A method and system for judging the quality of hip joint images
By comprehensively judging various evaluation indicators of hip X-ray images, including the offset position of the small rotor coordinate, shooting position and implant detection, the problem of strong subjectivity of image quality evaluation in the existing technology is solved, and hip imaging is realized Automatic quality judgment improves the accuracy of hip lesions diagnosis and preoperative planning.
Patent Information
- Application Number
- CN202111193855.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-10-13
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2041-10-13
AI Technical Summary
The existing technology lacks automated tools to evaluate the quality of hip X-ray images, which leads to strong subjectivity in image quality evaluation and difficulty in achieving a unified quality evaluation system, which affects the accuracy of hip lesions diagnosis and preoperative planning.
By obtaining hip X-ray images, pre-processing and comprehensive judgment of multiple evaluation indicators, including judging the offset position of the small rotor coordinate, whether the shooting position is positive and whether there is an implant, using the key point model and deep learning model , establish a variety of implant detection models to achieve automatic judgment of hip image quality.
It improves the accuracy of judging hip image quality, ensures the quality control requirements for automatic hip prosthesis planning, enhances the accuracy of doctors' preoperative planning and the completeness of surgical plans, and reduces subjective deviations.
Smart Images
Figure CN113907775B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of medical imaging, and in particular to a method and system for judging the quality of hip joint images. Background Art
[0002] Standard anteroposterior hip images are of great significance for the diagnosis of hip joint diseases, preoperative planning, formulation and review of surgical plans, and are also an important form of clinical reference. In the clinical trial stage, the most important thing is the quality of the data. The differences in different data sets have a great impact on the product performance evaluation results in the clinical use stage, so it is necessary to strengthen the quality of the data itself.
[0003] At present, not all hip X-ray images are standard hip anteroposterior images. Due to the different operating levels of technicians in different hospitals, even if the same equipment is used, different technicians will present slightly different images, resulting in a wide range of data quality. At the same time, due to the lack of a unified quality evaluation system, it is too difficult to achieve image quality and format. In terms of hip image quality control, the most common and direct method of image quality evaluation is the subjective evaluation of the radiologist. Although some hip images may be usable, they have little clinical value, which can easily lead to the subjective nature of this evaluation method and prone to deviations. Therefore, it is crucial to find a method to automatically judge the quality of hip images. Summary of the invention
[0004] Based on this, the embodiments of the present application provide a method and system for judging the quality of hip joint images, which can determine whether the quality of hip joint images meets the quality control requirements of automatic planning of hip joint prostheses, thereby improving the accuracy of doctors' preoperative planning and formulating a more complete surgical plan.
[0005] In a first aspect, a method for determining the quality of a hip joint image is provided, the method comprising:
[0006] S1, obtaining an original X-ray image of a hip joint, performing preprocessing to obtain an image of a hip joint, and inputting the image of the hip joint into a left and right femur detection model to obtain an image of a left and right femur region;
[0007] S2, obtaining the coordinates of the lesser trochanter by using the key point model for the left and right femoral region images; obtaining a first evaluation result by determining the relationship between the lesser trochanter coordinate offset position and the maximum height of the hip joint X-ray image;
[0008] S3, obtaining a second evaluation result by determining whether the shooting position of the hip joint image is the positive position;
[0009] S4, obtaining a third evaluation result by determining whether there is an implant in the left and right femoral region images;
[0010] S5. Determine the hip joint image quality judgment result according to the first evaluation result, the second evaluation result and the third evaluation result.
[0011] Optionally, the obtaining of the original hip joint X-ray image and preprocessing to obtain the hip joint image includes:
[0012] The original X-ray image of the hip joint is converted into a picture format according to a certain window width and window position to obtain a jpg format image of the hip joint.
[0013] Optionally, obtaining the first evaluation result by determining the relationship between the lesser trochanter coordinate offset position and the maximum height of the hip joint X-ray image includes:
[0014] When the position 10 cm below the lesser trochanter coordinate exceeds the maximum height of the hip joint X-ray image, the first evaluation result is unqualified;
[0015] When the position 10 cm below the lesser trochanter coordinates does not exceed the maximum height of the hip joint X-ray image, the first evaluation result is qualified.
