WO2023160272A1

WO2023160272A1 - Deep learning-based hip replacement postoperative image evaluation method and system

Info

Publication number: WO2023160272A1
Application number: PCT/CN2023/070790
Authority: WO
Inventors: 张逸凌; 刘星宇
Original assignee: 北京长木谷医疗科技有限公司; 张逸凌
Priority date: 2022-02-24
Filing date: 2023-01-05
Publication date: 2023-08-31
Also published as: CN114742747A; CN114742747B

Abstract

The present application relates to the technical field of medicine. Provided are a deep learning-based hip replacement postoperative image evaluation method and system, which can accurately evaluate a postoperative condition of a patient having total hip replacement surgery. The method comprises: acquiring a hip joint image of a patient having undergone hip replacement surgery; on the basis of a deep learning-based object recognition network, recognizing key point positions and target areas in the hip joint image; according to the key point positions and the target areas, determining a leg length difference of both legs, an eccentric distance and a femoral prosthesis index of the patient; and according to the leg length difference of both legs, the eccentric distance and the femoral prosthesis index, evaluating the accuracy of the installation position of a femoral prosthesis of the patient. The system executes the method. In the present application, on the basis of the hip joint image of the patient having undergone hip replacement surgery, the leg length difference of both legs, the eccentric distance and the femoral prosthesis index of the patient having undergone hip replacement surgery are calculated so as to accurately evaluate a recovery condition of the patient having undergone hip replacement surgery.

Description

Image evaluation method and system after hip replacement based on deep learning

Cross References to Related Applications

This application claims the priority of the Chinese patent application with the application number 202210173937.3 and titled "Deep Learning-Based Evaluation Method and System for Imaging after Hip Joint Replacement" filed on February 24, 2022, which is fully incorporated by reference into this article.

technical field

The present application relates to the field of medical technology, in particular to a method and system for evaluating images after hip replacement surgery based on deep learning.

Background technique

Postoperative evaluation of hip replacement surgery plays a very important role in the success rate of the surgery in the medical field, so it is very important to provide accurate postoperative evaluation.

At present, the main preoperative evaluation method is manual measurement through various tools, which is inefficient and cannot guarantee accuracy. Therefore, it is urgent to provide a more convenient and accurate postoperative evaluation method.

Contents of the invention

The evaluation method and system for images after hip replacement surgery based on deep learning provided by this application are used for the above-mentioned problems in the prior art. Postoperative leg length difference, eccentricity and femoral prosthesis index of the patient, in order to achieve an accurate assessment of the recovery of the patient after hip replacement surgery.

This application provides a method for evaluating images after hip replacement surgery based on deep learning, including:

Obtaining hip images of patients after hip replacement surgery;

A target recognition network based on deep learning, identifying key point positions and target areas in the hip joint image;

According to the position of the key point and the target area, determine the patient's leg length difference, eccentricity and femoral prosthesis index;

According to the leg length difference of the two legs, the eccentric distance and the femoral prosthesis index, the accuracy of the installation of the patient's femoral prosthesis position is evaluated;

Wherein, the installation accuracy of the position of the femoral prosthesis is used to evaluate the postoperative recovery of the patient.

According to a deep learning-based image evaluation method after hip replacement surgery provided by the present application, the target recognition network is trained based on a point recognition neural network and a segmentation neural network; or,

It is trained based on the preset neural network model including stacked hourglass network structure, segmented Segment-Head network and keypoint Keypoint-Head network.

According to a deep learning-based image evaluation method after hip arthroplasty provided by this application, the

A target recognition network based on deep learning to identify key point positions and target areas in the hip joint image, including:

Input the hip joint image into the target recognition network to determine the first lower edge point, the second lower edge point, and the first tear on both sides of the ischium region corresponding to the bilateral lesser trochanter in the hip joint image Drop point, second tear drop point, pubic symphysis point, femoral prosthesis ball head area, healthy side femoral head area, bilateral cortical bone area and ischium area;

Determining the first lower edge point and the second lower edge point as the first key point position, and determining the first teardrop point and the second teardrop point as the second key point respectively The position and the pubic symphysis point are determined as the third key point position;

determining the first key point position, the second key point position, and the third key point position as the key point position;

The area of the ball head of the femoral prosthesis, the area of the femoral head on the healthy side, the area of the bilateral cortical bone, and the area of the ischium are determined as the target area.

According to a deep learning-based image evaluation method after hip arthroplasty provided by the present application, the determination of the patient's leg length difference according to the position of the key point and the target area includes:

Determine the leg length difference between the two legs according to the first key point position and the ischial tuberosity line; or

According to the position of the first key point and the line connecting the teardrop points on both sides, determine the leg length difference between the two legs;

Wherein, the ischial tuberosity line is determined according to the bilateral first and second lowest points of the ischial region;

The line connecting the bilateral teardrop points is determined according to the position of the second key point.

