Arthroscopic imaging-based joint injury identification method and system

Abstract

The invention discloses a joint injury identification method and system based on arthroscopic imaging. The method comprises the steps that an original image, collected by an arthroscope, of intra-articular tissue is acquired; sequentially performing denoising, artifact removal and deblurring processing on the original image to obtain a preliminary original image; performing corresponding semantic segmentation according to the initial original image so as to segment sub-regions corresponding to different tissues on the initial original image, and performing image enhancement on the different sub-regions by adopting corresponding image enhancement strategies to obtain a target image; and performing visual display on the target image through a display device so as to perform joint injury identification on the current joint. According to the method, arthroscope imaging is enhanced, and medical staff are assisted in recognizing the joint injury state.

Description

TECHNICAL FIELD

[0001] The present application relates to the medical technology field, in particular to a joint damage identification method and system based on arthroscopic imaging. BACKGROUND

[0002] Arthroscopy is a minimally invasive endoscopic examination technology, which is specially used for observing the internal structure of the joint. By making a small incision around the knee joint, an elongated endoscope (arthroscope) with a light source and a camera is inserted into the joint cavity, and physiological saline is injected to expand the joint cavity, so that the lens can clearly observe the structures such as cartilage, synovial membrane and bone surface inside the joint. Then the image inside the joint is transmitted to the display through the lens, which helps the doctor to clearly observe the condition inside the joint, and then accurately diagnose and treat various joint diseases.

[0003] However, when the arthroscope takes pictures, it is easily affected by factors such as device performance, operating environment, and joint conditions, for example, lens shaking during operation causes image blur, air bubbles brought in when physiological saline is injected cause air bubble artifacts, and turbidity of joint fluid causes low contrast, which ultimately leads to poor imaging quality of the arthroscope, affecting the judgment of medical personnel. SUMMARY

[0004] Therefore, the purpose of the present application is to provide a joint damage identification method and system based on arthroscopic imaging, which aims to solve the problem of poor imaging quality of the arthroscope in the prior art, which affects the judgment of medical personnel.

[0005] The embodiment of the present application is implemented as follows: On the one hand, a joint damage identification method based on arthroscopic imaging is proposed, which is applied to a display device in communication connection with an arthroscope, and the method comprises: obtaining an original image of the joint tissue collected by the arthroscope; performing denoising, artifact removal and deblurring on the original image in sequence to obtain a preliminary original image; performing corresponding semantic segmentation on the preliminary original image to segment different tissue corresponding sub-regions on the preliminary original image, and performing image enhancement on different sub-regions using corresponding image enhancement strategies to obtain a target image; visualizing and displaying the target image through the display device for joint damage identification of the current joint.

[0006] Further, the joint damage identification method based on arthroscopic imaging, wherein the step of sequentially performing denoising, artifact removal and deblurring on the original image to obtain a preliminary original image comprises: obtaining the noise type in the original image, and denoising the original image using different filtering algorithms according to different noise types; The artifact area in the original image is identified, and the artifact position is filled by combining the neighborhood pixel interpolation to remove the artifact; The original image is deblurred by using a preset deblurring algorithm to restore the clear image of the original image through iterative calculation. Among them, the median filter or bilateral filter is used to remove salt and pepper noise, retain the edge while eliminate isolated noise points, and the Gaussian filter or wavelet transform is used to remove Gaussian noise, while retaining the organization details while smoothing the noise.

[0007] Further, the above-mentioned arthroscopic imaging-based joint damage identification method, wherein the step of identifying the artifact area in the original image, combining the neighborhood pixel interpolation to fill the artifact position to remove the artifact comprises: Convert the original image to HSV space to extract the brightness channel, and mark the area with a gray value greater than a threshold in the brightness channel as a candidate highlight area based on the gray level histogram of the brightness channel; Perform connected component analysis on the candidate highlight area, calculate the number of pixels and the circularity of each connected component, and determine the area that meets the pixel number and circularity conditions as the bubble artifact area and generate a binary mask according to the pixel number and circularity; Edge detection is performed on the binary mask of the bubble artifact area to obtain the closed contour of the bubble artifact area, and the pixel points on the contour are marked as filling boundary points; For each bubble artifact area, calculate the minimum circumscribed rectangle of the bubble artifact area, determine the filling range, take the boundary of the bubble artifact area as the starting point, and expand a ring area with a preset pixel width as the effective neighborhood, extract the gray mean, standard deviation and texture direction of the pixels in the effective neighborhood, and combine the gray mean, standard deviation and texture direction of the neighborhood pixels to fill the bubble artifact area to remove the artifact.

