Bone Bank art injury auxiliary diagnosis method and system based on image recognition

Abstract

The invention discloses a bony Bank art injury auxiliary diagnosis method and system based on image recognition, and the method comprises the steps: carrying out the preprocessing and threshold segmentation of a CT image, can precisely distinguish different tissue structures in the CT image, and improves the diagnosis accuracy and intuition through the coincidence analysis of a marked region; edge pixel points are randomly selected and finely analyzed, whether the edge pixel points are normal or not is judged according to angle changes and gray values, interference of noise and artifacts is reduced, and the image is clearer. All edge pixel points are judged one by one, so that the overall improvement of the image quality is ensured, feature matching is combined, a three-dimensional model is helped to be constructed, Bank art injury recognition of the shoulder joint of the patient is helped to be completed, and powerful support is provided for formulating a treatment scheme of the patient.

Description

Technical Field

[0001] The present invention belongs to the technical field of image processing, and relates to an auxiliary diagnosis method and system for bone Bank art injury based on image recognition. Background Art

[0002] Bony Bankart injury is a common type of shoulder injury, which involves avulsion injury of the anterior and inferior part of the glenoid labrum and its attached ligament structure, usually accompanied by fracture or bone defect of the anterior and inferior part of the glenoid. This injury not only leads to instability of the shoulder joint, but also may cause a series of problems such as pain and dysfunction, seriously affecting the patient's quality of life. With the rapid development of computer technology and artificial intelligence, image recognition technology has gradually become an important auxiliary diagnostic method. When performing CT image examination on patients, the patient may have slight tremors during the CT image acquisition process due to pain, resulting in blurred images, affecting the quality of the CT image, and thus making the identification of bony Bankart injury of the shoulder joint abnormal. Summary of the invention

[0003] The purpose of the present invention is to solve the problem of inaccurate identification of bony Bankart lesions of the shoulder joint due to blurring of CT images in the prior art, and to provide an auxiliary diagnosis method and system for bony Bankart lesions based on image recognition.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions: The auxiliary diagnosis method of bone Bank art injury based on image recognition includes: Step 1, preprocessing the collected CT image of the patient's shoulder joint; Step 2, dividing the preprocessed shoulder joint CT image into regions based on threshold segmentation technology; Step 3, performing overlap analysis on the image after threshold segmentation and the original CT image, and marking each area segmented by threshold on the original CT image; Step 4, randomly selecting edge pixels of the region, and obtaining the possibility that the edge pixel is a normal edge pixel based on the angle change and grayscale value of the line connecting the edge pixel and adjacent edge pixels; Step 5, determining whether the possibility of the edge pixel point being a normal edge pixel point is greater than a preset threshold, if so, the edge pixel point is considered to be a normal pixel point; repeating steps 4 and 5 until all edge pixels are determined, and a clear CT image of the patient's shoulder joint is obtained; Step 6, performing adjacent image matching based on the acquired plurality of CT images of the patient's shoulder joint, and fusing the image information of the overlapping parts to construct a three-dimensional model of the patient's shoulder joint; Step 7: Input the patient's shoulder joint 3D model into the trained machine learning model for analysis and diagnosis to determine whether there is bony Bank art injury and the extent and location of the injury.

[0005] A further improvement of the present invention is: Furthermore, the collected CT images of the patient's shoulder joint are preprocessed, specifically: the collected shoulder joint CT images are denoised and linearly enhanced to reduce noise interference in the image and improve image clarity and contrast.

[0006] Furthermore, the preprocessed shoulder joint CT image is divided into regions based on the threshold segmentation technology. Specifically, according to the grayscale values ​​of different tissue structures in the shoulder joint CT image, the corresponding threshold is selected, and each pixel in the image is compared with the threshold to generate a binary image, thereby obtaining the regions of different tissues.

[0007] Furthermore, edge pixels of the region are randomly selected, and based on the angle change and grayscale value of the line between the edge pixel and the adjacent edge pixel, the possibility of the edge pixel being a normal edge pixel is obtained, specifically: Randomly select the edge pixels of the area and obtain the CT image of shoulder joint In the region Gray value of pixel ; Based on Region pixel points, select its adjacent edge pixel points, connect the two pixel points, and obtain the angle between the connecting line and the horizontal direction Repeat the operation to construct The angle data sequence between the outermost edge pixels of the region is obtained, and then the angle change trend of the line between adjacent edge pixels is obtained; Based on the angle change trend of the line between adjacent edge pixels and the Image The average gray value of the pixel points on the outermost edge of the region is used to determine the Image No. The outermost edge of the area The normal performance of each pixel point; then analyze the performance of the overall pixel points from the outside, Image No. The outermost edge of the area The probability that a pixel belongs to a normal edge pixel.

