Femoral head necrosis staging judgment method, system and equipment and storage medium

Abstract

The invention provides a femoral head necrosis staging judgment method, system and device and a storage medium, and the method comprises the steps: obtaining femoral head necrosis X-ray images, and carrying out the recognition of all femoral head necrosis X-ray images through a target detection model, and obtaining the recognition confidence; segmenting the femoral head necrosis X-ray image by using the target segmentation model to obtain a plurality of segmented regions, and endowing each segmented region with a region label; dividing the femoral head necrosis X-ray image into a plurality of regions of interest, and endowing all the regions of interest with corresponding region tags according to the positions of the regions of interest; according to the region labels, the segmentation confidence coefficient of each region of interest is calculated, the probability of the femoral head necrosis X-ray image in each stage is calculated according to the segmentation confidence coefficient and the recognition confidence coefficient, and the stage with the maximum probability is selected as the femoral head necrosis stage. According to the invention, the problem of low staging accuracy caused by manual staging of femoral head necrosis in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic digital data processing, and in particular to a method, system, device and storage medium for determining the stages of femoral head necrosis. Background Art

[0002] Osteonecrosis of the femoral head (AVNFH) refers to a disease in which the blood supply to the femoral head is reduced or interrupted, leading to the death of bone cells. In China, as of 2015, the number of AVNFH patients has reached about 8 million, with about 100,000 new cases each year. It is worth noting that the early asymptomatic and painless nature of AVNFH is difficult to detect, but more than 70% of asymptomatic cases can be converted to symptomatic and femoral head collapse, eventually requiring hip replacement, and the average interval between the first symptom and femoral head collapse is only 12 months. The success rate of treatment depends on the stage of starting care, so once femoral head necrosis is diagnosed, a scientific staging diagnosis should be made in a timely manner to guide the formulation of a reasonable treatment plan and accurately judge the prognosis.

[0003] The staging of osteonecrosis of the femoral head is mainly based on imaging examinations such as X-ray, CT, and MRI. Among them, X-ray examination is the most widely used due to its low price and low radiation. It is usually the preferred diagnostic imaging method for osteonecrosis of the femoral head. The current plan is that doctors manually stage X-ray data according to the Ficat staging standard (as shown in Table 1), judge the patient's condition and formulate a treatment plan based on the staging results. Therefore, accurate staging of osteonecrosis of the femoral head based on X-ray images has important guiding significance for the diagnosis and treatment of osteonecrosis of the femoral head. However, the current staging accuracy depends on the doctor's diagnostic experience, so staging is also challenging for doctors, especially for doctors with low seniority. How to improve the accuracy of staging and ensure the consistency of X-ray image staging of osteonecrosis of the femoral head between doctors with high and low seniority is a problem that needs to be solved to protect the rights and interests of patients. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a method for determining the staging of femoral head necrosis, which solves the problem of low staging accuracy caused by manual staging of femoral head necrosis in the prior art.

[0005] According to an embodiment of the present invention, a method for determining the stages of femoral head necrosis includes:

[0006] Obtain X-ray images of femoral head necrosis, use the target detection model to identify all X-ray images of femoral head necrosis, and obtain the recognition confidence;

[0007] The target segmentation model is used to segment the X-ray images of femoral head necrosis to obtain multiple segmentation regions, and each segmentation region is assigned a region label;

[0008] The X-ray image of femoral head necrosis is divided into multiple regions of interest, and corresponding regional labels are assigned to all regions of interest according to their locations;

[0009] According to the regional labels, the segmentation confidence of each region of interest is calculated, and the probability of femoral head necrosis X-ray images in each stage is calculated based on the segmentation confidence and recognition confidence, and the stage with the highest probability is selected as the femoral head necrosis stage.

[0010] Preferably, before using the target detection model and the target segmentation model, the target detection model and the target segmentation model need to be trained using a historical image dataset;

[0011] The training process of the target detection model is as follows:

[0012] Frame and mark the X-ray images of femoral head necrosis, and mark the stages of artificial femoral head necrosis;

[0013] The YOLOv8 algorithm was used to identify X-ray images of femoral head necrosis and obtain the algorithmic femoral head necrosis staging;

[0014] The artificial femoral head necrosis staging was compared with the algorithmic femoral head necrosis staging, and the YOLOv8 algorithm was tuned according to the comparison results until the artificial femoral head necrosis staging was the same as the algorithmic femoral head necrosis staging, and the target detection model was obtained.