[0016] Optionally, obtaining the second evaluation result by judging whether the shooting position of the hip joint image is a positive position includes:
[0017] By reading the Image Orientation data field of the hip joint image in the dicom format;
[0018] When the value of this field is empty, the pre-trained classification model is used to determine whether the hip joint image is in the positive position. When the judgment result is negative, the second evaluation result is unqualified; when the judgment result is positive, the second evaluation result is qualified.
[0019] When the value of the field is AP, the second evaluation result is qualified, and the value of the field is AP, which is used to indicate that the shooting position of the read hip joint image is the normal position.
[0020] Optionally, obtaining a third evaluation result by judging whether there is an implant in the left and right femoral region images includes:
[0021] The coordinates of the lesser trochanter and greater trochanter of the femur of the hip joint are obtained according to the key point model, and the coordinates of the lesser trochanter and greater trochanter are used as two boundaries of the area to be inspected to determine the selected inspection area;
[0022] The selected detection area is respectively subjected to Hough line feature detection, gobar filter to extract texture features, and LeNet model to obtain a detection feature map;
[0023] The third evaluation result is obtained by manually judging the detection feature graph.
[0024] Optionally, the detecting the selected detection areas respectively by Hough line feature detection specifically includes:
[0025] The implant threshold segmentation result is obtained by performing convolution on the selected detection area to extract edge information and performing threshold segmentation with a lower threshold;
[0026] The implant threshold segmentation result is subjected to morphological corrosion, and the corroded image is subjected to Hough line feature detection.
[0027] Optionally, during the model training process, the LeNet model selects focal loss as the loss function.
[0028] Optionally, obtaining the hip joint image quality judgment result according to the first evaluation result, the second evaluation result and the third evaluation result includes:
[0029] When any one of the first evaluation result, the second evaluation result and the third evaluation result is unqualified, the hip joint image quality judgment result is unqualified.
[0030] Optionally, before inputting the hip joint image into the left and right femur detection models to obtain the left and right femur region images, the method further includes:
[0031] The left and right femur detection models are trained using unlabeled data to obtain trained left and right femur detection models.
[0032] In a second aspect, a hip joint image quality judgment system is provided, the system comprising: a preprocessing module, a first evaluation module, a second evaluation module, a third evaluation module and a determination module;
[0033] The preprocessing module is used to obtain the original hip joint X-ray image, perform preprocessing to obtain a hip joint image picture, and input the hip joint image picture into the left and right femur detection models to obtain left and right femur region images;
[0034] The first evaluation module is used to obtain the coordinates of the lesser trochanter by using the key point model for the left and right femoral region images; and obtain the first evaluation result by determining the relationship between the lesser trochanter coordinate offset position and the maximum height of the hip joint X-ray image;
[0035] The second evaluation module is used to obtain a second evaluation result by determining whether the shooting position of the hip joint image is the positive position;
[0036] The third evaluation module is used to obtain a third evaluation result by judging whether there is an implant in the left and right femoral region images;
[0037] The determination module is used to obtain the hip joint image quality judgment result based on the first evaluation result, the second evaluation result and the third evaluation result.
[0038] In the technical solution provided in the embodiment of the present application, the original hip joint X-ray image is first obtained, and the hip joint image is preprocessed to obtain the hip joint image and input into the left and right femoral detection models to obtain the left and right femoral area images; the left and right femoral area images are then subjected to the key point model to obtain the lesser trochanter coordinates; the first evaluation result is obtained by judging the relationship between the lesser trochanter coordinate offset position and the maximum height of the hip joint X-ray image; the second evaluation result is obtained by judging whether the shooting position of the hip joint image is in the positive position; the third evaluation result is obtained by judging whether there is an implant in the left and right femoral area images; finally, the hip joint image quality judgment result is determined based on the first evaluation result, the second evaluation result and the third evaluation result. It can be seen that the beneficial effects of the present invention include:
[0039] (1) For the first time, multiple different evaluation indicators were used to comprehensively judge the image quality of the hip joint. The dimensions are not limited to the location of the hip joint, but also mainly include whether there is an implant in the femoral head of the hip joint and the detection of the femoral isthmus area of the hip joint. Through these three different evaluation methods, it is comprehensively judged whether the image quality of the hip joint meets the quality control requirements of the automatic planning of the hip prosthesis.
