmentation image;

[0075] Step S4: Acquire joint abnormality information data; perform multimodal feature analysis of the joint abnormality region on the abnormal joint structure part segmentation image according to the joint abnormality information data to generate multimodal feature analysis data of the joint abnormality region;

[0076] Step S5: Perform feature visualization on the joint normal structure detection data and the joint abnormal area multimodal feature analysis data to generate a joint abnormal area feature map and a joint normal area feature map; perform structural labeling on the gouty arthritis CT image through the joint abnormal area feature map and the joint normal area feature map to generate a joint CT feature image.

[0077] The present invention collects and prepares CT scan images of gouty arthritis for analysis. Edge detection is performed on the CT images to accurately extract the contour and edge information of the joint. Based on the edge detection results, the core area of ​​the joint is segmented for subsequent analysis and processing. Clear joint structure edge information is provided, which helps to accurately locate and analyze the morphology and position of the joint. The generated core area image can be used as the basis for further analysis and processing. Low-light areas are extracted from the joint CT core image, and these areas contain important lesion information. Super-resolution reconstruction is performed on the low-light area to enhance the brightness and clarity of its image. The brightness enhancement area is synthesized with the core image to generate an optimized joint CT image. The visual quality of the low-light area is improved, and the lesion site of gouty arthritis can be more accurately identified and analyzed. The optimized image is more information-rich and helps further structural labeling and analysis. Structural labeling is performed on the optimized joint CT image to identify the various components of the joint. The labeling data is analyzed to detect normal structures and abnormal structures, such as lesion areas. According to the abnormal structure detection data, the abnormal structure site in the joint is segmented to generate an abnormal joint structure image. Detailed structural markers and abnormal structural information are provided to help identify and understand the specific pathological conditions of gouty arthritis. The generation of abnormal structural images is helpful for quantitative analysis and treatment decisions, and improves the accuracy and efficiency of diagnosis. Based on the joint abnormality information data, the multimodal features of the abnormal area, such as shape, density, grayscale, etc., are analyzed. The feature data of normal structures and abnormal areas are visualized to generate feature maps so as to intuitively understand and evaluate the characteristics of the lesions. Intuitive and quantitative abnormal area feature analysis results are provided to help make more accurate decisions during diagnosis and treatment. Therefore, the present invention improves the accuracy of CT image feature extraction through deep learning and multimodal feature analysis, as well as reconstruction and replacement of low-light areas.

[0078] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of a method for extracting features from CT image data for gouty arthritis according to the present invention. In this example, the method for extracting features from CT image data for gouty arthritis comprises the following steps:

[0079] Step S1: acquiring a gouty arthritis CT image; performing joint edge detection on the gouty arthritis CT image to generate CT image joint edge detection data; performing CT image core area segmentation using the CT image joint edge detection data to generate a joint CT core image;

[0080] In an embodiment of the present invention, a CT image of a patient with gouty arthritis is obtained by using a medical CT scanning device. The acquired CT image data is stored in a secure data storage system to ensure the integrity and confidentiality of the data. The Canny algorithm is used to detect the edges of the CT image to obtain clear joint edge information. The Canny detection results are edge enhanced to improve the definition and accuracy of the joint edges. The CT image is preprocessed, including denoising, graying, etc., to optimize the subsequent edge detection effect. The Canny edge detection algorithm is applied to the preprocessed image to extract the position and shape information of the joint edge. The core area of ​​the joint is segmented based on the joint edge detection results using a regional growing algorithm. The seed point selection and growth conditions during the growth process are optimized to ensure the accuracy and integrity of the core area segmentation. The core area inside the joint is marked and segmented, including the structure and tissue inside the joint. The joint CT core image is extracted and saved to ensure that the extracted data contains important structural information of the joint and the characteristic area of ​​gouty arthritis.

[0081] Step S2: extracting the weak light area of ​​the joint CT core image to obtain the weak light area of ​​the joint CT image; performing super-resolution reconstruction on the weak light area of ​​the joint CT image to generate the joint CT brightness enhancement area; replacing the weak light area of ​​the joint CT brightness enhancement area with the joint CT core image to generate the joint CT core optimized image;