[0008] Further, the above-mentioned arthroscopic imaging-based joint damage identification method, wherein the step of combining the gray mean, standard deviation and texture direction of the neighborhood pixels to fill the bubble artifact area to remove the artifact comprises: For the artifact area at a first preset pixel distance from the boundary of the bubble artifact area, use bilinear interpolation to calculate the filling value of the gray value with the neighborhood pixels of the boundary points and their symmetric positions as reference; For the artifact area at a second preset pixel distance from the boundary of the bubble artifact area, use weighted mean interpolation to calculate the filling value of the gray value according to the Euclidean distance between the neighborhood pixels and the filling points; Wherein, the second preset pixel is greater than the first preset pixel, the gray value of the filled pixel falls within a preset range, and the preset range is obtained according to the gray mean and standard deviation of the neighborhood pixels.

[0009] Further, the above-mentioned arthroscopic imaging-based joint damage identification method, wherein the sub-regions include cartilage regions, bone cortex regions, and synovial membrane and effusion interface regions.

[0010] Further, the above-mentioned arthroscopic imaging-based joint damage identification method, wherein the step of performing image enhancement on different sub-regions by using corresponding image enhancement strategies to obtain a target image comprises: for the cartilage region, using Gamma correction to enhance brightness for image enhancement; for the bone cortex region, using Laplacian pyramid decomposition for image enhancement; for the synovial membrane and effusion interface region, using local contrast stretching for image enhancement.

[0011] Further, the above-mentioned arthroscopic imaging-based joint damage identification method, wherein the step of obtaining the original image of the intra-articular tissue collected by the arthroscope further comprises: determining whether the original image meets a preset standard; if yes, directly visualizing and displaying the original image by a display device for joint damage identification of the current joint; if no, performing the step of sequentially denoising, removing artifacts, and deblurring the original image to obtain a preliminary original image.

[0012] Another object of the present application is to provide an arthroscopic imaging-based joint damage identification system, which is applied to a display device in communication connection with an arthroscope, and the system comprises: an acquisition module configured to acquire an original image of intra-articular tissue collected by the arthroscope; a preprocessing module configured to sequentially denoise, remove artifacts, and deblur the original image to obtain a preliminary original image; an enhancement module configured to perform corresponding semantic segmentation according to the preliminary original image, to segment different tissue corresponding sub-regions on the preliminary original image, and to perform image enhancement on different sub-regions by using corresponding image enhancement strategies to obtain a target image; an identification module configured to visualize and display the target image by a display device for joint damage identification of the current joint.

[0013] In another aspect, the present application provides a readable storage medium having a computer program stored thereon, wherein the program is executed by a processor to implement the steps of the above-mentioned method.

[0014] In another aspect, the present application provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned method when executing the program.

[0015] The embodiment of the present application obtains the original image of intra-articular tissue collected by an arthroscope; sequentially performs denoising, artifact removal and deblurring processing on the original image to obtain a preliminary original image; performs corresponding semantic segmentation on the preliminary original image to segment out sub-regions corresponding to different tissues on the preliminary original image, and performs image enhancement on different sub-regions by using corresponding image enhancement strategies to obtain a target image; and visualizes the target image by using a display device for use in joint damage identification of the current joint. The image quality of arthroscopic imaging is improved by denoising, artifact removal and deblurring processing, and image enhancement is performed on different tissue regions respectively. The present application solves the problem of poor imaging quality of the arthroscope in the prior art, which affects the judgment of medical personnel. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 A flowchart of the joint damage identification method based on arthroscopic imaging proposed in the first embodiment of the present application; Figure 2 A structural schematic diagram of the joint damage identification system based on arthroscopic imaging in the third embodiment of the present application.

[0017] The following specific embodiments will further illustrate the present application in combination with the above drawings. DETAILED DESCRIPTION

[0018] In order to facilitate the understanding of the present application, the present application will be described more fully below with reference to the related drawings. Several embodiments of the present application are shown in the drawings. However, the present application can be realized in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive.

[0019] It should be noted that when an element is referred to as being "fixedly attached" to another element, it can be directly on the other element or there can be an intervening element. When an element is referred to as being "connected" to another element, it can be directly connected to the other element or intervening elements can be present. As used herein, the terms "vertical", "horizontal", "left", "right", and the like are merely for the purpose of illustration.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs. The terminology used in the description of the present application herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0021] Embodiment one Please refer toFigure 1 , which is a joint injury identification method based on arthroscopic imaging in the first embodiment of the present application, applied to a display device in communication connection with an arthroscope, and the method comprises steps S10-S13.