According to a deep learning-based image evaluation method after hip arthroplasty provided by the present application, the determination of the leg length difference between the two legs according to the position of the first key point and the ischial tuberosity line includes:

determining a first shortest distance between the first inferior border point and the ischial tuberosity line;

determining a second shortest distance between the second inferior border point and the ischial tuberosity line;

The leg length difference between the two legs is determined according to the difference between the first shortest distance and the second shortest distance.

According to a deep learning-based image evaluation method after hip replacement provided by the present application, the leg length difference between the two legs is determined according to the position of the first key point and the line connecting the teardrop points on both sides, include:

determining the third shortest distance between the first lower edge point and the line connecting the bilateral teardrop points;

determining the fourth shortest distance between the second lower edge point and the line connecting the bilateral teardrop points;

The leg length difference between the two legs is determined according to the difference between the third shortest distance and the fourth shortest distance.

According to a deep learning-based image evaluation method after hip arthroplasty provided by the present application, the determination of the patient's eccentricity according to the position of the key point and the target area includes:

According to the bilateral cortical bone area, determine the first femoral medullary canal centerline on the same side as the femoral prosthesis ball head area and the second femoral medullary canal centerline on the same side as the healthy side femoral head area;

determining the fifth shortest distance between the first center of rotation of the femoral prosthesis ball head area and the centerline of the first femoral canal;

Determine the sixth shortest distance between the second center of rotation of the femoral head region on the healthy side and the centerline of the second femoral canal;

determining the femoral offset according to the difference between the fifth shortest distance and the sixth shortest distance;

Wherein, the eccentricity includes the femoral eccentricity.

According to a deep learning-based image evaluation method after hip arthroplasty provided by the present application, the determination of the patient's eccentricity according to the position of the key point and the target area further includes:

Determine the acetabular cup eccentricity according to the first center of rotation of the femoral prosthesis ball head area, the second center of rotation of the healthy side femoral head area, the ischial tuberosity line and the pelvic central axis; or

Determine the eccentricity of the acetabular cup according to the first center of rotation, the second center of rotation, the line connecting the bilateral teardrop points and the central axis of the pelvis;

Wherein, the central axis of the pelvis is determined according to the position of the third key point and the ischial tuberosity line;

The eccentricity includes the acetabular cup eccentricity.

According to a deep learning-based image evaluation method after hip arthroplasty provided by the present application, the first rotation center point of the ball head area of the femoral prosthesis and the second rotation of the femoral head area of the healthy side The center point, the ischial tuberosity line, and the midline of the pelvis determine the acetabular cup offset, including:

determining a seventh shortest distance between the first center of rotation and the ischial tuberosity line;

determining an eighth shortest distance between the second center of rotation and the ischial tuberosity line;

determining a ninth shortest distance between the first center of rotation and the central axis of the pelvis;

determining a tenth shortest distance between the second center of rotation and the central pelvic axis;

The acetabular cup eccentricity is determined according to the difference between the seventh shortest distance and the eighth shortest distance and the difference between the ninth shortest distance and the tenth shortest distance.

According to a deep learning-based image evaluation method after hip joint replacement provided by the present application, according to the first rotation center point, the second rotation center point, the line connecting the teardrop points on both sides and the The central axis of the pelvis is used to determine the eccentricity of the acetabular cup, including:

determining the eleventh shortest distance between the first rotation center point and the line connecting the bilateral teardrop points;

determining the twelfth shortest distance between the second rotation center point and the line connecting the bilateral teardrop points;

determining a thirteenth shortest distance between the first center of rotation and the central pelvic axis;

determining a fourteenth shortest distance between the second center of rotation and the central pelvic axis;

Determining the acetabular cup eccentricity based on the difference between the eleventh shortest distance and the twelfth shortest distance and the difference between the thirteenth shortest distance and the fourteenth shortest distance distance.

According to a deep learning-based image evaluation method after hip arthroplasty provided by the present application, the determination of the patient's femoral prosthesis index according to the position of the key point and the target area includes:

According to the two outer diameter vertices of the femoral prosthesis in the ball head area of the femoral prosthesis, the two junction points of the femoral prosthesis and the ball head area of the femoral prosthesis and the ischial tubercle line, determine the femoral prosthesis body anteversion and abduction angles;

According to the anteversion angle and the abduction angle, the femoral prosthesis index of the patient is determined.

The present application also provides a deep learning-based image evaluation system after hip replacement surgery, including: an acquisition module, an identification module, a determination module, and an evaluation module;

The acquiring module is configured to acquire hip joint images of patients after hip joint replacement surgery;

The identification module is configured as a target recognition network based on deep learning to identify key point positions and target areas in the hip joint image;

The determination module is configured to determine the patient's leg length difference, eccentricity and femoral prosthesis index according to the key point position and the target area;

The evaluation module is configured to evaluate the installation accuracy of the patient's femoral prosthesis position according to the leg length difference between the two legs, the eccentric distance and the femoral prosthesis index;

The present application also provides an electronic device, including a processor and a memory storing a computer program. When the processor executes the program, the method for evaluating images after hip joint replacement based on deep learning as described above is implemented. .

The present application also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the image evaluation after hip joint replacement based on deep learning as described above can be realized. method.