[0008] Further, The angle data sequence between the outermost edge pixels of the region ; Get the angle change trend of the line between adjacent edge pixels, specifically: ; The judgment Image No. The outermost edge of the area The normal performance of each pixel is as follows: ; ; in, Indicates Image The average gray value of the pixels on the outermost edge of the region; Indicates the number of outermost edge pixels selected; when the grayscale difference of a pixel is small and the adjacent angle changes are small, it means that it is likely to be a normal edge pixel; The above analysis is based on the performance of the overall pixel points from the outside. Image No. The outermost edge of the area The probability that a pixel belongs to a normal edge pixel is: ; ; in, Indicates Image The average change in angle of the pixels on the outermost edge of the region, Indicates Image The number of pixels on the outermost edge of a region whose angle difference between adjacent pixels is greater than 10°, The value of is related to the state of the pixel. If the pixels are connected, then ; If these pixels are distributed separately, then .

[0009] Further, it is determined whether the possibility of the edge pixel being a normal edge pixel is greater than a preset threshold. If so, the edge pixel is considered to be a normal pixel. This is done until all edge pixels are determined to obtain a clear CT image of the patient's shoulder joint. Specifically: Setting Thresholds ,when , it indicates that the pixel is a normal edge pixel; when judging the outermost pixel, if there is a non-normal edge pixel, it will be removed from the area, and the pixel adjacent to it and the previous pixel will be selected to replace the non-positive pixel as the area edge; recalculate whether the selected pixel is a normal edge pixel until all pixels are normal edge pixels.

[0010] Furthermore, based on the acquired CT images of the patient's shoulder joint, adjacent images are matched and the image information of the overlapping parts is fused to construct a three-dimensional model of the patient's shoulder joint. Specifically, CT image and The reliability of the matching result of the hth group of matching points on the CT image. If the reliability of the matching result is greater than the set threshold , indicating that the matching result of the matching point on the CT image is correct; if the error rate of the matching result exceeds the preset value, re-shoot the angle and repeat the above steps until the matching result is correct; then fuse the image information of the overlapping parts, and use the registered and fused image information to construct a three-dimensional model of the patient's shoulder joint.

[0011] Further, get the CT image and The reliability of the matching result of the hth group of matching points on the CT image is: ; ; in, Indicates The distance between the pixel points corresponding to the hth group of matching points on the CT image and the pixel points at the edge of the bone joint along the horizontal direction. If there are two distances, the shortest distance is selected; Indicates the The first pixel of the nine pixels centered at the pixel corresponding to the hth group of matching points on the CT image is The gray value of each pixel; express The ratio of successful matches among the eight neighboring pixels corresponding to the hth group of matching points on the CT image; CT image and The hth group of matching points on the CT image is recorded as and ; Assume that there are n pixels in the 8-neighborhood around a certain pixel to match, and there are b groups of pixels that match each other. .

[0012] Furthermore, the three-dimensional model of the patient's shoulder joint is input into the trained machine learning model for analysis and diagnosis to determine whether there is bony Bankart injury and the extent and location of the injury. Specifically, the selected machine learning algorithm is trained using three-dimensional models of known bony Bankart injury and normal bones as training data; the three-dimensional model of the patient's shoulder joint is input into the trained machine learning model for analysis and diagnosis to determine whether there is bony Bankart injury and the extent and location of the injury.

[0013] The auxiliary diagnosis system of bone Bank art injury based on image recognition includes: A preprocessing module, wherein the preprocessing module preprocesses the collected CT image of the patient's shoulder joint; A region division module, wherein the region division module divides the preprocessed shoulder joint CT image into regions based on a threshold segmentation technique; A marking module, wherein the marking module performs overlap analysis on the image after the threshold segmentation and the original CT image, and marks each area segmented by the threshold on the original CT image; An acquisition module, wherein the acquisition module randomly selects edge pixels of a region and acquires the possibility that the edge pixel is a normal edge pixel based on the angle change and grayscale value of the line connecting the edge pixel and adjacent edge pixels; A judgment module, wherein the judgment module judges whether the possibility that the edge pixel point is a normal edge pixel point is greater than a preset threshold value, and if so, the edge pixel point is considered to be a normal pixel point, until all edge pixels are judged and a clear CT image of the patient's shoulder joint is obtained; A matching module, wherein the matching module matches adjacent images based on the acquired CT images of the patient's shoulder joint, and fuses the image information of the overlapping parts to construct a three-dimensional model of the patient's shoulder joint; A determination module, wherein the determination module inputs the patient's shoulder joint three-dimensional model into a trained machine learning model for analysis and diagnosis to determine whether there is a bony Bank art injury and the extent and location of the injury.