[0082] In an embodiment of the present invention, a threshold is set based on a grayscale value to extract a weak light area in a joint CT core image. Local contrast enhancement is performed on the extracted weak light area to enhance the detail information of the weak light area. The joint CT core image is preprocessed, including denoising, graying, etc., to optimize the extraction effect of the weak light area. Grayscale threshold segmentation or other appropriate image processing methods are applied to extract the weak light area in the joint CT image. A CNN model in deep learning, such as SRCNN (Super-Resolution Convolutional Neural Network), is used to perform super-resolution reconstruction of the weak light area. The extracted weak light area of ​​the joint CT image is input into the super-resolution reconstruction model to generate a high-resolution brightness enhancement area. The reconstruction result is quality evaluated and corrected to ensure that the generated brightness enhancement area has high quality and clarity. The reconstructed joint CT brightness enhancement area is synthesized with the original joint CT core image to replace the weak light area in the original image. It is ensured that the replaced joint CT core image is significantly improved in brightness and clarity, thereby improving the diagnosis and analysis effect of the image.

[0083] Step S3: marking the joint structure of the joint CT core optimized image to generate joint structure marking data; performing structure detection on the joint structure marking data to generate joint normal structure detection data and joint abnormal structure detection data; performing abnormal joint structure region segmentation on the joint CT core optimized image according to the joint abnormal structure detection data to generate an abnormal joint structure part segmentation image;

[0084] In an embodiment of the present invention, feature points or specific structures such as bones, soft tissues, etc. in the joint image are extracted and feature point matching is performed. During the marking process, the structures in the joint CT core optimized image are marked and segmented to distinguish different joint structures. The generated joint structure marking data is optimized and corrected to ensure the accuracy and completeness of the marking. The joint structure marking data is subjected to structural detection using feature detection algorithms such as edge detection and corner point detection. The structural detection results are divided into two categories: normal structure and abnormal structure, and joint normal structure detection data and joint abnormal structure detection data are generated. Based on the method of regional growing, the abnormal structure region detected in the joint CT core optimized image is segmented. By adjusting the parameters and conditions of the growth algorithm, the accuracy and completeness of the abnormal structure part segmentation image are improved. The abnormal parts identified in the abnormal structure detection data are subjected to detailed segmentation processing to generate an abnormal joint structure part segmentation image. The generated abnormal joint structure part segmentation image is integrated with the joint CT core optimized image to form a complete abnormal structure part display.

[0085] Step S4: Acquire joint abnormality information data; perform multimodal feature analysis of the joint abnormality region on the abnormal joint structure part segmentation image according to the joint abnormality information data to generate multimodal feature analysis data of the joint abnormality region;

[0086] In an embodiment of the present invention, joint abnormality information data, including arthritis type, location, severity, etc., is obtained from a public medical database or an internal hospital system. Joint abnormality information data can be obtained through manual annotation by experts or a medical image reporting system. Use a deep learning model such as a convolutional neural network (CNN) or a traditional image feature extraction method to extract the visual features of the abnormal joint structure segmentation image. The features extracted from different modalities (such as CT images, MRI images, etc.) are fused to form multimodal feature data. The multiple visual features extracted from the abnormal joint structure segmentation image are fused to enhance the richness and expression ability of the features. The fused multimodal feature data is statistically analyzed, including indicators such as mean, variance, and correlation. According to the joint abnormality information data, the abnormal areas identified in the abnormal joint structure segmentation image are refined. The multimodal features of each abnormal area are extracted, and these features are analyzed and compared in detail.

[0087] Step S5: Perform feature visualization on the joint normal structure detection data and the joint abnormal area multimodal feature analysis data to generate a joint abnormal area feature map and a joint normal area feature map; perform structural labeling on the gouty arthritis CT image through the joint abnormal area feature map and the joint normal area feature map to generate a joint CT feature image.

[0088] In an embodiment of the present invention, the spatial distribution and intensity of the multimodal feature analysis data of the joint abnormal area are displayed by using a heat map. The features extracted from the different modal data are mapped to the joint structure to show the distribution of different features in the joint CT image. The joint normal structure detection data and the multimodal feature analysis data of the joint abnormal area are merged and sorted for visualization. Image processing and visualization tools such as Matplotlib, Seaborn, etc. are used to generate the joint abnormal area feature map and the joint normal area feature map. Automated structural labeling is performed on the joint abnormal area feature map and the joint normal area feature map using machine learning or deep learning technology. On the basis of automated processing, further verification and correction are performed through manual labeling by medical experts. The generated joint abnormal area feature map and the joint normal area feature map are superimposed and combined with the original gouty arthritis CT image to generate a complete joint CT feature image. The generated joint CT feature image is analyzed and compared to evaluate the accuracy and completeness of the feature labeling.