The present application also provides a computer program product, including a computer program. When the computer program is executed by a processor, any one of the methods for evaluating images after hip arthroplasty based on deep learning described above can be implemented.

The image evaluation method and system after hip replacement surgery based on deep learning provided by this application, based on the hip joint image of the patient after hip joint replacement surgery, calculates the difference in leg length and eccentricity of the patient's legs after hip joint replacement surgery Calcar and femoral component metrics for accurate assessment of recovery in patients undergoing hip replacement surgery.

Description of drawings

In order to more clearly illustrate the technical solutions in this application or the prior art, the accompanying drawings that need to be used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings in the following description are the present For some embodiments of the application, those skilled in the art can also obtain other drawings based on these drawings without creative work.

Fig. 1 is a schematic flowchart of the image evaluation method after hip arthroplasty based on deep learning provided by the present application;

Fig. 2 is a schematic diagram of the lower edge of the bilateral femoral lesser trochanter in the recognized image of the hip joint provided by the present application;

Fig. 3 is a schematic diagram of the ischium region in the hip joint image provided by the present application;

Fig. 4 is a schematic diagram of bilateral teardrop points in the identified hip joint image provided by the present application;

Fig. 5 is a schematic diagram of the pubic symphysis point identified in the image of the hip joint provided by the present application;

FIG. 6 is a schematic structural diagram of a preset neural network model provided by the present application;

Fig. 7 is a schematic structural diagram of the target recognition network provided by the present application;

Fig. 8 is a schematic diagram of the position of the first lowest point on both sides of the ischium region in the hip joint image provided by the present application;

Fig. 9 is a schematic diagram of the ischial tuberosity line in the hip joint image provided by the present application;

Fig. 10 is a schematic diagram of the line of bilateral teardrop points in the hip joint image provided by the present application;

Figure 11 is one of the schematic diagrams for determining the leg length difference provided by the present application;

Fig. 12 is the second schematic diagram of determining the leg length difference provided by the present application;

Fig. 13 is a schematic diagram of the centerline of the bilateral femoral medullary cavity in the image of the hip joint provided by the present application;

Fig. 14 is the schematic diagram of the first center of rotation of the femoral prosthesis ball head area provided by the present application;

Fig. 15 is one of the schematic diagrams of determining the femoral eccentricity provided by the present application;

Fig. 16 is the second schematic diagram of determining the femoral eccentricity provided by the present application;

Fig. 17 is a schematic diagram of the central axis of the pelvis in the hip joint image provided by the present application;

Figure 18 is one of the schematic diagrams for determining the eccentricity of the acetabular cup provided by the present application;

Fig. 19 is the second schematic diagram of determining the eccentricity of the acetabular cup provided by the present application;

Fig. 20 is a schematic diagram of the outer diameter vertex and junction point in the hip joint image provided by the present application;

Fig. 21 is a schematic diagram of a fitting ellipse provided by the present application;

Fig. 22 is the schematic diagram of the abduction angle of the femoral prosthesis provided by the present application;

Fig. 23 is a schematic structural diagram of the image evaluation system after hip replacement based on deep learning provided by the present application;

FIG. 24 is a schematic diagram of the physical structure of the electronic device provided by the present application.

Detailed ways

In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below in conjunction with the accompanying drawings in this application. Obviously, the described embodiments are part of the embodiments of this application , but not all examples. Based on the embodiments in this application, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of this application.

Figure 1 is a schematic flowchart of the deep learning-based image evaluation method after hip replacement surgery provided by this application. As shown in Figure 1, the method includes:

S1. Acquiring hip joint images of patients after hip joint replacement surgery;

S2. Target recognition network based on deep learning to identify the key point position and target area in the hip joint image;

S3. Determine the patient's leg length difference, eccentricity and femoral prosthesis index according to the key point position and the target area;

S4. Evaluate the accuracy of the installation of the patient's femoral prosthesis position according to the leg length difference, eccentric distance and femoral prosthesis index;

Among them, the accuracy of the installation of the femoral prosthesis is used to evaluate the postoperative recovery of the patient.

It should be noted that, the execution subject of the above method may be a computer device.

Optionally, after the hip replacement operation, the doctor will perform postoperative evaluation on the patient based on the hip joint image of the patient after the hip joint replacement operation, by analyzing the key points in the hip joint image of the patient after the hip joint replacement operation The location and target area are identified to evaluate the recovery of patients after hip replacement surgery.

First of all, the hip joint image of the patient after the hip joint replacement operation is obtained, specifically, X-ray film, computer tomography (Computed Tomography, CT) or magnetic Magnetic Resonance Imaging (MRI) acquired hip images of the patient's hip.

Secondly, identify the key points and target areas of the obtained hip image of the patient after hip replacement surgery, and find the key point positions and target areas in the hip image for postoperative evaluation. For example, the hip Joint images are input to a pre-trained object recognition network to identify keypoint locations and object regions.

Thirdly, according to the key point positions and target areas identified above, the patient's postoperative leg length difference, the patient's eccentricity and the patient's femoral prosthesis index are determined.