[0014] Compared with the prior art, the present invention has the following beneficial effects: The present invention can accurately distinguish different tissue structures in CT images by preprocessing and threshold segmentation of CT images, and improve the accuracy and intuitiveness of diagnosis by overlapping analysis and marking areas; randomly select edge pixels and conduct detailed analysis, combine angle changes and grayscale values ​​to determine whether they are normal or not, reduce the interference of noise and artifacts, and make the image clearer. The one-by-one judgment of all edge pixels ensures the overall improvement of image quality, and combined with feature matching, helps to build a three-dimensional model, thereby helping to complete the identification of Bank art injuries in patients' shoulder joints; providing strong support for the formulation of patient treatment plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0016] Figure 1 It is a flow chart of the auxiliary diagnosis method of bone Bank art injury based on image recognition of the present invention; Figure 2 It is a structural schematic diagram of the image recognition-based auxiliary diagnosis system for bone Bank art injury of the present invention; Figure 3 Schematic diagram of replacing abnormal edge pixels. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0018] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0020] In the description of the embodiments of the present invention, it should be noted that if the terms "upper", "lower", "horizontal", "inner", etc. indicate an orientation or positional relationship based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the product of the invention is usually placed when in use, it is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. In addition, the terms "first", "second", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

[0021] In addition, if the term "horizontal" appears, it does not mean that the component must be absolutely horizontal, but can be slightly tilted. For example, "horizontal" only means that its direction is more horizontal than "vertical", which does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0022] In the description of the embodiments of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal connection of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0023] The present invention is further described in detail below in conjunction with the accompanying drawings: See also Figure 1 The present invention discloses an auxiliary diagnosis method for bone Bank art injury based on image recognition, comprising: S101, preprocessing the acquired CT image of the patient's shoulder joint; The acquired shoulder joint CT images are denoised and linearly enhanced to reduce noise interference in the images and improve image clarity and contrast.

[0024] S102, dividing the preprocessed shoulder joint CT image into regions based on a threshold segmentation technology; According to the grayscale values ​​of different tissue structures in the shoulder joint CT image, the corresponding threshold is selected, and each pixel in the image is compared with the threshold to generate a binary image, thereby obtaining the areas of different tissues.

[0025] S103, performing overlap analysis on the image segmented by the threshold value and the original CT image, and marking each region segmented by the threshold value on the original CT image; S104, randomly selecting an edge pixel point in the area, and obtaining the possibility that the edge pixel point is a normal edge pixel point based on the angle change and grayscale value of the line connecting the edge pixel point and adjacent edge pixel points; S104.1, randomly select edge pixels of the region and obtain CT image of shoulder joint In the region Gray value of pixel ; S104.2, based on Region pixel points, select its adjacent edge pixel points, connect the two pixel points, and obtain the angle between the connecting line and the horizontal direction Repeat the operation to construct The angle data sequence between the outermost edge pixels of the region is obtained, and then the angle change trend of the line between adjacent edge pixels is obtained; No. The angle data sequence between the outermost edge pixels of the region ; Get the angle change trend of the line between adjacent edge pixels, specifically: ; S104.3, based on the angle change trend of the line between adjacent edge pixels and the Image The average gray value of the pixel points on the outermost edge of the region is used to determine the Image No. The outermost edge of the area The normal performance of each pixel point; then analyze the performance of the overall pixel points from the outside, Image No. The outermost edge of the area The probability that a pixel belongs to a normal edge pixel.

[0026] The judgment Image No. The outermost edge of the area The normal performance of each pixel is as follows: ; ; in, Indicates Image The average gray value of the pixels on the outermost edge of the region; Indicates the number of outermost edge pixels selected; when the grayscale difference of a pixel is small and the adjacent angle changes are small, it means that it is likely to be a normal edge pixel; The above analysis is based on the performance of the overall pixel points from the outside. Image No. The outermost edge of the area The probability that a pixel belongs to a normal edge pixel is: ; ; in, Indicates Image The average change in angle of the pixels on the outermost edge of the region, Indicates Image The number of pixels on the outermost edge of a region whose angle difference between adjacent pixels is greater than 10°, The value of is related to the state of the pixel. If the pixels are connected, then ; If these pixels are distributed separately, then .