[0089] Preferably, step S1 comprises the following steps:

[0090] Step S11: obtaining CT images of gouty arthritis;

[0091] Step S12: performing image preprocessing on the gouty arthritis CT image to generate a standard CT image, wherein the image preprocessing includes image filtering, image denoising and contrast enhancement;

[0092] Step S13: Perform joint edge detection on the standard CT image to generate CT image joint edge detection data;

[0093] Step S14: Segment the core area of ​​the standard CT image using the CT image joint edge detection data to generate a joint CT core image.

[0094] In an embodiment of the present invention, CT image data of patients with gouty arthritis are obtained from a medical institution or a data provider. Ensure that the image data contains sufficient resolution and quality to support subsequent processing and analysis steps. Use appropriate filtering techniques (such as Gaussian filtering, median filtering, etc.) to remove noise and smooth the image to improve the stability and accuracy of subsequent processing steps. Apply denoising algorithms (such as wavelet denoising, neighborhood-based denoising methods, etc.) to further reduce noise in the image and ensure image clarity and information retention. Use contrast enhancement techniques (such as histogram equalization, adaptive contrast enhancement, etc.) to enhance local contrast in the image and highlight details and features in the image. Select an edge detection algorithm suitable for medical image processing, such as Canny edge detection, Sobel operator, Laplace operator, etc. Consider the noise level and complexity of the image, and select an algorithm that can accurately extract joint edges. Apply the selected edge detection algorithm to the preprocessed standard CT image to extract edge information of the joint part. Use the joint edge detection data obtained in the above steps, combined with an appropriate segmentation algorithm (such as threshold segmentation, region growing method, etc.), to segment the joint area from the overall image. Ensure that the segmentation results can accurately reflect the anatomical structure of the joint and provide a clear regional image for subsequent analysis and diagnosis.

[0095] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:

[0096] Step S21: performing brightness detection on the joint CT core image to generate pixel brightness data of the joint CT core image;

[0097] Step S22: extracting the weak light area of ​​the joint CT core image based on the pixel brightness data of the joint CT core image to obtain the weak light area of ​​the joint CT image;

[0098] Step S23: super-resolution reconstruction is performed on the weak-light area of ​​the joint CT image to generate a joint CT brightness enhancement area;

[0099] Step S24: Replace the joint CT brightness enhancement area and the joint CT core image with the weak light area to generate a joint CT core optimized image.

[0100] In an embodiment of the present invention, by traversing each pixel of the joint CT core image, its brightness value or grayscale value is obtained. The average, median or other statistics of the pixel value can be used to represent the overall brightness situation. Based on the pixel brightness data, a threshold is set or an adaptive algorithm is used to determine the pixel set of the low-light area. Algorithms such as threshold segmentation and regional growth can be used to mark the area where the pixel brightness is lower than the set threshold as a low-light area. Super-resolution technology (such as super-resolution algorithms based on deep learning, interpolation methods, etc.) is used to supplement and reconstruct the low-light area at the pixel level. The goal is to improve the resolution and clarity of the low-light area so that it can be seamlessly connected with the high-brightness area. The joint CT brightness enhancement area generated in step S23 and the low-light area in the original joint CT core image are replaced at the pixel level. Ensure a smooth transition during the replacement process so that the overall image remains visually consistent and natural.

[0101] Preferably, step S23 includes the following steps:

[0102] Step S231: performing low-light region threshold segmentation on the low-light region of the joint CT image to generate low-light region mask data;

[0103] Step S232: performing multi-view image acquisition on the weak-light region of the joint CT image based on the weak-light region mask data to obtain a multi-view CT image of the weak-light region;

[0104] Step S233: spatially registering the multi-view CT images in the weak-light region to generate a multi-view CT registered image in the weak-light region; and reconstructing a super-resolution regional image on the multi-view CT registered image in the weak-light region to generate a super-resolution reconstructed image.

[0105] Step S234: pyramid decompose the super-resolution reconstructed image and the weak-light area of ​​the joint CT image to generate a reconstructed image decomposition image and a joint CT image weak-light area decomposition image; fuse the reconstructed image decomposition image and the joint CT image weak-light area decomposition image to generate a joint CT brightness enhancement area.