[0022] Step S10, obtaining the original images of intra-articular tissues collected by the arthroscope.

[0023] Among them, the original images of intra-articular tissues collected by the arthroscope during the examination of the joint are obtained. These original images are directly derived from the shooting of cartilage, synovial membrane, bone surface and other intra-articular structures after the arthroscope is inserted into the joint cavity. Although they contain basic information of intra-articular tissues, there may be problems affecting image quality due to factors such as operating environment and intra-articular conditions. Obtaining these original images is the basis for subsequent joint injury identification related processing, and provides original data support for further processing and analysis of images in subsequent steps. In addition, in some optional embodiments of the present application, the step of obtaining the original images of intra-articular tissues collected by the arthroscope further comprises: determining whether the original image meets the preset standard; If yes, the original image is directly visualized and displayed by the display device for joint injury identification of the current joint; If not, the step of obtaining the preliminary original image by sequentially performing denoising, artifact removal and deblurring processing on the original image is performed.

[0024] Among them, after obtaining the original images of intra-articular tissues collected by the arthroscope, instead of directly entering the subsequent image processing flow, a judgment link is first added, that is, to evaluate whether the original image meets the preset quality standard, such as the proportion of noise points in the image not exceeding a certain proportion, no obvious bubble artifact, and the main tissue contour is clear and distinguishable, etc. If the original image meets all these preset standards, it means that its quality is sufficient to support medical staff to identify joint injury, at this time the original image is directly visualized and displayed by the display device for medical staff to directly observe and judge. If the original image does not meet the preset standard, there are too many noise points, obvious bubble artifact, blurred image and other problems affecting observation, then the subsequent processing flow is started, that is, the original image is sequentially denoised, artifact removed and deblurred to eliminate or reduce these quality problems, and finally a preliminary original image is obtained to prepare for the subsequent steps of semantic segmentation and image enhancement. Such design can not only save unnecessary processing steps and improve efficiency when the quality of the original image meets the standard, but also improve the quality through targeted processing when the image quality is poor, ensuring to provide reliable image basis for joint injury identification.

[0025] Step S11, sequentially performing denoising, artifact removal and deblurring processing on the original image to obtain a preliminary original image.

[0026] After obtaining the original image collected by the arthroscope, a series of processing of denoising, artifact removal and deblurring is carried out on the original image in order to solve the quality problems existing therein, and finally a preliminary original image is obtained. The denoising processing removes the generated noise points, thereby preliminarily improving the image clarity and reducing the influence of noise points on subsequent analysis. The artifact removal accurately identifies and eliminates the bubble artifact brought in when physiological saline is injected, so that the intra-articular tissue is more clearly presented, and the artifact interference in observation is avoided. The deblurring processing analyzes the image blur characteristics and degree caused by intraoperative lens shaking, reversely calculates the possible state of the original clear image, effectively improves the image clarity, restores the true form and details of the tissue, and after the three-step orderly processing, the problems of noise points, bubble artifacts and blur in the original image are effectively solved, and the image quality is significantly improved. Step S12, performing corresponding semantic segmentation on the preliminary original image to segment out the sub-regions corresponding to different tissues on the preliminary original image, and performing image enhancement on different sub-regions by using corresponding image enhancement strategies to obtain a target image.

[0027] After the preliminary original image is obtained through denoising, artifact removal and deblurring processing, the image is segmented by using deep learning technology and the like, the characteristics of different tissues in the image are accurately identified, and the sub-regions corresponding to various tissues such as cartilage, synovial membrane and bone surface are clearly segmented on the preliminary original image. Then, according to the imaging characteristics (such as the texture performance of cartilage and the color characteristics of synovial membrane) of these different sub-regions in the arthroscope image and the importance (such as the judgment of cartilage damage depending on the texture details and the recognition of synovial inflammation needing to highlight the color difference) in diagnosis, the image enhancement strategy matched therewith is used, such as highlighting the texture details of the cartilage sub-region to highlight the subtle damage and highlighting the color characteristics of the synovial membrane sub-region to assist in inflammation judgment. Through this individualized enhancement processing, the key information of each tissue sub-region can be clearly presented, and finally the target image accurately reflecting the conditions of various tissues in the joint is obtained.