Finally, the accuracy of the installation of the patient's femoral prosthesis is evaluated by using the obtained patient's leg length difference after hip replacement surgery, the patient's femoral eccentricity and the patient's femoral prosthesis index, in order to achieve Accurately assess the recovery of patients after hip replacement surgery. The image evaluation method after hip replacement surgery based on deep learning provided by this application is based on the hip joint images of patients after hip joint replacement surgery, and calculates the difference in leg length, eccentricity and Femoral prosthesis metrics to enable accurate assessment of recovery in patients undergoing hip replacement surgery.

Further, in one embodiment, the target recognition network is trained based on the point recognition neural network and the segmentation neural network; or,

Further, in one embodiment, step S2 may specifically include:

S21. Input the hip joint image into the target recognition network to determine the first lower edge point, the second lower edge point, and the first teardrop point on both sides of the ischium region corresponding to the bilateral lesser trochanter in the hip joint image position, the second teardrop point, the pubic symphysis point, the femoral prosthesis ball head area, the contralateral femoral head area, the bilateral cortical bone area and the ischial area;

S22. Determine the first lower border point and the second lower border point as the first key point position, the first tear drop point and the second tear drop point as the second key point position and the pubic symphysis point respectively Determined as the third key point position;

S23. Determine the position of the key point according to the position of the first key point, the position of the second key point and the position of the third key point;

S24. Determine the area of the ball head of the femoral prosthesis, the area of the uninjured femoral head, the area of the bilateral cortical bone, and the area of the ischium as target areas.

Optionally, as shown in Fig. 2-Fig. 5, the hip joint images are input into the pre-trained target recognition network to identify the bilateral femoral lesser trochanters in the hip joint images of patients after hip joint replacement surgery respectively corresponding to The positions of the first inferior border point and the second inferior border point (that is, the lower border points of the lesser trochanter on both sides, such as the first inferior border point A1 and the second inferior border point A2 in Figure 2), bilateral The first teardrop point, the second teardrop point (that is, the bilateral teardrop point, such as the first teardrop point D1 and the second teardrop point D2 in Figure 4), the pubic symphysis point (such as G point shown in Figure 5) and the target area (including the ischium area shown in Figure 3, the femoral prosthesis ball head area, the healthy side femoral head area and the bilateral cortical bone area), wherein the target recognition network can be specifically composed of points It can be trained by identifying neural network and segmenting neural network, or it can be trained by preset neural network model (including stacked hourglass network structure (Stacked Hourglass Networks, SHM), segmented Segment-Head network and keypoint Keypoint-Head network) .

Specifically, the midpoint recognition neural network of the target recognition network can be used to recognize the position of the lower edge of the bilateral lesser trochanter and the position of the bilateral teardrop in the image of the patient's hip joint marked in advance, so as to obtain the position of the patient's surgery. The bilateral femoral lesser trochanter in the image of the posterior hip joint corresponds to the first inferior edge point, the second inferior edge point, the first teardrop point and the second teardrop point respectively; and using the segmentation in the target recognition network The neural network converts the hip image of the patient after hip replacement surgery into a 0-255 grayscale image, and classifies each pixel of the image. For example, each pixel of the image can be classified according to the ischium area and the background area Category classification is carried out to determine the ischial area in the hip joint images of patients after hip replacement surgery. The identification method is the same, which is not specifically limited in this application.

Among them, the point recognition neural network can be specifically target positioning network LocNet, image segmentation network SegNet, regional convolutional neural network R-CNN, fast regional convolutional neural network Fast R-CNN, regional full convolutional neural network R-FCN, and target Detect network SSD.

Among them, the segmentation neural network can specifically be a full convolutional neural network FCN, SegNet, a hole convolutional neural network, an efficient neural network ENet, and an instance segmentation network DeepMask.

Train the preset neural network model to obtain the target recognition network. The specific steps are as follows:

First, obtain a dataset of hip images of patients after hip replacement surgery;

Secondly, input the hip joint image data set into the preset neural network model for training, and determine the model output result;

Finally, adjust the parameters of the preset neural network model based on the output result and loss function until the trained deep learning model is determined;

Wherein, the loss function is determined based on the loss function and the first weight corresponding to the segmented Segment-Head network, and the loss function and the second weight corresponding to the keypoint Keypoint-Head network.

It can be understood that before acquiring the hip joint image data set, the collected hip joint images of patients after hip joint replacement surgery can be preprocessed. The image format can be a digital imaging and communication in medicine (Digital Imaging and Communications in Medicine, DICOM) format file.

In actual implementation, the image format of the hip joint image of the patient after the hip joint replacement operation is first converted into JPG format, and the converted image will have problems of different sizes and diverse contrasts.

For the problem of different sizes, if the image is directly scaled to the target pixel, the image will be deformed and the subsequent measurement will be inaccurate. Therefore, the following method can be used to deal with it: align the image with the ratio of the longer side of the image to the target pixel Perform proportional scaling, and then perform zero padding on the scaled image to avoid deformation of the converted image. Wherein, the target pixel can be set to 512×512 pixels.