[0106] In an embodiment of the present invention, an appropriate threshold segmentation algorithm is applied to the preprocessed joint CT image, such as an adaptive threshold segmentation based on the Otsu method or a global threshold segmentation based on a fixed threshold. These algorithms can segment the image into regions with obvious grayscale changes according to the pixel grayscale value. According to the result of the threshold segmentation, a binary mask image is generated, in which the pixel value of the low-light area is 1 and the pixel value of other areas is 0. This mask will help determine which areas are regarded as low-light areas for the processing and analysis of subsequent steps. The generated low-light area mask data is used to guide the acquisition of multi-view images. In the joint CT image, multiple view images of the same low-light area are collected by selecting different angles or applying rotation and translation transformations. Ensure that each view covers all or most of the low-light area so that the subsequent steps can perform spatial registration and reconstruction. Use image registration techniques, such as feature point matching, mutual information matching, or registration methods based on deep learning, to spatially align the collected multi-view CT images. The goal of registration is to accurately correspond image pixels at different perspectives to eliminate visual offsets and geometric deformations caused by different perspectives. First, the multi-view CT images in the low-light area that have been spatially registered are reconstructed with super-resolution. A super-resolution algorithm or interpolation method based on deep learning is used to increase the details and clarity at the pixel level. The reconstructed image and the low-light area of ​​the original joint CT image are pyramid decomposed to generate multi-scale image data. This step can provide image details at different scales, gradually decomposing from coarse to fine. Finally, image fusion technology, such as weighted fusion or hybrid methods, is used to integrate the reconstructed image decomposition image and the joint CT image low-light area decomposition image. Ensure smooth transition of pixel values ​​during the fusion process to maintain the visual consistency and medical information integrity of the generated joint CT brightness enhancement area.

[0107] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes:

[0108] Step S31: Calculate the image gray value of the joint CT core optimized image to obtain image gray value data;

[0109] Step S32: performing image texture feature analysis on the joint CT core optimized image to generate image texture feature data;

[0110] Step S33: marking the joint structure of the joint CT core optimized image by using the image gray value data and the image texture feature data to generate joint structure marking data;

[0111] Step S34: perform structural detection on the joint structure marking data to generate normal joint structure detection data and abnormal joint structure detection data; perform abnormal joint structure area segmentation on the joint CT core optimized image according to the abnormal joint structure detection data to generate an abnormal joint structure part segmentation image.

[0112] In an embodiment of the present invention, the grayscale value of the pre-processed joint CT core optimized image is calculated pixel by pixel. A simple average grayscale value calculation or a more complex statistical method, such as grayscale histogram analysis, can be used to obtain information on the overall grayscale distribution of the image. Texture features are extracted from the joint CT core optimized image using texture feature analysis methods, such as grayscale co-occurrence matrix (GLCM), grayscale difference matrix (GLDM), directional gradient histogram (HOG), etc. These features can include texture energy, contrast, directionality, etc., which are used to describe the subtle texture changes and structural information of the image. Combined with the image grayscale value data and image texture feature data obtained in steps S31 and S32, the joint CT core optimized image is structurally labeled using a preset algorithm or machine learning model. Areas of different structures in the image, such as bones, soft tissues, etc., can be marked according to the threshold of the grayscale value or the statistical analysis of the texture features. Use the labeled joint structure data for structural detection and analysis. The structure labeling data can be divided into normal structure and abnormal structure by thresholding or a machine learning classifier. Generate joint normal structure detection data and joint abnormal structure detection data to describe which parts are abnormal or have a high or low probability of abnormality. According to the joint abnormal structure detection data, the area marked as abnormal structure in the joint CT core optimization image is segmented. The pixel-level segmentation method or regional growing algorithm can be used to extract and separate the specific parts of the abnormal structure. Finally, a segmented image of the abnormal joint structure is generated for further observation and diagnosis by analysts.

[0113] Preferably, marking the joint structure of the joint CT core optimized image by using the image gray value data and the image texture feature data includes:

[0114] Performing grayscale weight division on the image grayscale value data based on a preset grayscale threshold range to generate grayscale weight division data, wherein the grayscale weight division data includes a first grayscale weight, a second grayscale weight, and a third grayscale weight;

[0115] Performing texture morphology analysis on the image texture feature data to generate texture morphology analysis data; performing texture regularity analysis on the image texture feature data to generate texture regularity analysis data; performing image texture density division according to the texture morphology analysis data and the texture regularity analysis data to generate image texture density division data, wherein the image texture density data includes texture dense data, texture uniform data and texture sparse data;

[0116] The joint structure is marked on the joint CT core optimized image through grayscale weight division data and image texture density division data to generate joint structure marking data.