[0028] Specifically, in the process of semantic segmentation based on the preliminary raw image, the core is to accurately identify and regionally divide different joint tissues (such as cartilage, synovium, and bone surface) in the image with the help of deep learning algorithms. First, a deep learning model for arthroscopic images needs to be constructed, which is usually based on a convolutional neural network (CNN) architecture (such as U-Net, DeepLab series, etc.). These networks can gradually extract low-level features (such as edges, textures) and high-level features (such as tissue morphology, structural relationships) in the image through multiple convolutional operations. During the model training phase, a large number of annotated arthroscopic image datasets are used, i.e., each image is pre-labeled with different tissue pixel regions. Through the backpropagation algorithm, the model parameters are continuously optimized, enabling the model to learn the unique visual features of each type of tissue, such as the smooth texture of cartilage, the red tone distribution of synovium, and the hard texture of bone surface. When the model is trained, the pre-processed preliminary raw image is input into the model, which classifies and judges each pixel point in the image to determine which tissue the pixel belongs to. Finally, clear boundaries are generated on the image, and different tissue sub-regions are completely segmented out, just like "labeling" different tissues and dividing independent regions. This achieves automatic and high-precision segmentation of multiple tissues in a complex intra-articular environment, providing accurate regional positioning for subsequent individualized image enhancement of different tissue sub-regions. It should be noted that how to perform semantic segmentation is understood by those skilled in the art, and excessive repetition is not necessary here.

[0029] More specifically, in some optional embodiments of the present application, the sub-regions include cartilage regions, bone cortex regions, and synovium and effusion interface regions; the step of performing image enhancement on different sub-regions using corresponding image enhancement strategies to obtain a target image includes: Gamma correction is used to enhance the brightness of the cartilage region; Laplacian pyramid decomposition is used to enhance the image of the bone cortex region; Local contrast stretching is used to enhance the image of the synovium and effusion interface region.

[0030] Since the cartilage region is often dark due to insufficient illumination, Gamma correction can be used to brighten it. Guided filtering can enhance texture while suppressing noise, and edge replacement can prevent filter blurring of the boundary, making the cartilage details clearer. Therefore, the cartilage region image can be extracted based on the sub-region mask, the gray mean value is calculated, and if the mean value is less than a threshold, Gamma correction is performed with a corresponding gamma value to enhance the brightness. For the corrected cartilage image, a corresponding guided filter is applied to enhance the cartilage surface texture (such as micro-cracks). Finally, the Canny operator is used to extract the cartilage region edge, and the edge region pixel value of the filtered image is replaced with the corresponding pixel value after Gamma correction to preserve the edge clarity.

[0031] Further, since the Laplacian pyramid decomposition can separate the high-frequency details (fracture lines) and low-frequency noise of the bone cortex, enhancing the high-frequency components of specific layers can accurately highlight the fracture lines, and the reconstructed image ensures the integrity of the overall structure; therefore, based on the mask extraction of the bone cortex region image, Gaussian pyramid decomposition (decomposed into 5 layers) is performed, and 4 layers of Laplacian pyramid images are obtained by subtracting adjacent layers; the 3rd and 4th layers of Laplacian images containing high-frequency information are selected, and the pixel values are multiplied by 1.5 to enhance the high-frequency components (highlight the fracture lines), and the remaining layers remain unchanged; the processed Laplacian images of each layer are reconstructed by inverse pyramid transformation to obtain the enhanced bone cortex sub-region image; Further, the gray difference between the synovial membrane and the effusion is small, and the local contrast stretching can expand the difference and clearly define the boundary area; the gray limit can avoid abnormal pixels and ensure the natural vision of the image; the fusion step can integrate the enhanced results of each region to obtain a complete target image; the boundary area between the synovial membrane and the effusion can be determined according to the sub-region masks of the synovial membrane and the effusion, and the boundary area and the surrounding area of 10 pixels are taken as the processing area. The gray value range of the processing area is counted, and if the main gray value is concentrated in 120-150 (0-255 range), the gray value range is stretched to 100-180 through the linear transformation formula "output gray value=((input gray value-120) / (150-120))×(180-100)+100". The stretched image is subjected to gray limit processing, and the pixel values exceeding 255 are set to 255, and the values below 0 are set to 0 to avoid gray overflow. Finally, the enhanced sub-region images can be fused into the same image according to the position information of the original image through the mask to obtain the final target image.

[0032] Step S13, the target image is visually displayed through the display device for joint damage identification of the current joint.