[0015] Preferably, the training process of the target segmentation model is as follows:

[0016] Frame and mark the X-ray images of femoral head necrosis, marking all the segmented areas of the femoral head;

[0017] The U-Net segmentation algorithm is used to identify the X-ray images of femoral head necrosis and obtain the image segmentation results;

[0018] The image segmentation result is compared with the segmentation area, and the U-Net segmentation algorithm is tuned according to the comparison result until the image segmentation result is the same as the segmentation area, and the target segmentation model is obtained.

[0019] Preferably, the segmented area includes the femoral head shape area P1, the bone joint space area P2 and the femoral head lesion area P3;

[0020] The region labels of the femoral head shape region include a femoral head shape normal region s1, a femoral head shape interruption region s2, a femoral head flattened region s3, and a femoral head collapsed region s4;

[0021] The regional labels of the bone joint gap area include a normal bone joint gap area s5, a narrowed bone joint gap area s6, and a disappeared bone joint gap area s7;

[0022] The regional labels of the femoral head lesion area include the femoral head sclerosis area s8, the femoral head cystic change area s9, the femoral head half-moon sign area s10, the mild osteoporosis area s11, and the lamellar osteoporosis area s12.

[0023] Preferably, the calculation formula of the segmentation confidence is as follows:

[0024]

[0025] Among them, siPX is the pixel set of the region of interest with region label si, and ti is the weight of the region label si.

[0026] Preferably, each region of interest has one or more region labels.

[0027] When k = 1 or 2, the segmentation confidence corresponding to the largest region label is used as the segmentation confidence of the region of interest;

[0028] When k=3, the segmentation confidence corresponding to the region label that is greater than the preset value is used as the segmentation confidence of the region of interest.

[0029] Preferably, the calculation formula for calculating the probability of femoral head necrosis X-ray images in each stage according to the segmentation confidence and recognition confidence is as follows:

[0030]

[0031] Among them, P0, PI, PIIA, PIIB, PIII, and PIV are the recognition confidences of different stages obtained by the target detection model.

[0032] On the other hand, according to an embodiment of the present invention, a method for determining the staging of femoral head necrosis is also provided, wherein the system uses a method for determining the staging of femoral head necrosis as described in any one of claims 1 to 7, comprising:

[0033] A training module, wherein the training module is used to train the target detection module and the target segmentation module;

[0034] A data acquisition module, wherein the data acquisition module is used to acquire X-ray images of femoral head necrosis and pre-process the X-ray images of femoral head necrosis;

[0035] A target detection module, wherein a target recognition model is provided in the target detection module, and is used to respectively recognize the processed femoral head necrosis X-ray images to obtain recognition confidences;

[0036] A target segmentation module, wherein a target segmentation model is provided in the target segmentation module, and is used to segment the processed femoral head necrosis X-ray image to obtain a segmentation area, and divide the segmentation area into a plurality of regions of interest;

[0037] A decision module is used to calculate the segmentation confidence of each region of interest and the probability of each stage of femoral head necrosis X-ray images, and make a decision on the femoral head stage in the femoral head necrosis X-ray images.

[0038] On the other hand, according to an embodiment of the present invention, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the above-mentioned method for determining the staging of femoral head necrosis.

[0039] On the other hand, according to an embodiment of the present invention, there is also provided a computer storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the above-mentioned method for determining the staging of femoral head necrosis.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] The present invention preliminarily detects the bone necrosis X-ray image through the target detection model to obtain the preliminary staging confidence, and then adopts the target segmentation model to segment the various lesion areas in the bone necrosis X-ray image, and calculates the confidence of each area. The confidence obtained by the comprehensive target detection model and the target segmentation model is used to judge the femoral head necrosis stage in the femoral head necrosis X-ray image, thereby converting various lesion features into quantifiable and calculable features, providing an easy-to-understand and explainable feature description representation for doctors to understand and judge the model staging logic, thereby improving the accuracy of doctors' judgment on the staging of femoral head necrosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 The following is a flow chart of the installment determination according to an embodiment of the present invention.