[0117] In an embodiment of the present invention, the gray value data of the joint CT core optimized image is divided according to a preset gray threshold range. Multiple gray thresholds can be set to divide the gray value into different intervals. The corresponding gray weights, such as the first gray weight, the second gray weight and the third gray weight, are calculated for each interval to describe the gray distribution characteristics of each interval. Appropriate texture analysis methods, such as gray level co-occurrence matrix (GLCM), directional gradient histogram (HOG), etc., are used to analyze the texture features of the joint CT core optimized image. Texture morphology analysis can extract texture shape information in the image, such as texture energy, contrast, etc. Texture regularity analysis can evaluate the uniformity and regularity of texture in the image, such as the uniformity and regularity of texture. Combined with texture morphology analysis data and texture regularity analysis data, the joint CT core optimized image is divided into texture density. The image can be divided into different regions according to texture morphology and regularity, such as texture dense region, texture uniform region and texture sparse region. The division of each region is based on its performance in texture features, reflecting the spatial distribution and changes of texture information in the image. Combine the grayscale weight division data and image texture density division data obtained in steps S31 and S32 to perform structural labeling on the joint CT core optimized image. Appropriate classifiers or rules can be used to label different joint structures, such as bone structure, soft tissue structure, etc., based on the combined features of grayscale weight and texture density. Finally, joint structure labeling data is generated to provide basic data for subsequent structure detection and abnormality analysis.

[0118] As an example of the present invention, refer to Figure 4 As shown, in this example, step S4 includes:

[0119] Step S41: Acquire joint abnormality information data based on the medical information database;

[0120] Step S42: performing abnormal feature association on the abnormal joint structure segmentation image according to the joint abnormality information data to generate joint abnormality structure feature association data;

[0121] Step S43: constructing a model based on the joint abnormality information data to obtain an abnormal joint structure area detection model;

[0122] Step S44: Performing joint abnormality region multimodal feature analysis on the joint abnormality region feature association data through the abnormal joint structure region detection model to generate joint abnormality region multimodal feature analysis data.

[0123] In an embodiment of the present invention, by using a medical information system or database, joint abnormality information data is obtained from an existing joint disease diagnosis report or medical image database. The data includes information such as different types of joint diseases, the location of abnormal structures, and morphological feature descriptions. The joint abnormality information data obtained in step S41 is associated with the abnormal joint structure part segmentation image in the joint CT image for analysis. Image processing technology and data matching algorithms can be used to correspond the abnormal features in the medical information database to the structures in the actual image. As a result, joint abnormal structure feature association data is generated to describe the specific location and morphological features of each abnormal structure part in the image. Using machine learning or deep learning methods, an abnormal joint structure area detection model is constructed based on the joint abnormality information data obtained in step S41. Appropriate algorithms such as convolutional neural networks (CNNs), support vector machines (SVMs), etc. can be selected to design and train models according to the characteristics of the data. The goal is to identify and locate abnormal structure areas in joint CT images through models. The abnormal joint structure area detection model constructed in step S43 is used to analyze the joint abnormal structure feature association data generated in step S42. The features of the abnormal structure, such as morphological features, texture features, grayscale distribution, etc., can be analyzed from multiple perspectives and modalities. The analysis results generate multimodal feature analysis data of abnormal joint areas, providing detailed quantitative and qualitative descriptions for medical imaging diagnosis and research.

[0124] Preferably, step S43 includes the following steps:

[0125] Step S431: extracting joint abnormality features from the joint abnormality information data to obtain joint abnormality information feature data;

[0126] Step S432: dividing the joint abnormality information feature data into a data set to obtain a model training set and a model test set;

[0127] Step S433: performing model training on the model training set by using a convolutional neural network algorithm to generate an abnormal joint structure area training model;

[0128] Step S434: Use the model test set to perform model optimization iteration on the abnormal joint structure area training model, thereby generating an abnormal joint structure area training model.

[0129] In an embodiment of the present invention, key features such as the location of the abnormal structure, morphological feature description, lesion type, etc. are extracted from the joint abnormality information data obtained in step S41. Image processing and feature engineering techniques can be used to convert medical information data into feature vectors or feature descriptions that can be used to train models. The joint abnormality information feature data obtained in step S431 is divided into two parts: a model training set and a model test set. Commonly used division methods include random division, cross-validation, etc., to ensure that the data distribution and feature representation of the training set and the test set are representative and consistent. The model training set in step S432 is trained using a convolutional neural network (CNN) or other deep learning model. Design a suitable network structure and parameter settings, and optimize and train the model according to the complexity and characteristics of the data. The goal is to generate a deep learning model that can accurately identify and locate the abnormal structure area of ​​the joint through training. Use the model test set in step S432 to evaluate and optimize the trained model. The generalization ability and prediction accuracy of the model can be improved by adjusting model parameters, increasing the amount of training data, and applying regularization techniques. Finally, an optimized and verified training model of abnormal joint structure area is generated for subsequent multimodal feature analysis and application of joint abnormality area.