For the problem of diversification of contrast, the following methods can be used to deal with it:

1. According to the distribution of the pixel values of each image, do mean value processing. Then, threshold screening is performed on all images, and contrast enhancement operation is performed on the screened images with abnormal contrast, so that all images are in the same contrast range.

2. Diversify the contrast of the image through gamma transformation, and increase the various scenes of the data to adapt to the scene of unknown contrast.

The above-mentioned image processing methods can increase image definition and reduce noise. Of course, in other embodiments, the image processing method can also be expressed in other forms, including but not limited to using Laplacian operator for image enhancement or image enhancement based on object Log transformation, etc., which can be determined according to actual needs. The application does not specifically limit this.

For non-DICOM format pictures, deep learning is used to calibrate the scale of the entire hip joint image according to the reference scale on the hip joint image to ensure the accuracy of subsequent measurement data. , for a hip joint image with a scale, the hip joint image can be corrected directly by referring to a scale of known size. For hip images without scales, the hip image can be corrected with reference to the known cup diameter.

Optionally, after the preprocessing operation is completed, a hip joint image data set of patients after hip joint replacement surgery can be obtained. The dataset consists of two parts: keypoint location and region segmentation. The key point positions include five key points in each hip joint image, namely, the first inferior border point, the second inferior border point corresponding to the bilateral lesser trochanter, the first teardrop point on both sides, the second The teardrop point and the pubic symphysis point; the region segmentation refers to the target segmentation region is the femoral prosthesis ball head region, the contralateral femoral head region, the bilateral cortical bone region and the ischial region. When training the preset neural network model, it is necessary to continuously iterate the training result and the real value to reduce the error and improve the prediction accuracy. Therefore, before model training, the hip joint image data set can be divided into training set, verification set and test set according to the target ratio. For example, the target ratio of training set, validation set and test set can be set as 6:2:2.

Specifically, a deep learning model is built according to different neural network structures, and the training set is input to a preset neural network model for training until each neural network converges to obtain an initial neural network model. The initial neural network model is optimized according to the test set, the optimal neural network model after training is obtained, and the weight parameters of the optimal neural network model are determined. Then input the verification set into the trained optimal neural network model for verification, and verify the output result of the optimal neural network model. During the training process, the multi-weight loss function is used for error calculation, and the back propagation algorithm is used to continuously update the weight parameters of the model until the preset neural network model reaches the expected goal, and finally completes the training.

Optionally, the loss function in this application includes two parts, which respectively correspond to the position of the key point and the error corresponding to the region segmentation result. In order to improve the prediction accuracy of the preset neural network model, during the training process, the weight changes of the error function corresponding to the key point position and the error function corresponding to the region segmentation are observed until the errors of the two can be balanced.

Among them, the loss function corresponds to two different neural network structures and different weights.

In actual implementation, as shown in FIG. 6 , the network structure of the preset neural network model may include SHM network, Segment-Head network and Keypoint-Head network. The preset neural network model uses the Adam optimizer. Adam combines the advantages of the adaptive learning rate gradient descent algorithm (Adagrad) and the momentum gradient descent algorithm, which can not only adapt to sparse gradients (ie, natural language and computer vision problems), but also ease the gradient Concussion problem.

The loss function of the preset neural network model corresponds to the two heads, and the loss function of Keypoint-Head is the mean absolute value error (MAE), which is the average of the absolute value of the difference between all network prediction points and the corresponding points in the gold standard. The loss function of Segment-Head is the Dice coefficient + BCEloss loss function. The overall loss function is aMAE+b(Dice+BCEloss), where a is the first weight and b is the second weight, which can balance the error between key points and region segmentation.

The preset neural network model is evaluated by the following indicators: the evaluation indicator of Keypoints refers to the evaluation indicator oks of the key points of the human body, and the evaluation indicator of Segment is the Dice coefficient.

After obtaining the target neural network model, the SHM network and Segment-Head network based on the target neural network model identify the target area in the hip joint image of the patient after hip replacement surgery. The target area is the sciatic area as an example for detailed description ,specifically:

As shown in Figure 7, the Hourglass structure is a classic encoding Encoder-decoding Decoder structure. The Encoder structure is composed of convolution and pooling. The Decoder is composed of deconvolution and convolution. After the first feature is extracted through the SHM network, the Keypoint-Head Share the feature extraction layer with Segment-Head, and further extract the second feature through two convolutions on this basis, and finally change the number of channels through 1×1 convolution, the output is the logits layer, and Segment-Head passes the logits layer Do softmax normalization, and extract the area corresponding to the maximum probability value as the final segmentation result, that is, the ischial area.

The identified femoral prosthesis ball head area, uninjured femoral head area, bilateral cortical bone area, and ischium area were determined as target areas.