[0040] (2) In the process of hip prosthesis implant detection, a variety of different types of implant detection models are established through the cascade model, making implant detection more accurate and robust.
[0041] (3) During the positioning of the femoral isthmus, the position of the lesser trochanter is located through a deep learning key point model, and then the isthmus region is automatically positioned, achieving more robust positioning of the femoral isthmus region. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the implementation methods or the description of the prior art. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.
[0043] Figure 1 A flowchart of a hip joint image quality judgment method provided in an embodiment of the present application.
[0044] Figure 2A schematic diagram of selecting an implant area provided in an embodiment of the present application;
[0045] Figure 3 A schematic diagram of the implant threshold segmentation result provided in an embodiment of the present application;
[0046] Figure 4 A schematic diagram of the morphological corrosion results provided in the embodiments of the present application;
[0047] Figure 5 A schematic diagram of a hip joint image quality judgment system provided in an embodiment of the present application. DETAILED DESCRIPTION
[0048] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0049] In the description of the present invention, unless otherwise specified, "plurality" means two or more. The terms "first", "second", "third", "fourth", etc. in the specification and claims of the present invention and the above-mentioned drawings are intended to distinguish the objects referred to. For schemes with a sequential flow, this terminology does not have to be understood as describing a specific order or sequence. For schemes with device structures, this terminology does not distinguish between importance, positional relationships, etc.
[0050] In addition, the terms "comprises", "has" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus that includes a series of steps or units is not necessarily limited to those steps or units that have been explicitly listed, but may also include other steps or units that are not explicitly listed but are inherent to these processes, methods, products or apparatuses, or steps or units that are added based on further optimization schemes conceived by the present invention.
[0051] At present, the quality of hip joint images in hip X-ray images is mainly measured manually, and there are no automated tools and methods. With the development of convolutional neural networks, a multi-dimensional joint evaluation method is used to determine whether the quality of hip joint images meets the quality control requirements of automatic planning of hip prostheses, thereby improving the accuracy of doctors' preoperative planning and formulating more complete surgical plans. Therefore, this method has high application value and clinical significance.
[0052] Please refer to Figure 1 , which shows a flow chart of a hip joint image quality judgment method provided in an embodiment of the present application, and the specific steps include:
[0053] S1. Obtain the original X-ray image of the hip joint, perform preprocessing to obtain a hip joint image, and input the hip joint image into the left and right femur detection models to obtain left and right femur region images.
[0054] In the embodiment of the present application, step S1 is to detect the hip joint X-ray image using a deep learning method to obtain the left and right femoral images of the hip joint, which specifically includes:
[0055] S11. Preprocessing of the original image of the coronary CTA sequence: converting the original image of the hip joint X-ray image into a picture format according to a certain window width and window position to obtain a picture of the hip joint image in jpg format;
[0056] S12. Obtaining left and right femur regions: Obtaining left and right femur detection models through unlabeled data training, and then obtaining left and right femur regions.
[0057] S2. Obtain the coordinates of the lesser trochanter from the images of the left and right femoral regions through a key point model; obtain the first evaluation result by determining the relationship between the offset position of the lesser trochanter coordinates and the maximum height of the hip joint X-ray image.
[0058] In the embodiment of the present application, the purpose of step S2 is to determine whether the narrowest part of the medullary cavity below the lesser trochanter of the left and right femurs can be located, which specifically includes:
[0059] S21. Obtaining the coordinates of the lesser rotor: obtaining the coordinates of the lesser rotor through the key point model.
[0060] S22. Obtain the medullary cavity area 10 cm below the lesser trochanter: First, obtain the position of 10 cm below the lesser trochanter according to S21, and then calculate whether the position of 10 cm below the lesser trochanter exceeds the maximum height of the X-ray image. If the position exceeds the maximum height of the hip joint X-ray image, it is considered that it does not meet the quality requirements of artificial intelligence prosthesis planning. Among them, selecting 10 cm below the lesser trochanter is the optimal choice based on the actual narrowest part of the medullary cavity.