[0027] S105, determining whether the possibility of the edge pixel point being a normal edge pixel point is greater than a preset threshold, if so, the edge pixel point is considered to be a normal pixel point; repeating S104 and S105 until all edge pixels are determined, and a clear CT image of the patient's shoulder joint is obtained; Setting Thresholds ,when , it indicates that the pixel is a normal edge pixel; when judging the outermost pixel, if there is a non-normal edge pixel, it will be removed from the area, and the pixel adjacent to it and the previous pixel will be selected to replace the non-positive pixel as the area edge; recalculate whether the selected pixel is a normal edge pixel until all pixels are normal edge pixels.

[0028] S106, performing adjacent image matching based on the acquired plurality of CT images of the patient's shoulder joint, and fusing the image information of the overlapping parts to construct a three-dimensional model of the patient's shoulder joint; Get the CT image and The reliability of the matching result of the hth group of matching points on the CT image. If the reliability of the matching result is greater than the set threshold , indicating that the matching result of the matching point on the CT image is correct; if the error rate of the matching result exceeds the preset value, re-shoot the angle and repeat the above steps until the matching result is correct; then fuse the image information of the overlapping parts, and use the registered and fused image information to construct a three-dimensional model of the patient's shoulder joint.

[0029] Get the CT image and The reliability of the matching result of the hth group of matching points on the CT image is: ; ; in, Indicates The distance between the pixel points corresponding to the hth group of matching points on the CT image and the pixel points at the edge of the bone joint along the horizontal direction. If there are two distances, the shortest distance is selected; Indicates the The first pixel of the nine pixels centered at the pixel corresponding to the hth group of matching points on the CT image is The gray value of each pixel; express The ratio of successful matches among the eight neighboring pixels corresponding to the hth group of matching points on the CT image; CT image and The hth group of matching points on the CT image is recorded as and ; Assume that there are n pixels in the 8-neighborhood around a certain pixel to match, and there are b groups of pixels that match each other. .

[0030] S107, input the patient's shoulder joint three-dimensional model into the trained machine learning model for analysis and diagnosis to determine whether there is bony Bank art injury and the extent and location of the injury.

[0031] The selected machine learning algorithm is trained using three-dimensional models of known bony Bankart lesions and normal bones as training data; the patient's shoulder joint three-dimensional model is input into the trained machine learning model for analysis and diagnosis to determine whether bony Bankart lesions exist and the extent and location of the lesions.

[0032] See also Figure 2 The present invention discloses an auxiliary diagnosis system for bone Bank art injury based on image recognition, comprising: A preprocessing module, wherein the preprocessing module preprocesses the collected CT image of the patient's shoulder joint; A region division module, wherein the region division module divides the preprocessed shoulder joint CT image into regions based on a threshold segmentation technique; A marking module, wherein the marking module performs overlap analysis on the image after the threshold segmentation and the original CT image, and marks each area segmented by the threshold on the original CT image; An acquisition module, wherein the acquisition module randomly selects edge pixels of a region and acquires the possibility that the edge pixel is a normal edge pixel based on the angle change and grayscale value of the line connecting the edge pixel and adjacent edge pixels; A judgment module, wherein the judgment module judges whether the possibility that the edge pixel point is a normal edge pixel point is greater than a preset threshold value, and if so, the edge pixel point is considered to be a normal pixel point, until all edge pixels are judged and a clear CT image of the patient's shoulder joint is obtained; A matching module, wherein the matching module matches adjacent images based on the acquired CT images of the patient's shoulder joint, and fuses the image information of the overlapping parts to construct a three-dimensional model of the patient's shoulder joint; A determination module, wherein the determination module inputs the patient's shoulder joint three-dimensional model into a trained machine learning model for analysis and diagnosis to determine whether there is a bony Bank art injury and the extent and location of the injury.

[0033] Example: The present invention discloses an auxiliary diagnosis method for bone Bank art injury based on image recognition, comprising: Step 1: Use a CT scanner to acquire a CT image of the patient's shoulder joint.