Based on the SHM network and the Keypoint-Head network, the key point position in the hip joint image of the patient after hip replacement surgery is identified, specifically:

As shown in Figure 7, after the first feature is extracted through the SHM network, Keypoint-Head and Segment-Head share the feature extraction layer, and on this basis, the third feature is further extracted through two convolutions, and finally through 1× 1 Convolution changes the number of channels, and the output is the logits layer. Keypoint-Head generates a thermal heatmap, and uses the point with the maximum probability value in the heatmap as a feature point, that is, a key point, specifically including the position of the first key point determined by the first lower edge point and the second lower edge point, and by The position of the second key point determined by the position of the first teardrop and the second position of the teardrop, and the position of the third key point determined by the position of the pubic symphysis.

The deep learning-based image evaluation method after hip replacement surgery provided by this application, combined with the deep learning method, evaluates the accuracy of the installation position of the femoral prosthesis in patients undergoing hip replacement surgery, so as to realize the accuracy of hip replacement surgery. Rapid and accurate assessment of postoperative recovery in patients undergoing replacement surgery.

Further, in one embodiment, step S3 may specifically include:

S30. Determine the leg length difference between the legs according to the position of the first key point and the ischial tuberosity line; or

S31. According to the position of the first key point and the line connecting the teardrop points on both sides, determine the leg length difference between the two legs;

The line connecting the teardrop points on both sides is determined according to the position of the second key point.

Optionally, after identifying the ischial region of the hip joint image of the patient after hip replacement surgery, the ischial tuberosity line is obtained by determining the bilateral first lowest point and the second lowest point of the ischial region, specifically :

Using image processing technology to extract the lowest point of bilateral ischia from the segmented ischial area, that is, take the lowest point of the bilateral ischial area, assuming it is the first lowest point, and draw a horizontal straight line along the first lowest point, as shown in Figure 8 Show.

Then rotate the horizontal straight line obtained above around the first lowest point (rotate counterclockwise when the lowest point is on the left side, and clockwise when it is on the right side), until the second intersection point is generated with the ischium area, the second The intersection point is the second lowest point, specifically as shown in FIG. 9 , the ischial tuberosity line CD is obtained by connecting the first lowest point with the intersection point.

Alternatively, after the ischium region is acquired, a set of ischial edge points of the ischia region is determined. And automatically scan each row of pixels in the sciatic region. The scanning method is as follows:

Step 1. Use the horizontal scanning line to scan upwards from the bottom of the ischial area, and judge whether the scanning line passes through the pixel points on the edge of the ischial for each row of pixels. In a case where it is determined that the scanning line passes the first pixel corresponding to the edge of the ischia for the first time, the scanning line stops moving up. Or when it is judged that the point on the scanning line exists in the set of ischial edge points, the scanning line stops moving up, and the first pixel point is determined, and the first pixel point is assumed to be the first lowest point.

Step 2. Taking the first pixel as the center of rotation, it is judged whether the scan line passes through the pixel on the edge of the ischia every time one degree of rotation is performed. In a case where it is determined that the scanning line passes through the second pixel point corresponding to the edge of the ischium for the first time, the scanning line stops rotating. Or when it is judged that the point on the scanning line exists in the point concentration of the edge of the ischia, the scanning line stops moving up, and the second pixel point is determined, and the second pixel point is the second lowest point.

Step 3. Determine the line connecting the first pixel point and the second pixel point as the ischial tuberosity line CD.

According to the position of the first key point corresponding to the bilateral femoral lesser trochanter and the ischial tuberosity line CD in the hip joint image of the patient after hip joint replacement surgery identified above, the leg length difference between the legs of the patient is determined.

Alternatively, according to the position of the first key point corresponding to the position of the first key point of the bilateral femoral lesser trochanter in the hip joint image of the patient after hip joint replacement surgery identified above and the line ab of the bilateral tear drop points, determine the patient's legs Length difference, wherein, the line ab of the bilateral teardrop point is obtained after connecting the first teardrop point and the second teardrop point, see FIG. 10 .

The application provides a deep learning-based image evaluation method after hip replacement surgery, using deep learning methods to identify the corresponding key points and target areas in the hip joint images of patients after hip joint replacement surgery and calculate the leg length difference, It laid the foundation for the subsequent rapid assessment of the postoperative recovery of patients undergoing hip replacement surgery based on the leg length difference.

Further, in one embodiment, step S30 may specifically include:

S301. Determine the first shortest distance between the first inferior border point and the ischial tuberosity line;

S302. Determine the second shortest distance between the second inferior border point and the ischial tuberosity line;

S303. Determine the leg length difference between the two legs according to the difference between the first shortest distance and the second shortest distance.

Optionally, as shown in FIG. 11 , it is assumed that the positions of the lower edge of the bilateral femoral lesser trochanter in the image of the identified hip joint of the patient after hip joint replacement surgery are A1 and A2 respectively and the ischial tuberosity line is cd.

Then draw a vertical line from the points A1 and A2 at the lower edge of the lesser trochanter to the ischial tuberosity line CD respectively to obtain the first line segment A1a1 and the second line segment A2a2. The distance between the first line segment is also the first The first shortest distance between the edge point A1 and the ischial tuberosity line, and the distance between the second line segment is the second shortest distance between the second lower edge point A2 and the ischial tuberosity line.