[0130] Preferably, step S5 comprises the following steps:

[0131] Step S51: Performing feature visualization on the joint normal structure detection data and the joint abnormal region multimodal feature analysis data, thereby generating a joint abnormal region feature map and a joint normal region feature map;

[0132] Step S52: Structural marking is performed on the gouty arthritis CT image using the joint abnormal region feature map and the joint normal region feature map to generate a joint CT feature image.

[0133] In an embodiment of the present invention, feature visualization is performed by combining the joint normal structure detection data generated in step S34 and the joint abnormal region multimodal feature analysis data generated in step S44. Image processing tools or data visualization software can be used to display key features in the form of images or charts, such as grayscale distribution maps, texture feature maps, morphological analysis results, etc. The purpose is to intuitively display the characteristic differences and distribution of normal and abnormal structures in joint images. Based on the joint abnormal region feature map and joint normal region feature map generated in step S51, the gouty arthritis CT image is structurally labeled. Computer-aided diagnosis (CAD) systems or image processing algorithms can be used to mark or highlight key abnormal structures and normal structures on the original image. As a result, a joint CT feature image is generated, which shows the characteristic area of ​​joint inflammation after structural labeling, which is helpful for further diagnosis and analysis.

[0134] In this specification, a CT image data feature extraction system for gouty arthritis is provided, which is used to execute the above-mentioned CT image data feature extraction method for gouty arthritis. The CT image data feature extraction system for gouty arthritis includes:

[0135] The region segmentation module is used to obtain CT images of gouty arthritis; perform joint edge detection on the CT images of gouty arthritis to generate CT image joint edge detection data; perform CT image core region segmentation based on the CT image joint edge detection data to generate joint CT core images;

[0136] The low-light reconstruction module is used to extract the low-light area of ​​the joint CT core image to obtain the low-light area of ​​the joint CT image; perform super-resolution reconstruction on the low-light area of ​​the joint CT image to generate the joint CT brightness enhancement area; replace the low-light area of ​​the joint CT brightness enhancement area with the joint CT core image to generate the joint CT core optimized image;

[0137] The structure segmentation module is used to mark the joint structure of the joint CT core optimized image and generate joint structure marking data; perform structure detection on the joint structure marking data to generate joint normal structure detection data and joint abnormal structure detection data; perform abnormal joint structure region segmentation on the joint CT core optimized image according to the joint abnormal structure detection data to generate abnormal joint structure part segmentation image;

[0138] The abnormal feature analysis module is used to obtain joint abnormality information data; perform multimodal feature analysis of the joint abnormality region on the segmented image of the abnormal joint structure part according to the joint abnormality information data, and generate multimodal feature analysis data of the joint abnormality region;

[0139] The feature labeling module is used to perform feature visualization on the normal joint structure detection data and the multimodal feature analysis data of the abnormal joint area, so as to generate a feature map of the abnormal joint area and a feature map of the normal joint area; the gouty arthritis CT image is structurally labeled through the feature map of the abnormal joint area and the feature map of the normal joint area to generate a joint CT feature image.

[0140] The beneficial effect of the present invention is that by performing joint edge detection and core area segmentation on gouty arthritis CT images, the morphological features of the joint area can be extracted, which helps to better understand and analyze the location and range of the lesion. There are some weak light areas in the joint CT images, and the information of these areas is not clear enough. By extracting the weak light areas and performing super-resolution reconstruction, the brightness and details of these areas can be enhanced, and the visualization effect of the image can be improved. Structural marking of the joint CT core optimized image can mark different tissue structures of the joint, such as bones, soft tissues, etc., to provide a basis for subsequent analysis and diagnosis. Through structural detection, the normal structure and abnormal structure of the joint can be further identified and segmented, which is helpful to find lesions or abnormal conditions. According to the joint abnormal information data, the abnormal joint structure part segmentation image is segmented into joint abnormal area, which can accurately locate and extract the abnormal part of the joint. Through multimodal feature analysis, these abnormal areas can be more deeply extracted and analyzed, which helps to understand the nature and extent of the lesion. The joint normal structure detection data and the joint abnormal area multimodal feature analysis data are visualized to generate joint abnormal area feature map and joint normal area feature map. By using these feature maps to structurally mark CT images of gouty arthritis, joint CT feature images can be generated, providing more intuitive and comprehensive information for reference. Therefore, the present invention improves the accuracy of CT image feature extraction through deep learning and multimodal feature analysis, as well as reconstruction and replacement of weak light areas.