[0033] Wherein, after the target image is obtained through semantic segmentation and targeted image enhancement processing, the image is transmitted to a display device (such as a high-definition display and the like) in communication connection with the arthroscopic system, and the target image is presented in a clear and intuitive visual form through the display device. At this time, the target image not only eliminates the noise, artifacts and blur problems in the original image, but also performs individualized enhancement on different tissue sub-regions, such as more prominent texture details in the cartilage region and more obvious color differences in the synovial membrane region, so that the morphology, structure and possible damage characteristics (such as cracks in the cartilage, hyperemia and swelling of the synovial membrane, and wear on the bone surface) of various tissues in the joint can be clearly and accurately displayed. Medical personnel can intuitively and accurately identify whether the current joint is damaged, the location, type and severity of the damage, and other information by observing the target image on the display device, thereby providing a reliable visual basis for the diagnosis of joint damage and the development of subsequent treatment plans, and effectively improving the accuracy and efficiency of joint damage identification.

[0034] In summary, the joint damage identification method based on arthroscopic imaging in the above embodiments of the present application acquires an original image of the intra-articular tissue collected by an arthroscope; sequentially performs denoising, artifact removal and deblurring processing on the original image to obtain a preliminary original image; performs corresponding semantic segmentation on the preliminary original image to segment different tissue corresponding sub-regions on the preliminary original image, and performs image enhancement on different sub-regions using corresponding image enhancement strategies to obtain a target image; and visually displays the target image through a display device for joint damage identification of the current joint. By denoising, artifact removal and deblurring processing, and image enhancement according to different tissue regions, the image quality of arthroscopic imaging is improved. The present application solves the problem of poor imaging quality of the arthroscope in the prior art, which affects the judgment of medical personnel.

[0035] Embodiment Two The present embodiment also proposes a joint damage identification method based on arthroscopic imaging. The joint damage identification method based on arthroscopic imaging in the present embodiment differs from the joint damage identification method based on arthroscopic imaging in Embodiment One in that: The step of sequentially performing denoising, artifact removal and deblurring processing on the original image to obtain a preliminary original image comprises: Acquire the noise type in the original image, and use different filtering algorithms to denoise the original image according to different noise types; Identify the artifact region in the original image, and fill the artifact position by combining with the neighborhood pixel interpolation to remove the artifact; Use a preset deblurring algorithm to restore the clear image of the original image through iterative calculation for deblurring processing of the original image; In the method, the salt and pepper noise is removed by using median filtering or bilateral filtering, the isolated noise points are removed while the edges are preserved, and the Gaussian noise is removed by using Gaussian filtering or wavelet transform, the noise is smoothed while the details of the intra-articular tissue are preserved as much as possible.

[0036] In the process of sequentially performing the denoising, artifact removal and deblurring on the original image to obtain the preliminary original image, a targeted processing method is adopted: firstly, the denoising processing, the noise types existing in the original image are first determined, and then corresponding filtering algorithms are selected according to different noise types, wherein for the salt and pepper noise, median filtering or bilateral filtering is used for removal, these two algorithms can effectively remove isolated noise points while preserving the edge information of the image, and avoid edge blurring; for the Gaussian noise, Gaussian filtering or wavelet transform is used, the noise is smoothed while the details of the intra-articular tissue are preserved as much as possible, and sufficient tissue information can be obtained for subsequent analysis; then, the artifact removal, by identifying the artifact area in the original image caused by bubbles and the like, the information of the surrounding neighborhood pixels of the area is used for interpolation filling to fill the artifact position, so that the artifact area in the image is repaired and the complete presentation of the tissue is restored; finally, the deblurring processing, a preset deblurring algorithm is used to process the image by iterative calculation, and the original image blurred due to lens shaking and the like is gradually restored to a clear state, through this series of targeted processing steps, the noise, artifact and blurring problems in the original image are effectively solved, and finally the preliminary original image with improved quality is obtained, which lays a good foundation for subsequent semantic segmentation and the like.

[0037] Specifically, the step of identifying the artifact area in the original image and filling the artifact position by interpolation of the neighborhood pixels for artifact removal comprises: Converting the original image into HSV space to extract the brightness channel, and marking the area with a gray value greater than a threshold in the brightness channel as a candidate highlight area based on a gray histogram of the brightness channel; Performing connected component analysis on the candidate highlight area, calculating the number of pixels and the circularity of each connected component, and determining the area meeting the pixel number and circularity conditions as a bubble artifact area and generating a binary mask according to the pixel number and the circularity; Performing edge detection on the binary mask of the bubble artifact area to obtain a closed contour of the bubble artifact area, and marking the pixel points on the contour as filling boundary points; For each bubble artifact area, calculating the minimum circumscribed rectangle of the bubble artifact area to determine the filling range, taking the boundary of the bubble artifact area as the starting point, expanding a ring-shaped area with a preset pixel width outward as an effective neighborhood, extracting the gray mean value, standard deviation and texture direction of the pixels in the effective neighborhood, and filling the bubble artifact area by using the gray mean value, standard deviation and texture direction of the neighborhood pixels for artifact removal.