According to the distance between the first line segment A1a1 and the second line segment A2a2, calculate the difference between the first line segment A1a1 and the second line segment A2a2, and use the absolute value of the difference as the actual length difference of the lower limbs of the patient, that is The leg length difference between the two legs can be used to judge the recovery of the leg length of the lower limbs of patients after joint replacement surgery. Among them, the length values of A1a1 and A2a2 are positive if they are below the ischial tuberosity line CD, and negative if they are above .

The application provides a deep learning-based image evaluation method after hip replacement surgery, using the deep learning method to analyze the corresponding key points (the first lower edge point and the second edge point) in the hip joint image of the patient's hip joint after hip joint replacement surgery. two lower edge points) and the target area to identify and calculate the leg length difference, so as to realize the rapid assessment of postoperative recovery of patients undergoing total hip replacement surgery.

Further, in one embodiment, step S31 may specifically include:

S311. Determine the third shortest distance between the first lower edge point and the line connecting the bilateral teardrop points;

S312. Determine the fourth shortest distance between the second lower edge point and the line connecting the bilateral teardrop points;

S313. Determine the leg length difference between the two legs according to the difference between the third shortest distance and the fourth shortest distance.

Optionally, as shown in FIG. 12 , it is assumed that the positions of the lower edge of the bilateral femoral lesser trochanter in the image of the identified hip joint of the patient after hip joint replacement surgery are A1 and A2 respectively, and the first teardrop After the point D1 is connected with the second teardrop point D2, a bilateral teardrop point connection line ab is obtained.

Then draw a vertical line from the points A1 and A2 at the lower edge of the lesser trochanter on both sides to the line ab connecting the teardrop points on both sides to get the third line segment A1b1 and the fourth line segment A2b2. The distance between the third line segment is also the first The third shortest distance between the lower edge point A1 and the line d between the teardrop points on both sides, and the distance between the fourth line segment is the connection line ab between the second lower edge point A2 and the teardrop points on both sides The fourth shortest distance between .

According to the distance between the third line segment A1b1 and the fourth line segment A2b2, calculate the difference between the third line segment A1b1 and the fourth line segment A2b2, and use the absolute value of the difference as the actual length difference of the lower limbs of the patient, that is, the two legs Leg length difference, which can be used to judge the recovery of lower extremity leg length of patients after joint replacement surgery.

According to the difference in the lengths of the legs and legs of patients after hip replacement surgery obtained above, the postoperative recovery of patients undergoing hip replacement surgery is evaluated. If the difference in leg length is within the preset range (for example, less than 3mm), it is determined Patients undergoing total hip replacement surgery recover well after surgery.

The application provides a deep learning-based image evaluation method after hip replacement surgery, using deep learning methods to analyze the corresponding key points in the hip joint image of the patient's hip joint after hip joint replacement surgery (the first lower edge point, the second The two lower edge points, the first teardrop point and the second teardrop point) are identified and the difference in leg length is calculated, so as to realize the rapid assessment of postoperative recovery of patients undergoing total hip replacement surgery.

Further, in one embodiment, step S3 may also specifically include:

S32. Determine the first femoral medullary canal centerline on the same side as the femoral prosthesis ball head area and the second femoral medullary canal centerline on the same side as the femoral head area on the healthy side according to the bilateral cortical bone areas;

S33. Determine the fifth shortest distance between the first rotation center point of the ball head region of the femoral prosthesis and the centerline of the first femoral medullary canal;

S34. Determine the sixth shortest distance between the second center of rotation of the femoral head region on the healthy side and the centerline of the second femoral canal;

S35. Determine the femoral eccentricity according to the difference between the fifth shortest distance and the sixth shortest distance;

Wherein, the eccentricity includes the femoral eccentricity.

Optionally, after identifying the bilateral cortical area of the hip joint image of the patient after hip replacement surgery, and calculating the centerline e1 of the first femoral medullary canal and the healthy The centerline e2 of the second femoral medullary canal on the same side as the lateral femoral head region is specifically shown in FIG. 13 .

It should be noted that the centerline of the femoral medullary cavity is fitted by mathematical means based on the segmentation of the cortical bone area. First, the image is cut into two parts, left and right, and then the segmented cortical bone area is retained in a certain proportion. And take points, the point method is to take the ordinate of the reserved area as a group, retain the intersection point with the ordinate as the axis and the reserved area, and select the midpoint of the two adjacent points with the largest distance to save, and traverse all the ordinates of the reserved area Coordinates, a series of points are obtained, and a straight line is obtained by least squares fitting, which is the required centerline of the femoral medullary cavity.

According to the identified femoral prosthesis ball head area and the healthy side femoral head area of the hip joint image of the patient after hip joint replacement surgery, the first rotation center point F1 of the femoral prosthesis ball head area is calculated respectively (see Figure 14 ) and the second center of rotation F2 of the femoral head region on the uninjured side.