[0043] Figure 2 4 is a diagram of the system architecture of the installment judgment according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0045] like Figure 1 As shown, the embodiment of the present invention proposes a method for determining the stages of femoral head necrosis, including:

[0046] Obtain X-ray images of femoral head necrosis, use the target detection model to identify all X-ray images of femoral head necrosis, and obtain the recognition confidence;

[0047] Before using the target detection model, it is necessary to use the X-ray images (historical image dataset) that have been used to accurately determine the stage of femoral head necrosis to train the target detection model;

[0048] The training process of the target detection model is as follows:

[0049] Frame and mark the X-ray images of femoral head necrosis, and mark the stages of artificial femoral head necrosis;

[0050] Use the YOLOv8 algorithm (including but not limited to YOLOv3, YOLOv5, YOLOx, YOLOv8, Faster RCNN and other target detection networks) to identify X-ray images of femoral head necrosis and obtain the algorithmic femoral head necrosis staging;

[0051] The artificial femoral head necrosis staging was compared with the algorithmic femoral head necrosis staging, and the YOLOv8 algorithm was tuned according to the comparison results until the artificial femoral head necrosis staging was the same as the algorithmic femoral head necrosis staging, and the target detection model was obtained. In theory, the termination node of parameter tuning is that the artificial femoral head necrosis staging is the same as the algorithmic femoral head necrosis staging, but due to the errors caused by various components themselves in the actual operation process, in the process of multiple parameter tuning, it is only necessary that the artificial femoral head necrosis staging result and the algorithmic femoral head necrosis staging result have a probability of more than 90% being the same.

[0052] After the training is completed, new X-ray images of femoral head necrosis are obtained, and the target detection model is used to identify all X-ray images of femoral head necrosis, and the following recognition results are obtained:

[0053] Label set: nVNFH, I, IIA, IIB, III, IV, respectively indicating that the femoral head staging in X-ray images of femoral head necrosis is stage 0 (preclinical stage), I (preradiological stage), IIA (necrosis formation), IIB (transitional stage), III (collapse stage), and IV (arthritis stage).

[0054] Identification confidence: P0, PI, PIIA, PIIB, PIII, PIV, corresponding to the femoral head stages nVNFH, I, IIA, IIB, III, and IV in the label set, respectively.

[0055] The target segmentation model is used to segment the X-ray images of femoral head necrosis to obtain multiple segmentation regions, and each segmentation region is assigned a region label;

[0056] Before using the target segmentation model, it is necessary to use the X-ray images (historical image dataset) that have been used to accurately determine the stage of femoral head necrosis to train the target segmentation model;

[0057] The training process of the target segmentation model is as follows:

[0058] Frame and mark the X-ray images of femoral head necrosis, marking all the segmented areas of the femoral head;

[0059] Use U-Net segmentation algorithm (including but not limited to U-Net, U-Net++, SegNet, DeepLab, RefineNet and other target segmentation algorithms) to identify X-ray images of femoral head necrosis and obtain image segmentation results;

[0060] The image segmentation result is compared with the segmentation area, and the U-Net segmentation algorithm is tuned according to the comparison result until the image segmentation result is the same as the segmentation area to obtain the target segmentation model. In theory, the termination node of parameter tuning is that the image segmentation result is the same as the segmentation area, but due to the errors caused by various components themselves in the actual operation process, in the process of multiple parameter tuning, it is only necessary for the image segmentation result to be the same as the segmentation area in more than 90% of the area.

[0061] After the training is completed, new X-ray images of femoral head necrosis are obtained, and the target segmentation model is used to identify all X-ray images of femoral head necrosis, obtain segmented regions, and assign regional labels to each segmented region:

[0062] The segmented area includes a femoral head shape area P1, a bone joint space area P2 and a femoral head lesion area P3;

[0063] The region labels of the femoral head shape region include a femoral head shape normal region s1, a femoral head shape interruption region s2, a femoral head flattened region s3, and a femoral head collapsed region s4;

[0064] The regional labels of the bone joint gap area include a normal bone joint gap area s5, a narrowed bone joint gap area s6, and a disappeared bone joint gap area s7;

[0065] The regional labels of the femoral head lesion area include the femoral head sclerosis area s8, the femoral head cystic change area s9, the femoral head half-moon sign area s10, the mild osteoporosis area s11, and the lamellar osteoporosis area s12.

[0066] Among them, P1-P3 and S1-S12 are area labels.

[0067] The X-ray image of femoral head necrosis is divided into multiple regions of interest, and corresponding regional labels are assigned to all regions of interest according to their locations;

[0068] According to the segmentation results of all X-ray images of femoral head necrosis by the target segmentation model, the segmented area is divided into multiple regions of interest. The coordinates of the center point of the segmented area are set to (xd, yd). The region width wROI = wd + f1 and the region height hROI = hd + f2 of each region of interest RIOm, where m = 1, 2, 3..., where wd is the pixel width of the mth target region, hd is the pixel height of the mth target region, and f1 and f2 are the coordinate change thresholds of the region of interest.