[0038] Firstly, the original image is converted to the HSV color space, and the brightness channel is extracted from it because the bubble artifact often has a significant difference in brightness characteristics from the surrounding tissue. Based on the gray level histogram of the brightness channel, a specific threshold is set to mark the regions with a gray level value exceeding the threshold as candidate high-light regions, preliminarily locking the range of possible bubble artifacts. Then, connected component analysis is performed on these candidate high-light regions. By calculating the number of pixels and the circularity (bubble usually presents a shape close to a circle) contained in each connected component, regions that meet the conditions are selected according to the preset pixel number range and circularity standard, and are determined as bubble artifact regions. A corresponding binary mask (i.e., distinguishing the artifact region from the normal region by the pixel values of 0 and 1, with the artifact region marked as 1 and other regions as 0) is generated to achieve accurate positioning of the bubble artifact. Then, edge detection is performed on the binary mask of the bubble artifact region to obtain the closed contour of the bubble artifact region, and the pixel points on the contour are marked as filling boundary points to clearly define the range boundary for subsequent interpolation filling. Finally, for each bubble artifact region, the minimum bounding rectangle is calculated to determine the specific range of filling, and a ring-shaped region with a preset pixel width is expanded outward from the boundary of the bubble artifact region as an effective neighborhood (this region contains the pixel information of the normal tissue around the artifact), and the gray level mean, standard deviation (reflecting the distribution characteristics of pixel gray levels) and texture direction (reflecting the structure direction of the tissue) of the pixels in the effective neighborhood are extracted. The filling of the bubble artifact region is calculated by interpolation using these neighborhood pixel feature information, so that the artifact is removed while the filled region and the surrounding normal tissue are naturally connected in terms of gray level, texture, etc., maximizing the restoration of the true state of the image and eliminating the interference of the bubble artifact on the observation of the intra-articular tissue. Specifically, the step of filling the bubble artifact region by combining the gray level mean, standard deviation and texture direction of the neighborhood pixels for artifact removal includes: For the artifact region at a first preset pixel distance from the boundary of the bubble artifact region, a bilinear interpolation is used to calculate the filling value of the gray level based on the neighborhood pixels of the boundary points and their symmetric positions. For the artifact region at a second preset pixel distance from the boundary of the bubble artifact region, a weighted mean interpolation is used to calculate the filling value of the gray level based on the Euclidean distance between the neighborhood pixels and the filling point. Wherein, the second preset pixel is greater than the first preset pixel, and the gray level value of the filled pixel falls within a preset range, which is obtained based on the gray level mean and standard deviation of the neighborhood pixels.

[0039] In a specific implementation, when filling the bubble artifact area by combining the gray mean value, standard deviation and texture direction of the neighborhood pixels to achieve artifact removal, a differentiated interpolation filling strategy is adopted: first, the bubble artifact area is divided into two parts according to the distance from the boundary, for the artifact area within the first preset pixel range from the boundary of the bubble artifact area, a bilinear interpolation method is adopted, the pixels on the boundary of the area and the neighborhood pixels at the symmetrical position thereof are taken as references, and the gray value of the filling point is determined by calculating the linear combination of the gray values of these reference pixels, which can make the filling area transition naturally with the boundary and well fit the pixel characteristics at the boundary; for the artifact area within the second preset pixel range (and the second preset pixel is greater than the first preset pixel) from the boundary, a weighted mean interpolation is adopted, different weights are given according to the Euclidean distance between the neighborhood pixels and the filling point, the closer the neighborhood pixel, the greater the weight, and the farther the weight, the smaller the weight, and the gray value of the filling point is obtained by calculating the gray values of the neighborhood pixels by weighting, so as to better utilize the pixel information of a wider range of neighborhood, and make the gray distribution in the filling area more consistent with the overall characteristics of the surrounding tissue; at the same time, in order to ensure the consistency and authenticity of the filled image, the gray value of the filled pixel is limited in a preset range, and the preset range is determined according to the gray mean value and standard deviation of the pixels in the effective neighborhood extracted before (for example, a reasonable fluctuation interval is determined in the center of the gray mean value in combination with the standard deviation), so that the filling value can avoid abnormal gray and the gray characteristics of the filling area can be coordinated with the neighborhood tissue, and finally the natural filling and effective removal of the bubble artifact area are realized, so that the repaired image is closer to the real intra-articular tissue state.