[0034] The present invention is mainly used to identify and analyze whether a patient has bony Bankart lesions in the shoulder joint. CT images can clearly show the patient's shoulder labrum and the bone structure in the anterior and inferior part of the glenoid fossa, which helps doctors make early diagnosis of the patient's bone quality and helps complete the accurate identification of bony Bankart lesions and the formulation of treatment plans. The specific process is as follows: First, make sure that the CT equipment is in good condition, and then use high-resolution CT equipment to scan the patient's shoulder joint to obtain a CT image of the patient's shoulder joint. Note: When the CT scanner is scanning, the patient needs to control his or her breathing to minimize the impact of unstable breathing on the quality of the acquired CT images.

[0035] The acquired shoulder joint CT images are subjected to denoising and linear enhancement, thereby completing the preprocessing of the CT images to improve the image quality.

[0036] Step 2: Analyze the acquired image information and complete the three-dimensional reconstruction of the patient's shoulder joint.

[0037] Processing is performed based on the collected CT images of the patient's shoulder joint. When a patient has a Bankart injury to the shoulder joint, pain will occur in the shoulder joint, causing the patient (especially adolescent patients) to experience slight body tremors due to pain during the CT image scan, resulting in blurring of the collected CT images, which affects the doctor's identification of the patient's Bankart injury to the shoulder joint. Therefore, the CT image of the patient's shoulder joint is collected by a CT scanner for analysis to eliminate the blurring effect, and then further three-dimensional reconstruction is performed to help identify the bony Bankart injury of the patient's shoulder joint. Therefore, the process of helping to complete the three-dimensional reconstruction of the patient's shoulder joint is as follows: Step 2.1: Analyze the acquired CT images of the patient and complete the refined processing of the shoulder joint edge.

[0038] The analysis is performed based on the collected CT images of the patient's shoulder joint. In a normal shoulder joint CT image, the shoulder joint has a relatively clear bone structure, that is, the bone edge is clear; when the collected shoulder joint CT image is blurred, the shoulder joint edge line on the CT image is unclear, affecting the subsequent three-dimensional CT reconstruction. Therefore, based on the analysis of the CT image after threshold segmentation, whether it is a normal shoulder joint or a damaged shoulder joint, the edge curve of the bone is relatively smooth when it is not affected by the ambiguity (even if bone fragments appear, they are only individual small lines that protrude, and the overall appearance is still smooth). Based on this, the CT images of the patient's shoulder joint are analyzed to complete the fine processing of the bone edge.

[0039] The specific analysis is as follows: (1) First, the shoulder joint CT image is processed using threshold segmentation technology.

[0040] The threshold value for threshold segmentation should be set smaller. This is because the patient's shoulder CT image is blurred, which will cause the grayscale values ​​of some pixels at the edge of the shoulder joint on the CT image to be relatively dark. Therefore, a smaller threshold is set so that these pixels are included in the shoulder joint area (divided into it), and then they are distinguished and processed.

[0041] (2) Based on the CT image after threshold segmentation, select the pixel with a gray value of 1 on the image, and analyze whether there is a pixel with a gray value of 1 in its 8-neighborhood. If so, connect them. Repeat the steps until the pixel is connected to the initial pixel, and locate the area surrounded by it as area 1. Repeat the above steps to complete the regional division of the shoulder joint CT image.

[0042] Note: All connected pixels with grayscale value of 1 are divided into one area. It should be noted that there are also pixels with grayscale value of 0 inside the area connected by pixels with grayscale value of 1. These pixels are divided into the adjacent areas (i.e., the areas covering these pixels) divided above.

[0043] (3) Since the shoulder joint CT image is blurred, the edge of the shoulder joint area on the image is not accurate enough and needs to be analyzed specifically. Therefore, the present invention analyzes the area divided on the threshold segmentation CT image, as follows: (3.1) First, the image after threshold segmentation is overlapped with the original CT image, and each area segmented by the threshold is marked on the original CT image. CT image of shoulder joint In the region Gray value of pixel .

[0044] (3.2) Then along the Region Pixels (pixels on the edge) are analyzed, and the adjacent edge pixels are selected, and the two pixels are connected to obtain the angle between the connecting line and the horizontal direction. Repeat the steps to complete The angle between the outermost edge pixels of the region is calculated to construct a data sequence: ; (3.3) Calculate the angle change between adjacent edge pixels: ; (3.4) The edge of a normal human shoulder joint is relatively smooth, and the grayscale values ​​of the pixels on the edge are relatively uniform. Image No. The outermost edge of the area The normality of each pixel is determined by the pixel itself: ; ; In the formula, Indicates Image The average gray value of the pixels on the outermost edge of the region ( indicates the outermost side). Indicates the number of outermost edge pixels selected.