[0061] S3. Obtain a second evaluation result by determining whether the shooting position of the hip joint image is the positive position.
[0062] In the embodiment of the present application, the purpose of step S3 is to determine whether the photographed position of the hip joint is the positive position, which specifically includes:
[0063] S31. Read the Image Orientation data field in the hip joint DICOM to determine whether it is the AP position.
[0064] S32, judging whether the photographed position of the hip joint is the AP position: if the value of this field is AP, it is judged to be the AP position. When this field is empty, it is necessary to judge according to the classification model, and use the pre-trained classification model to judge whether the hip joint is the AP position.
[0065] S4. Obtain a third evaluation result by determining whether there is an implant in the left and right femoral region images.
[0066] In the embodiment of the present application, the purpose of step S4 is to determine whether there is an implant in the femur, which specifically includes:
[0067] S41. Select implant detection area: First, obtain the coordinates of the lesser trochanter and greater trochanter of the hip joint according to the key point model, and use the coordinates of the greater and lesser trochanter as the two boundaries of the area to be detected to determine the selected detection area. In order to avoid the influence of the outer boundary of the femur on implant detection, the frame selection area can be determined and selected based on the positional relationship between the coordinates of the lesser trochanter and the greater trochanter, such as Figure 2 As shown, the large box represents the selected area obtained according to the coordinates of the greater rotor, and the small box represents the selected area obtained according to the coordinates of the lesser rotor.
[0068] S42. Detection of steel nail-shaped implants with obvious features: Extract Hough line features to detect simple implants: Figure 2 The edge information is extracted by convolution of the framed area obtained by S41 and the threshold segmentation is performed at a lower threshold. Figure 3 After segmentation, a small-scale morphological erosion is performed to obtain Figure 4 , perform Hough line detection on the eroded image. If there is an implant, the lines can be obtained. Check whether the lines are the implants we need.
[0069] S43. Detection of missed steel nail-shaped implants: For steel nail implants with unclear features, gobar filtering is used to extract texture features, and based on the extraction of some relatively weak texture information, the missed steel nail-shaped implants are further detected.
[0070] S44. Detection of implants without nails: For the implant scenario without nails, a LeNet model classifier is designed. The most important function of this model is to detect implants with less obvious or irregular texture information. During the model training process, due to the imbalance of positive and negative samples, focal loss is selected as the loss function to reduce the weight impact brought by a large number of positive samples.
[0071] S5. Determine a hip joint image quality judgment result based on the first evaluation result, the second evaluation result, and the third evaluation result.
[0072] In the embodiment of the present application, the purpose of step S4 is to comprehensively judge the quality of the hip joint image, which specifically includes:
[0073] S51. Obtain the results of the above three different judgments: make a comprehensive judgment based on the results of S2, S3, and S4 to form a specific score. If the result of S2 does not detect the isthmus area, the score is 0 points. If it can be located in the area 10 cm below, it is 10 points. The result of S3 is an anteroposterior film, then the score is 10 points, otherwise it is 0 points. If the result of S4 is that there is an implant, the hip joint image quality is 5 points. The purpose is that the image can be used for the doctor's own judgment and cannot be used for automatic judgment by artificial intelligence. If there is no implant, the score is 10 points.
[0074] S52. Comprehensively judge the hip joint image quality score. The S51 score is comprehensively judged. When any score is 0, the image quality is judged to be unqualified. When the score is 25 points, the hip joint image quality is judged to meet the standard of manual judgment by doctors, but does not meet the standard of artificial intelligence prosthesis planning. When the score is 30 points, the hip joint image quality is considered to meet the quality standard of manual judgment by doctors and the quality standard of artificial intelligence prosthesis planning.
[0075] This application determines whether the quality of hip joint X-ray images meets the quality control requirements of hip joint prosthesis automatic planning. By using a method for automatically determining the quality of hip joint images, the quality of hip joint lesion diagnosis can be improved, and the accuracy of hip joint preoperative planning can also be improved.