[0040] In summary, the joint damage identification method based on arthroscopic imaging in the above embodiments of the present application, by acquiring the original image of the intra-articular tissue collected by the arthroscope, sequentially performing denoising, artifact removal and deblurring processing on the original image to obtain a preliminary original image, performing corresponding semantic segmentation on the preliminary original image to segment different tissue corresponding sub-regions on the preliminary original image, and performing image enhancement on different sub-regions by corresponding image enhancement strategies to obtain a target image, and visualizing the target image through a display device for joint damage identification of the current joint. By denoising, artifact removal and deblurring processing, and image enhancement according to different tissue regions, the image quality of arthroscopic imaging is improved. The present application solves the problem of poor imaging quality of the arthroscope in the prior art, which affects the judgment of medical personnel.

[0041] Embodiment three Please refer to Figure 2 , which shows the joint damage identification system based on arthroscopic imaging proposed in the third embodiment of the present application, applied to a display device in communication connection with an arthroscope, the system comprises: The acquisition module 100 is configured to acquire an original image of intra-articular tissue collected by an arthroscope. The preprocessing module 200 is configured to sequentially perform denoising, artifact removal and deblurring on the original image to obtain a preliminary original image. The enhancement module 300 is configured to perform corresponding semantic segmentation on the preliminary original image to segment different sub-regions corresponding to different tissues on the preliminary original image, and perform image enhancement on different sub-regions by using corresponding image enhancement strategies to obtain a target image. The identification module 400 is configured to visually display the target image by using a display device, so as to identify joint damage of the current joint.

[0042] The functions or operation steps realized when the above modules are executed are basically the same as those of the above method embodiments, and thus will not be described herein again.

[0043] Embodiment Four Another aspect of the present application further provides a readable storage medium having a computer program stored thereon, wherein the program is executed by a processor to realize the steps of the method according to any one of the above embodiments one to two.

[0044] Embodiment Five Another aspect of the present application further provides an electronic device, which comprises a memory, a processor and a computer program stored in the memory and running on the processor, wherein the processor realizes the steps of the method according to any one of the above embodiments one to two when executing the program.

[0045] The technical features of the above embodiments can be combined in any manner, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described herein, however, as long as the combinations of the technical features do not contradict, they should be considered as the scope of the present application.

[0046] Those skilled in the art can understand that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a list of executable instructions for realizing the logic function, which can be embodied in any computer readable storage medium for use by or in connection with an instruction execution system, apparatus or device, such as a computer-based system, a system including a processor or other system that can fetch and execute instructions from the instruction execution system, apparatus or device. For the present specification, the "computer readable storage medium" can be any device that can contain, store, communicate, propagate or transport programs for use by or in connection with the instruction execution system, apparatus or device, or in conjunction with these instruction execution systems, apparatus or devices.

[0047] More specific examples (a non-exhaustive list) of the computer-readable storage medium include the following: an electrical connection having one or more wires (electrical, optical, and the like) a portable computer diskette (magnetic, or optical, e.g., Blu-ray® disk, etc.) a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, and a portable compact disc read-only memory (CDROM). Additionally, the computer readable storage medium can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for instance via an optical scanner, then compiled, interpreted, or otherwise processed, using an appropriate medium, into a computer program in a suitable language.

[0048] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the embodiments described above, various steps or methods can be implemented, for example, by software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, and in another embodiment, any of the following techniques, which are well known in the art, can be used to implement the application: a hybrid of the techniques mentioned above; a combination of one or more of the techniques mentioned above; or one or more other techniques suitable for use in the computer hardware devices described above.

[0049] In the description of the specification, the description of the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the application. In the specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in one or more embodiments or examples.

[0050] The above-described embodiments only express several implementation manners of the application, which are described in a more specific and detailed manner, but cannot be understood as a limitation on the patent scope of the application. It should be noted that for those skilled in the art, several modifications and improvements can be made without departing from the concept of the application, which are all within the protection scope of the application. Therefore, the patent protection scope of the application should be subject to the appended claims.

Claims

1. A method for identifying joint injuries based on arthroscopic imaging, characterized in that: Applied to a display device in communication with an arthroscope, the method comprises: Obtain original images of intra-articular tissues collected by arthroscopy; The original image is subjected to denoising, artifact removal and deblurring processing in sequence to obtain a preliminary original image; Performing corresponding semantic segmentation based on the preliminary original image to segment sub-regions corresponding to different tissues on the preliminary original image, and performing image enhancement using corresponding image enhancement strategies on different sub-regions to obtain the target image; The target image is visualized through a display device to be used for identifying joint damage of the current joint.