[0069] Since each ROI is a rectangle, some ROIs may be in multiple segmented regions at the same time and have multiple region labels.

[0070] According to the regional labels, the segmentation confidence of each region of interest is calculated, and the probability of femoral head necrosis X-ray images in each stage is calculated based on the segmentation confidence and recognition confidence, and the stage with the highest probability is selected as the femoral head necrosis stage.

[0071] According to the region label, the segmentation confidence of each region of interest is calculated according to the following formula:

[0072]

[0073] Among them, siPX is the pixel set of the region of interest with region label si, and ti is the weight of the region label si.

[0074] When k=1, t1+t2+t3+t4=1, and the segmentation confidence corresponding to the largest regional label is used as the segmentation confidence of the region of interest, and the femoral head feature description corresponding to the regional label is used as the disease description of the femoral head, and the segmentation confidence corresponding to other regional labels is set to zero;

[0075] When k=2, t5+t6+t7=1, and the segmentation confidence corresponding to the largest regional label is used as the segmentation confidence of the region of interest, and the femoral head feature description corresponding to the regional label is used as the disease description of the femoral head, and the segmentation confidence corresponding to other regional labels is set to zero;

[0076] When k=3, t8+t9+t10+t11+t12=1, the segmentation confidence corresponding to the regional label greater than the preset value is used as the segmentation confidence of the region of interest, and the femoral head feature description corresponding to the regional label is used as the disease description of the femoral head, while the segmentation confidence corresponding to other regional labels is set to zero.

[0077] Afterwards, based on the segmentation confidence calculated above and combined with the recognition confidence, the probability of each stage of the femoral head in the X-ray image of the bone necrosis stage is calculated:

[0078]

[0079] Among them, P0, PI, PIIA, PIIB, PIII, and PIV are the recognition confidences of different stages obtained by the target detection model.

[0080] Select the X-ray image with the highest probability from PFIcat0, PFIcatI, PFIcatIIA, PFIcatIIB, PFIcatIII, and PFIcatIV as the staging of bone necrosis, and describe the femoral head in the X-ray image as shown in Table 1.

[0081] Table 1: Correspondence between femoral head necrosis stages and Ficat staging characteristics

[0082]

[0083] The target detection model is used to preliminarily detect the bone necrosis X-ray image and obtain the preliminary staging confidence. Then the target segmentation model is used to segment the various lesion areas in the femoral head necrosis X-ray image, and the confidence of each area is calculated. The confidence obtained by the target detection model and the target segmentation model is combined to determine the femoral head necrosis stage in the femoral head necrosis X-ray image. In this way, the various lesion features are converted into quantifiable and computable features, which provides doctors with an easy-to-understand and explainable feature description representation to understand and judge the model staging logic, thereby improving the accuracy of doctors' judgment of bone necrosis staging.

[0084] On the other hand, Figure 2 As shown, an embodiment of the present invention further provides a femoral head necrosis staging judgment system, which uses a femoral head necrosis staging judgment method as described in any one of claims 1 to 7, including:

[0085] A training module, wherein the training module is used to train the target detection module and the target segmentation module;

[0086] A data acquisition module, wherein the data acquisition module is used to acquire X-ray images of femoral head necrosis and pre-process the X-ray images of femoral head necrosis;

[0087] A processing module, wherein the processing module is used to use a target detection module and a target segmentation module to respectively identify and segment the processed femoral head necrosis X-ray image, obtain an identification confidence and a segmentation area, and divide the segmentation area into a plurality of regions of interest;

[0088] A decision module is used to calculate the segmentation confidence of each region of interest and the probability of each stage of femoral head necrosis X-ray images, and make a decision on the femoral head stage in the femoral head necrosis X-ray images.

[0089] On the other hand, an embodiment of the present invention further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the above-mentioned method for determining the staging of femoral head necrosis.

[0090] On the other hand, an embodiment of the present invention further provides a computer storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the above-mentioned method for determining the stages of femoral head necrosis.