[0141] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0142] 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 method for extracting features from CT image data of gouty arthritis. It is characterized in that The following steps are involved: Step S1: acquiring a gouty arthritis CT image; performing joint edge detection on the gouty arthritis CT image to generate CT image joint edge detection data; performing CT image core area segmentation using the CT image joint edge detection data to generate a joint CT core image; Step S2: extracting the weak light area of ​​the joint CT core image to obtain the weak light area of ​​the joint CT image; performing super-resolution reconstruction on the weak light area of ​​the joint CT image to generate the joint CT brightness enhancement area; replacing the weak light area of ​​the joint CT core image with the joint CT brightness enhancement area to generate the joint CT core optimized image; Step S3: marking the joint structure of the joint CT core optimized image to generate joint structure marking data; Performing structural detection on joint structure label data to generate joint normal structure detection data and joint abnormal structure detection data; According to the joint abnormal structure detection data, the joint CT core optimized image is segmented into abnormal joint structure areas to generate abnormal joint structure segmentation images; Step S4: Acquire joint abnormality information data; Performing multimodal feature analysis of the abnormal joint region on the segmented image of the abnormal joint structure according to the abnormal joint information data, and generating multimodal feature analysis data of the abnormal joint region; Step S4 includes the following steps: Step S41: Acquire joint abnormality information data based on the medical information database; Step S42: performing abnormal feature association on the abnormal joint structure segmentation image according to the joint abnormality information data to generate joint abnormality structure feature association data; Step S43: constructing a model based on the joint abnormality information data to obtain an abnormal joint structure area detection model; Step S44: performing joint abnormal region multimodal feature analysis on the joint abnormal structure feature associated data using the abnormal joint structure region detection model to generate joint abnormal region multimodal feature analysis data; Step S5: Perform feature visualization on the joint normal structure detection data and the joint abnormal area multimodal feature analysis data to generate a joint abnormal area feature map and a joint normal area feature map; perform structural labeling on the gouty arthritis CT image through the joint abnormal area feature map and the joint normal area feature map to generate a joint CT feature image.

2. The method for extracting features from CT image data of gouty arthritis according to claim 1, It is characterized in that Step S1 includes the following steps: Step S11: obtaining CT images of gouty arthritis; Step S12: performing image preprocessing on the gouty arthritis CT image to generate a standard CT image, wherein the image preprocessing includes image filtering, image denoising and contrast enhancement; Step S13: Perform joint edge detection on the standard CT image to generate CT image joint edge detection data; Step S14: Segment the core area of ​​the standard CT image using the CT image joint edge detection data to generate a joint CT core image.

3. The method for extracting features from CT image data of gouty arthritis according to claim 1, It is characterized in that Step S2 includes the following steps: Step S21: performing brightness detection on the joint CT core image to generate pixel brightness data of the joint CT core image; Step S22: extracting the weak light area of ​​the joint CT core image based on the pixel brightness data of the joint CT core image to obtain the weak light area of ​​the joint CT image; Step S23: super-resolution reconstruction is performed on the weak-light area of ​​the joint CT image to generate a joint CT brightness enhancement area; Step S24: Replace the joint CT brightness enhancement area and the joint CT core image with the weak light area to generate a joint CT core optimized image.

4. The method for extracting features from CT image data of gouty arthritis according to claim 3, It is characterized in that Step S23 includes the following steps: Step S231: performing low-light region threshold segmentation on the low-light region of the joint CT image to generate low-light region mask data; Step S232: performing multi-view image acquisition on the weak-light region of the joint CT image based on the weak-light region mask data to obtain a multi-view CT image of the weak-light region; Step S233: spatially registering the multi-view CT images in the weak-light region to generate a multi-view CT registered image in the weak-light region; and reconstructing a super-resolution regional image on the multi-view CT registered image in the weak-light region to generate a super-resolution reconstructed image. Step S234: pyramid decompose the super-resolution reconstructed image and the weak-light area of ​​the joint CT image to generate a reconstructed image decomposition image and a joint CT image weak-light area decomposition image; fuse the reconstructed image decomposition image and the joint CT image weak-light area decomposition image to generate a joint CT brightness enhancement area.