[0045] Logic: The above formula is mainly based on the difference in the grayscale value of the pixel points and the difference in the angle of adjacent pixel points. For the shoulder joint of the human body, its edge is relatively smooth (the bone is relatively round). As the position of the pixel point changes, its angle will change to a certain extent, but the angle of change is generally small, and the grayscale performance is relatively similar. Therefore, when the grayscale difference of a certain pixel point is small and the adjacent angle changes are small, it means that it is likely to be a normal edge pixel point.

[0046] (3.5) Further, we analyze the performance of the overall pixels on the outside to determine the Image No. The outermost edge of the area The probability that a pixel belongs to a normal edge pixel: ; ; In the formula, Indicates the image The average change in angle of the pixels on the outermost edge of the region. Indicates the image The number of pixels on the outermost edge of the region whose angle difference between adjacent pixels is greater than 10°. If these pixels are connected, then ; If these pixels are distributed separately, then .

[0047] Logic: The above formula mainly uses the pixel points calculated in step (3.4) to judge the normality of the pixel points themselves, and analyzes and calculates the pixel points as a whole. When the angle difference between the overall edge pixels is small, combined with the individual pixel points, it means that the shoulder joint edge is relatively smooth. If the angle difference between adjacent pixels is slightly larger and continuous, it may be caused by joint cracks, and it is recognized as a normal pixel point with hidden dangers; but when these pixels are distributed discontinuously, they are scattered, and they are identified as abnormal edge pixels.

[0048] (3.6) Setting thresholds , when the calculated , it means that the pixel is a normal edge pixel. Repeat the above steps to complete the judgment of the outermost pixel. When there is an abnormal edge pixel, it will be removed from the area, and the pixel adjacent to it and the previous pixel will be selected to replace the abnormal pixel as the area edge. Figure 3 shown.

[0049] (3.7) Recalculate whether the other selected pixels are normal edge pixels, until all pixels are normal edge pixels. Here, 5 layers of pixels are selected to complete the calculation.

[0050] At this point, the processing of the bone edge on the patient's shoulder joint CT image is completed.

[0051] Step 2.2, based on the above-mentioned refined CT image of the patient, further construct a three-dimensional model of the patient's shoulder joint.

[0052] The present invention completes the processing of the bone edge on the shoulder joint CT image, making the edge of the shoulder joint on the CT image clearer and more reliable, and then the multiple two-dimensional shoulder joint CT images collected are reconstructed in three dimensions, where a certain overlapping area is required between the multiple CT images. According to the above processed CT image, based on the specific information shown on the image, the adjacent images are regionally matched, and then a smooth and continuous three-dimensional model is generated to help the subsequent analysis of whether there is bony Bank art injury in the shoulder joint. The specific analysis is as follows: First, based on step 2.1, the patient CT images collected at different angles are processed to eliminate the ambiguity. Then, the feature matching algorithm is used to perform feature matching on the collected adjacent images. CT image and The first The group matching points are recorded as and .

[0053] Then analyze the CT image and The first Is the matching result of the group matching points reliable? CT image and The first Matching reliability of group matching points: ; ; In the formula, Indicates The first The distance between the pixel point corresponding to the group matching point and the pixel point at the edge of the bone joint along the horizontal direction (if there are two distances, choose the shortest distance). Indicates the The first The first of the nine pixels centered on the pixel corresponding to the group matching point The gray value of a pixel. express The first The ratio of successful matches among the eight neighboring pixels corresponding to the group matching point.

[0054] Note: Assume that there are n pixels in the 8-neighborhood around a certain pixel that are matched, and there are b groups of pixels that are matched with each other. .

[0055] Logic: Since the pixel points in the bone area on the shoulder joint CT images are similar, there may be mismatches. Therefore, based on the invariance of the positional relationship of the matching points on the two CT images, the analysis is combined with the invariance of their grayscale color. When a group of matching points on the two images have the same distance from the bone edge, and the surrounding pixel points have the same color, the matching group is consistent, indicating that the pixel matching result is correct.

[0056] Setting Thresholds , when the calculated When , it means that the matching result of the matching point on the CT image is correct. When the number of matching errors exceeds 5%, re-shoot the angle and repeat the above steps.

[0057] Based on the results of the above analysis, the image information of the overlapping parts is fused using existing technology, and a complete three-dimensional model can be constructed using the registered and fused image information.

[0058] At this point, the three-dimensional reconstruction of the patient's shoulder joint CT image has been completed.