[0076] Please refer to Figure 5 , which shows a schematic diagram of a hip joint image quality judgment system 200 provided in an embodiment of the present application, the hip joint image quality judgment system 200 includes a preprocessing module 201, a first evaluation module 202, a second evaluation module 203, a third evaluation module 204 and a determination module 205;
[0077] The preprocessing module 201 is used to obtain the original image of the hip joint X-ray image, and perform preprocessing to obtain the hip joint image picture, and input the hip joint image picture into the left and right femur detection models to obtain the left and right femur area images;
[0078] The first evaluation module 202 is used to obtain the coordinates of the lesser trochanter from the left and right femoral region images through the key point model; and obtain the first evaluation result by determining the relationship between the offset position of the lesser trochanter coordinates and the maximum height of the hip joint X-ray image;
[0079] The second evaluation module 203 is used to obtain a second evaluation result by determining whether the shooting position of the hip joint image is the positive position;
[0080] A third evaluation module 204, configured to obtain a third evaluation result by determining whether there is an implant in the left and right femoral region images;
[0081] The determination module 205 is used to obtain a hip joint image quality judgment result according to the first evaluation result, the second evaluation result and the third evaluation result.
[0082] Each module in the above-mentioned hip joint image quality judgment system can be implemented in whole or in part by software, hardware and their combination. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.
[0083] The implementation principle and technical effects of the hip joint image quality judgment system are similar to those of the above-mentioned method embodiment. Its specific limitations can be found in the above limitations on the intelligent logistics sending and receiving method, and will not be repeated here.
[0084] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0085] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the patent application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent application shall be subject to the attached claims.
Claims
1. A method for judging the quality of hip joint images, characterized in that: The method comprises: S1, obtaining an original X-ray image of a hip joint, performing preprocessing to obtain an image of a hip joint, and inputting the image of the hip joint into a left and right femur detection model to obtain an image of a left and right femur region; S2, obtaining the coordinates of the lesser trochanter by using the key point model for the left and right femoral region images; obtaining a first evaluation result by determining the relationship between the lesser trochanter coordinate offset position and the maximum height of the hip joint X-ray image; S3, obtaining a second evaluation result by determining whether the shooting position of the hip joint image is the positive position; S4, obtaining a third evaluation result by determining whether there is an implant in the left and right femoral region images; S5, determining the hip joint image quality judgment result according to the first evaluation result, the second evaluation result and the third evaluation result; Obtaining an original X-ray image of the hip joint, and performing preprocessing to obtain an image of the hip joint, including: converting the original X-ray image of the hip joint into an image format according to a certain window width and window position, and obtaining an image of the hip joint in a jpg format; The first evaluation result is obtained by judging the relationship between the offset position of the lesser trochanter coordinates and the maximum height of the hip joint X-ray image, including: when the position 10 cm below the lesser trochanter coordinates exceeds the maximum height of the hip joint X-ray image, the first evaluation result is unqualified; when the position 10 cm below the lesser trochanter coordinates does not exceed the maximum height of the hip joint X-ray image, the first evaluation result is qualified; The second evaluation result is obtained by judging whether the shooting position of the hip joint image is the anteroposterior position, including: reading the Image Orientation data field of the hip joint image in the dicom format; when the value of the field is empty, judging whether the hip joint image is the anteroposterior position according to a pre-trained classification model, and when the judgment result is non-anteroposterior position, the second evaluation result is unqualified; when the judgment result is anteroposterior position, the second evaluation result is qualified; when the value of the field is AP, the second evaluation result is qualified, and the value of the field is AP, which is used to indicate that the shooting position of the read hip joint image is the anteroposterior position; The third evaluation result is obtained by judging whether there is an implant in the left and right femoral region images, including: obtaining the coordinates of the lesser trochanter and the greater trochanter of the hip joint femur according to the key point model, taking the coordinates of the lesser trochanter and the greater trochanter as two boundaries of the area to be inspected, and determining the selected inspection area; obtaining a detection feature map of the selected inspection area by Hough straight line feature detection, gobar filter extraction of texture features, and LeNet model; and obtaining the third evaluation result by manually judging the detection feature map; The selected detection areas are respectively subjected to Hough line feature detection, specifically comprising: extracting edge information from the selected detection areas by convolution and performing threshold segmentation with a lower threshold to obtain an implant threshold segmentation result; performing morphological corrosion on the implant threshold segmentation result, and performing Hough line feature detection on the corroded image; During the model training process, the LeNet model selects focal loss as the loss function; Obtaining the hip joint image quality judgment result according to the first evaluation result, the second evaluation result and the third evaluation result, including: when any one of the first evaluation result, the second evaluation result and the third evaluation result is unqualified, the hip joint image quality judgment result is unqualified; Before inputting the hip joint image into the left and right femur detection models to obtain the left and right femur region images, the method further includes: training the left and right femur detection models using unlabeled data to obtain trained left and right femur detection models.