2. The joint injury identification method based on arthroscopic imaging according to claim 1, characterized in that: The step of sequentially performing denoising, artifact removal, and deblurring on the original image to obtain a preliminary original image comprises: Obtain the noise type in the original image and use different filtering algorithms to denoise the original image according to different noise types; Identify the artifact area in the original image and remove the artifact by filling the artifact position with neighboring pixel interpolation; Using the preset deblurring algorithm, the original image is deblurred through iterative calculation to restore the clear image of the original image; Among them, median filtering or bilateral filtering is used to remove salt and pepper noise, preserving edges while eliminating isolated noise points, and Gaussian filtering or wavelet transform is used to remove Gaussian noise, smoothing noise while preserving tissue details.

3. The joint injury identification method based on arthroscopic imaging according to claim 1, characterized in that: The steps of identifying the artifact area in the original image and filling the artifact position by combining neighboring pixel interpolation to remove the artifact include: The original image is converted into HSV space to extract the brightness channel. Based on the grayscale histogram of the brightness channel, the areas with grayscale values ​​greater than the threshold in the brightness channel are marked as candidate highlight areas. Perform connected domain analysis on the candidate highlight areas, calculate the number of pixels and circularity of each connected domain, retain the areas that meet the pixel number and circularity conditions according to the pixel number and circularity, determine them as bubble artifact areas, and generate a binary mask; Perform edge detection on the binary mask of the bubble artifact area to obtain the closed contour of the bubble artifact area, and mark the pixels on the contour as filling boundary points; For each bubble artifact area, the minimum circumscribed rectangle of the bubble artifact area is calculated to determine the filling range. Starting from the boundary of the bubble artifact area, a ring area with a preset pixel width is expanded outward as the effective neighborhood. The grayscale mean, standard deviation and texture direction of the pixels in the effective neighborhood are extracted. The bubble artifact area is filled and artifacts are removed by combining the grayscale mean, standard deviation and texture direction of the neighborhood pixels.

4. The joint injury identification method based on arthroscopic imaging according to claim 3, characterized in that: The step of removing artifacts by filling the bubble artifact area in combination with the grayscale mean, standard deviation and texture direction of the neighborhood pixels includes: For the artifact area that is the first preset pixel away from the boundary of the bubble artifact area, bilinear interpolation is used to calculate the grayscale filling value with reference to the boundary point and the neighboring pixels at the symmetrical position thereof; For the artifact area that is second preset pixels away from the boundary of the bubble artifact area, weighted mean interpolation is used to calculate the filling value of the grayscale value according to the Euclidean distance between the neighboring pixels and the filling point; The second preset pixel is larger than the first preset pixel, and the grayscale value of the filled pixel falls within a preset range, which is obtained based on the grayscale mean and standard deviation of the neighboring pixels.

5. The method for identifying joint injuries based on arthroscopic imaging according to claim 1, characterized in that: The subregions include the cartilage region, the cortical bone region, and the synovium-effusion junction.

6. The joint injury identification method based on arthroscopic imaging according to claim 5, characterized in that: The step of performing image enhancement on different sub-regions using corresponding image enhancement strategies to obtain a target image includes: Gamma correction is used to enhance the brightness of the cartilage area for image enhancement; Image enhancement is performed on the bone cortical area using Laplacian pyramid decomposition; Local contrast stretching was used to enhance the image at the junction of the synovium and effusion.

7. The method for identifying joint injuries based on arthroscopic imaging according to claim 1, characterized in that: After the step of obtaining the original image of the intra-articular tissue collected by the arthroscopic method, the method further includes: Determine whether the original image meets the preset standards; If yes, the original image is directly visualized through a display device to be used for joint injury identification of the current joint; If not, the steps of performing denoising, artifact removal and deblurring on the original image in sequence are executed to obtain a preliminary original image.

8. A joint injury identification system based on arthroscopic imaging, characterized in that: Applicable to a display device in communication with an arthroscopy, the system comprising: An acquisition module is used to obtain the original image of the intra-articular tissue collected by the arthroscopic method; A preprocessing module is used to perform denoising, artifact removal and deblurring on the original image in sequence to obtain a preliminary original image; An enhancement module is used to perform corresponding semantic segmentation based on the preliminary original image, so as to segment sub-regions corresponding to different tissues on the preliminary original image, and to enhance the different sub-regions using corresponding image enhancement strategies to obtain a target image; The recognition module is used to visualize the target image through a display device so as to identify joint damage of the current joint.

9. A readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. An electronic device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the steps of the method according to any one of claims 1 to 7 are implemented when the processor executes the program.

Images