[0059] Step 3: Based on the above three-dimensional reconstruction of the patient's shoulder joint CT image, help complete the identification of the bony Bank art injury of the patient's shoulder joint.

[0060] In the above process, the present invention completes the three-dimensional reconstruction of the patient's shoulder joint CT image and generates a three-dimensional model of the patient's shoulder joint. Further based on the three-dimensional model of the patient's shoulder joint, the existing machine learning model can be used for identification and analysis. The specific process is as follows: The selected machine learning algorithm was trained using 3D models of known bony Bankart lesions and normal bones as training data. The generated 3D model of the patient's shoulder joint was input into the trained machine learning model for analysis and diagnosis to determine whether there was a bony Bankart lesion and the extent and location of the lesion.

[0061] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An auxiliary diagnosis method for bone Bank art injury based on image recognition, characterized in that: include: Step 1, preprocessing the collected CT image of the patient's shoulder joint; Step 2, dividing the preprocessed shoulder joint CT image into regions based on threshold segmentation technology; Step 3, performing overlap analysis on the image after threshold segmentation and the original CT image, and marking each area segmented by threshold on the original CT image; Step 4, randomly selecting edge pixels of the region, and obtaining the possibility that the edge pixel is a normal edge pixel based on the angle change and grayscale value of the line connecting the edge pixel and adjacent edge pixels; Step 5, determining whether the possibility of the edge pixel point being a normal edge pixel point is greater than a preset threshold, if so, the edge pixel point is considered to be a normal pixel point; repeating steps 4 and 5 until all edge pixels are determined, and a clear CT image of the patient's shoulder joint is obtained; Step 6, performing adjacent image matching based on the acquired plurality of CT images of the patient's shoulder joint, and fusing the image information of the overlapping parts to construct a three-dimensional model of the patient's shoulder joint; Step 7: Input the patient's shoulder joint 3D model into the trained machine learning model for analysis and diagnosis to determine whether there is bony Bank art injury and the extent and location of the injury.

2. The image recognition-based auxiliary diagnosis method for bone Bank art injury according to claim 1, characterized in that: The preprocessing of the collected CT images of the patient's shoulder joint is specifically: denoising and linear enhancement of the collected CT images of the shoulder joint to reduce noise interference in the image and improve the clarity and contrast of the image.

3. The image recognition-based auxiliary diagnosis method for bone Bank art injury according to claim 2 is characterized in that: The threshold segmentation technology is used to divide the preprocessed shoulder joint CT image into regions. Specifically, according to the grayscale values ​​of different tissue structures in the shoulder joint CT image, a corresponding threshold is selected, each pixel in the image is compared with the threshold, a binary image is generated, and the regions of different tissues are obtained.

4. The image recognition-based auxiliary diagnosis method for bone Bank art injury according to claim 3 is characterized in that: The edge pixel point of the randomly selected area is obtained based on the angle change and grayscale value of the line between the edge pixel point and the adjacent edge pixel point to obtain the possibility that the edge pixel point is a normal edge pixel point, specifically: Randomly select the edge pixels of the area and obtain the CT image of shoulder joint In the region Gray value of pixel ; Based on Region pixel points, select its adjacent edge pixel points, connect the two pixel points, and obtain the angle between the connecting line and the horizontal direction Repeat the operation to construct The angle data sequence between the outermost edge pixels of the region is obtained, and then the angle change trend of the line between adjacent edge pixels is obtained; Based on the angle change trend of the line between adjacent edge pixels and the Image The average gray value of the pixel points on the outermost edge of the region is used to determine the Image No. The outermost edge of the area The normal performance of each pixel point; then analyze the performance of the overall pixel points from the outside, Image No. The outermost edge of the area The probability that a pixel belongs to a normal edge pixel.

5. The image recognition-based auxiliary diagnosis method for bone Bank art injury according to claim 4 is characterized in that: The said The angle data sequence between the outermost edge pixels of the region ; Get the angle change trend of the line between adjacent edge pixels, specifically: ； The judgment Image No. The outermost edge of the area The normal performance of each pixel is as follows: ； ； in, Indicates Image The average gray value of the pixels on the outermost edge of the region; Indicates the number of outermost edge pixels selected; when the grayscale difference of a pixel is small and the adjacent angle changes are small, it means that it is likely to be a normal edge pixel; The above analysis is based on the performance of the overall pixel points from the outside. Image No. The outermost edge of the area The probability that a pixel belongs to a normal edge pixel is: ； ； in, Indicates Image The average change in angle of the pixels on the outermost edge of the region, Indicates Image The number of pixels on the outermost edge of a region whose angle difference between adjacent pixels is greater than 10°, The value of is related to the state of the pixel. If the pixels are connected, then ; If these pixels are distributed separately, then .