2. A hip joint image quality judgment system, characterized in that: The system comprises: a preprocessing module, a first evaluation module, a second evaluation module, a third evaluation module and a determination module; The preprocessing module is used to obtain the original hip joint X-ray image, perform preprocessing to obtain a hip joint image picture, and input the hip joint image picture into the left and right femur detection models to obtain left and right femur region images; The first evaluation module is used to obtain the coordinates of the lesser trochanter by using the key point model for the left and right femoral region images; and obtain the first evaluation result by determining the relationship between the lesser trochanter coordinate offset position and the maximum height of the hip joint X-ray image; The second evaluation module is used to obtain a second evaluation result by determining whether the shooting position of the hip joint image is the positive position; The third evaluation module is used to obtain a third evaluation result by judging whether there is an implant in the left and right femoral region images; The determination module is used to obtain the hip joint image quality judgment result according to the first evaluation result, the second evaluation result and the third evaluation result; Obtaining an original X-ray image of the hip joint, and performing preprocessing to obtain an image of the hip joint, including: converting the original X-ray image of the hip joint into an image format according to a certain window width and window position, and obtaining an image of the hip joint in a jpg format; The first evaluation result is obtained by judging the relationship between the offset position of the lesser trochanter coordinates and the maximum height of the hip joint X-ray image, including: when the position 10 cm below the lesser trochanter coordinates exceeds the maximum height of the hip joint X-ray image, the first evaluation result is unqualified; when the position 10 cm below the lesser trochanter coordinates does not exceed the maximum height of the hip joint X-ray image, the first evaluation result is qualified; The second evaluation result is obtained by judging whether the shooting position of the hip joint image is the anteroposterior position, including: reading the Image Orientation data field of the hip joint image in the dicom format; when the value of the field is empty, judging whether the hip joint image is the anteroposterior position according to a pre-trained classification model, and when the judgment result is non-anteroposterior position, the second evaluation result is unqualified; when the judgment result is anteroposterior position, the second evaluation result is qualified; when the value of the field is AP, the second evaluation result is qualified, and the value of the field is AP, which is used to indicate that the shooting position of the read hip joint image is the anteroposterior position; The third evaluation result is obtained by judging whether there is an implant in the left and right femoral region images, including: obtaining the coordinates of the lesser trochanter and the greater trochanter of the hip joint femur according to the key point model, taking the coordinates of the lesser trochanter and the greater trochanter as two boundaries of the area to be inspected, and determining the selected inspection area; obtaining a detection feature map of the selected inspection area by Hough straight line feature detection, gobar filter extraction of texture features, and LeNet model; and obtaining the third evaluation result by manually judging the detection feature map; The selected detection areas are respectively subjected to Hough line feature detection, specifically comprising: extracting edge information from the selected detection areas by convolution and performing threshold segmentation with a lower threshold to obtain an implant threshold segmentation result; performing morphological corrosion on the implant threshold segmentation result, and performing Hough line feature detection on the corroded image; During the model training process, the LeNet model selects focal loss as the loss function; Obtaining the hip joint image quality judgment result according to the first evaluation result, the second evaluation result and the third evaluation result, including: when any one of the first evaluation result, the second evaluation result and the third evaluation result is unqualified, the hip joint image quality judgment result is unqualified; Before inputting the hip joint image into the left and right femur detection models to obtain the left and right femur region images, the method further includes: training the left and right femur detection models using unlabeled data to obtain trained left and right femur detection models.