[0091] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A method for determining the stages of femoral head necrosis, characterized in that: include: Obtain X-ray images of femoral head necrosis, use the target detection model to identify all X-ray images of femoral head necrosis, and obtain the recognition confidence; The target segmentation model is used to segment the X-ray images of femoral head necrosis to obtain multiple segmentation regions, and each segmentation region is assigned a region label; The X-ray image of femoral head necrosis is divided into multiple regions of interest, and corresponding regional labels are assigned to all regions of interest according to their locations; According to the regional labels, the segmentation confidence of each region of interest is calculated, and the probability of femoral head necrosis X-ray images in each stage is calculated based on the segmentation confidence and recognition confidence, and the stage with the highest probability is selected as the femoral head necrosis stage.

2. A method for determining the stages of femoral head necrosis as claimed in claim 1, characterized in that: Before using the target detection model and the target segmentation model, it is necessary to train the target detection model and the target segmentation model using historical image datasets; The training process of the target detection model is as follows: Frame and mark the X-ray images of femoral head necrosis, and mark the stages of artificial femoral head necrosis; The YOLOv8 algorithm was used to identify X-ray images of femoral head necrosis and obtain the algorithmic femoral head necrosis staging; The artificial femoral head necrosis staging was compared with the algorithmic femoral head necrosis staging, and the YOLOv8 algorithm was tuned according to the comparison results until the artificial femoral head necrosis staging was the same as the algorithmic femoral head necrosis staging, and the target detection model was obtained.

3. A method for determining the stages of femoral head necrosis as claimed in claim 2, characterized in that: The training process of the target segmentation model is as follows: Frame and mark the X-ray images of femoral head necrosis, marking all the segmented areas of the femoral head; The U-Net segmentation algorithm is used to identify the X-ray images of femoral head necrosis and obtain the image segmentation results; The image segmentation result is compared with the segmentation area, and the U-Net segmentation algorithm is tuned according to the comparison result until the image segmentation result is the same as the segmentation area, and the target segmentation model is obtained.

4. A method for determining the stages of femoral head necrosis according to claim 1, characterized in that: The segmented area includes a femoral head shape area P1, a bone joint space area P2 and a femoral head lesion area P3; The region labels of the femoral head shape region include a femoral head shape normal region s1, a femoral head shape interruption region s2, a femoral head flattened region s3, and a femoral head collapsed region s4; The regional labels of the bone joint gap area include a normal bone joint gap area s5, a narrowed bone joint gap area s6, and a disappeared bone joint gap area s7; The regional labels of the femoral head lesion area include the femoral head sclerosis area s8, the femoral head cystic change area s9, the femoral head half-moon sign area s10, the mild osteoporosis area s11, and the lamellar osteoporosis area s12.

5. A method for determining the stages of femoral head necrosis as claimed in claim 4, characterized in that: The calculation formula of the segmentation confidence is as follows: Among them, siPX is the pixel set of the region of interest with region label si, and ti is the weight of the region label si.

6. A method for determining the stages of femoral head necrosis according to claim 5, characterized in that: Each region of interest has one or more region labels. When k = 1 or 2, the segmentation confidence corresponding to the largest region label is used as the segmentation confidence of the region of interest; When k=3, the segmentation confidence corresponding to the region label that is greater than the preset value is used as the segmentation confidence of the region of interest.

7. A method for determining the stages of femoral head necrosis according to claim 5, characterized in that: The calculation formula for calculating the probability of femoral head necrosis X-ray images in each stage based on segmentation confidence and recognition confidence is as follows: Among them, P0, PI, PIIA, PIIB, PIII, and PIV are the recognition confidences of different stages obtained by the target detection model.

8. A femoral head necrosis staging judgment system, characterized in that: The system uses a method for determining the stages of femoral head necrosis as described in any one of claims 1 to 7, comprising: A training module, wherein the training module is used to train the target detection module and the target segmentation module; A data acquisition module, wherein the data acquisition module is used to acquire X-ray images of femoral head necrosis and pre-process the X-ray images of femoral head necrosis; A target detection module, wherein a target recognition model is provided in the target detection module, and is used to respectively recognize the processed femoral head necrosis X-ray images to obtain recognition confidences; A target segmentation module, wherein a target segmentation model is provided in the target segmentation module, and is used to segment the processed femoral head necrosis X-ray image to obtain a segmentation area, and divide the segmentation area into a plurality of regions of interest; A decision module is used to calculate the segmentation confidence of each region of interest and the probability of each stage of femoral head necrosis X-ray images, and make a decision on the femoral head stage in the femoral head necrosis X-ray images.

9. A computer device, characterized in that: It includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor executes a method for determining the staging of femoral head necrosis as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a processor, the processor executes a method for determining the staging of femoral head necrosis as described in any one of claims 1 to 7.

Images