6. The image recognition-based auxiliary diagnosis method for bone Bank art injury according to claim 5 is characterized in that: The possibility of judging whether the edge pixel point is a normal edge pixel point is greater than a preset threshold, and if so, the edge pixel point is considered to be a normal pixel point; until all edge pixels are judged, a clear CT image of the patient's shoulder joint is obtained, specifically: Setting Thresholds ,when , it indicates that the pixel is a normal edge pixel; when judging the outermost pixel, if there is a non-normal edge pixel, it will be removed from the area, and the pixel adjacent to it and the previous pixel will be selected to replace the non-positive pixel as the area edge; recalculate whether the selected pixel is a normal edge pixel until all pixels are normal edge pixels.

7. The image recognition-based auxiliary diagnosis method for bone Bank art injury according to claim 6 is characterized in that: The method is based on the acquired plurality of CT images of the patient's shoulder joint to perform adjacent image matching and fuse the image information of the overlapping parts to construct a three-dimensional model of the patient's shoulder joint, specifically: obtaining the first CT image and The reliability of the matching result of the hth group of matching points on the CT image. If the reliability of the matching result is greater than the set threshold , indicating that the matching result of the matching point on the CT image is correct; if the error rate of the matching result exceeds the preset value, re-shoot the angle and repeat the above steps until the matching result is correct; then fuse the image information of the overlapping parts, and use the registered and fused image information to construct a three-dimensional model of the patient's shoulder joint.

8. The image recognition-based auxiliary diagnosis method for bone Bank art injury according to claim 7 is characterized in that: The acquisition CT image and The reliability of the matching result of the hth group of matching points on the CT image is: ； ； in, Indicates The distance between the pixel points corresponding to the hth group of matching points on the CT image and the pixel points at the edge of the bone joint along the horizontal direction. If there are two distances, the shortest distance is selected; Indicates the The first pixel of the nine pixels centered at the pixel corresponding to the hth group of matching points on the CT image is The gray value of each pixel; express The ratio of successful matches among the eight neighboring pixels corresponding to the hth group of matching points on the CT image; CT image and The hth group of matching points on the CT image is recorded as and ; Assume that there are n pixels in the 8-neighborhood around a certain pixel to match, and there are b groups of pixels that match each other. .

9. The image recognition-based auxiliary diagnosis method for bone Bank art injury according to claim 8, characterized in that: The three-dimensional model of the patient's shoulder joint is input into the trained machine learning model for analysis and diagnosis to determine whether there is bony Bankart injury and the extent and location of the injury. Specifically, the three-dimensional model of the patient's shoulder joint is used as training data to train the selected machine learning algorithm; the three-dimensional model of the patient's shoulder joint is input into the trained machine learning model for analysis and diagnosis to determine whether there is bony Bankart injury and the extent and location of the injury.

10. The image recognition-based auxiliary diagnosis system for bone Bank art injury is characterized by: include: A preprocessing module, wherein the preprocessing module preprocesses the collected CT image of the patient's shoulder joint; A region division module, wherein the region division module divides the preprocessed shoulder joint CT image into regions based on a threshold segmentation technique; A marking module, wherein the marking module performs overlap analysis on the image after the threshold segmentation and the original CT image, and marks each area segmented by the threshold on the original CT image; An acquisition module, wherein the acquisition module randomly selects edge pixels of a region and acquires the possibility that the edge pixel is a normal edge pixel based on the angle change and grayscale value of the line connecting the edge pixel and adjacent edge pixels; A judgment module, wherein the judgment module judges whether the possibility that the edge pixel point is a normal edge pixel point is greater than a preset threshold value, and if so, the edge pixel point is considered to be a normal pixel point, until all edge pixels are judged and a clear CT image of the patient's shoulder joint is obtained; A matching module, wherein the matching module matches adjacent images based on the acquired CT images of the patient's shoulder joint, and fuses the image information of the overlapping parts to construct a three-dimensional model of the patient's shoulder joint; A determination module, wherein the determination module inputs the patient's shoulder joint three-dimensional model into a trained machine learning model for analysis and diagnosis to determine whether there is a bony Bank art injury and the extent and location of the injury.

Images