5. The method for extracting features from CT image data for gouty arthritis according to claim 1, It is characterized in that Step S3 includes the following steps: Step S31: Calculate the image gray value of the joint CT core optimized image to obtain image gray value data; Step S32: performing image texture feature analysis on the joint CT core optimized image to generate image texture feature data; Step S33: marking the joint structure of the joint CT core optimized image by using the image gray value data and the image texture feature data to generate joint structure marking data; Step S34: perform structural detection on the joint structure marking data to generate normal joint structure detection data and abnormal joint structure detection data; perform abnormal joint structure area segmentation on the joint CT core optimized image according to the abnormal joint structure detection data to generate an abnormal joint structure part segmentation image.

6. The method for extracting features from CT image data for gouty arthritis according to claim 5, It is characterized in that The joint structure marking of the joint CT core optimized image using image gray value data and image texture feature data includes: Performing grayscale weight division on the image grayscale value data based on a preset grayscale threshold range to generate grayscale weight division data, wherein the grayscale weight division data includes a first grayscale weight, a second grayscale weight, and a third grayscale weight; Performing texture morphology analysis on the image texture feature data to generate texture morphology analysis data; performing texture regularity analysis on the image texture feature data to generate texture regularity analysis data; performing image texture density division according to the texture morphology analysis data and the texture regularity analysis data to generate image texture density division data, wherein the image texture density data includes texture dense data, texture uniform data and texture sparse data; The joint structure is marked on the joint CT core optimized image through grayscale weight division data and image texture density division data to generate joint structure marking data.

7. The method for extracting features from CT image data for gouty arthritis according to claim 1, It is characterized in that Step S43 includes the following steps: Step S431: extracting joint abnormality features from the joint abnormality information data to obtain joint abnormality information feature data; Step S432: dividing the joint abnormality information feature data into a data set to obtain a model training set and a model test set; Step S433: performing model training on the model training set by using a convolutional neural network algorithm to generate an abnormal joint structure area training model; Step S434: Use the model test set to perform model optimization iteration on the abnormal joint structure area training model, thereby generating an abnormal joint structure area training model.

8. The method for extracting features from CT image data for gouty arthritis according to claim 1, It is characterized in that Step S5 includes the following steps: Step S51: Performing feature visualization on the joint normal structure detection data and the joint abnormal region multimodal feature analysis data, thereby generating a joint abnormal region feature map and a joint normal region feature map; Step S52: Structural marking is performed on the gouty arthritis CT image using the joint abnormal region feature map and the joint normal region feature map to generate a joint CT feature image.

9. A CT image data feature extraction system for gouty arthritis, It is characterized in that For executing the CT image data feature extraction method for gouty arthritis as claimed in claim 1, the CT image data feature extraction system for gouty arthritis comprises: The region segmentation module is used to obtain CT images of gouty arthritis; perform joint edge detection on the CT images of gouty arthritis to generate CT image joint edge detection data; perform CT image core region segmentation based on the CT image joint edge detection data to generate joint CT core images; The weak light reconstruction module is used to extract the weak light area of ​​the joint CT core image to obtain the weak light area of ​​the joint CT image; perform super-resolution reconstruction on the weak light area of ​​the joint CT image to generate the joint CT brightness enhancement area; replace the weak light area of ​​the joint CT core image with the joint CT brightness enhancement area to generate the joint CT core optimized image; The structure segmentation module is used to mark the joint structure of the joint CT core optimized image and generate joint structure marking data; perform structure detection on the joint structure marking data to generate joint normal structure detection data and joint abnormal structure detection data; perform abnormal joint structure region segmentation on the joint CT core optimized image according to the joint abnormal structure detection data to generate abnormal joint structure part segmentation image; The abnormal feature analysis module is used to obtain joint abnormality information data; perform multimodal feature analysis of the joint abnormality region on the segmented image of the abnormal joint structure part according to the joint abnormality information data, and generate multimodal feature analysis data of the joint abnormality region; The feature labeling module is used to perform feature visualization on the normal joint structure detection data and the multimodal feature analysis data of the abnormal joint area, so as to generate a feature map of the abnormal joint area and a feature map of the normal joint area; the gouty arthritis CT image is structurally labeled through the feature map of the abnormal joint area and the feature map of the normal joint area to generate a joint CT feature image.

Images