The first rotation center F1 of the femoral prosthesis ball head area is extracted from the femoral prosthesis ball head area, and the edge contour is extracted through traditional image processing. The intersection point of the two vertical lines is the center of the ball head of the prosthesis. The femoral head center can be obtained by the centroid formula of the region of interest. The centroid formula is:

Among them, the coordinate of each pixel in the x direction in the image is: x _i , the corresponding pixel value is: P _i , the coordinate of the centroid in the x direction is: x ₀ , and the coordinate of each pixel in the image in the direction is: y _j , the corresponding pixel value is: P _i , the coordinate of the centroid in the x direction is: y ₀ , and n represents the number of image pixels.

As shown in Figure 15, it is assumed that the first center of rotation F1 of the ball head area of the femoral prosthesis, the second center of rotation F2 of the femoral head area of the healthy side, the centerline e1 of the first femoral medullary canal, and the centerline of the second femoral medullary canal e2.

Draw a perpendicular line from the first rotation center point F1 to the first femoral medullary canal centerline e1 to obtain line segment F1d1, and draw a perpendicular line from the second rotation center point F2 to the first femoral medullary canal centerline e2 to obtain line segment F2d2 and line segment F1d1 That is, the fifth shortest distance between the first rotation center point F1 and the first femoral medullary canal centerline e1, and the line segment F2d2 is the sixth shortest distance between the second rotation center point F2 and the second femoral medullary canal centerline e2 shortest distance.

According to the distance of the line segment F1d1 and the distance of the line segment F2d2, calculate the difference between the line segment F1d1 and the line segment F2d2, the difference is the patient's femoral eccentricity, which can be used to judge the patient's lower limb after joint replacement surgery Restoration of leg length.

It should be noted that the femoral eccentricity can also be calculated by the following method, specifically: draw a vertical line from the lower edge points A1 and A2 of the bilateral lesser trochanter to the ischial tuberosity line CD respectively, and obtain the first lower edge point The shortest distance between A1 and the ischial tuberosity line CD (the first line segment A1a1) and the shortest distance between the second lower border point A2 and the ischial tuberosity line CD (the second line segment A2a2), based on which the first line segment A2a2 can be obtained The first intersection a1 and the second intersection a2 of the first segment A1a1 and the second segment A2a2 with the ischial tuberosity line CD respectively.

The first straight line is obtained by extending the first line segment A1a1, the second straight line is obtained by extending the second line segment A2a2, and then the shortest distance between the pubic symphysis point G and the first straight line and the pubic symphysis point G and the second straight line are calculated The shortest distance between, for example, the longitudinal axis perpendicular to the ischial tuberosity line CD along the pubic symphysis point G, and the vertical axis between the first line and the second line along the pubic symphysis point G respectively Vertical line, the shortest distance between the longitudinal axis and the first straight line is determined according to the vertical distance between the pubic symphysis point G and the first straight line, and the vertical distance between the pubic symphysis point G and the second straight line is determined Determine the shortest distance between the longitudinal axis and the second straight line, and calculate the shortest distance between the pubic symphysis point G and the first straight line and the shortest distance between the pubic symphysis point G and the second straight line, namely Is the femoral eccentricity (see Figure 16).

According to the femoral eccentricity obtained above, the postoperative recovery of patients undergoing hip joint replacement surgery is evaluated. If the femoral eccentricity is within the preset range (for example, 31mm to 45mm), the postoperative recovery of patients undergoing hip joint replacement surgery is determined. good.

The eccentric structure of the femur affects the strength and performance of the hip abductors. An appropriate femoral eccentricity can balance the muscle strength of the hip abductors to obtain the maximum abductor force and the minimum joint interface stress, that is, Pelvic balance can also be achieved with minimal abductor muscle force. Increased eccentricity increases the corresponding abductor muscle force arm, correspondingly reduces abductor muscle force, reduces joint contact stress, and reduces prosthesis wear. At the same time, the stress on the neck of the prosthesis is reduced. The corresponding parts of the femoral stress decreased.

This application provides a deep learning-based image evaluation method after hip replacement surgery, which identifies the corresponding key points in the hip joint image of patients after hip joint replacement surgery and calculates the femoral eccentricity, so as to realize the evaluation of total hip arthroplasty. Rapid assessment of postoperative recovery in patients undergoing replacement surgery.

Further, in one embodiment, step S3 may also specifically include:

S36. Determine the acetabular cup eccentricity according to the first rotation center point of the femoral prosthesis ball head area, the second rotation center point of the uninjured femoral head area, the ischial tuberosity line and the pelvic central axis; or

S37. Determine the acetabular cup eccentricity according to the first rotation center point, the second rotation center point, the line connecting the bilateral teardrop points and the pelvic central axis;

Among them, the central axis of the pelvis is determined according to the position of the third key point and the ischial tuberosity line;

The eccentricity includes the acetabular cup eccentricity.

Optionally, according to the first center of rotation F1 of the femoral prosthesis ball head area, the second center of rotation F2 of the femoral head area of the healthy side, the ischial tuberosity line CD and the pelvic axis EF, determine the acetabular cup eccentricity; Alternatively, the acetabular cup eccentricity can be determined according to the first rotation center point F1, the second rotation center point F2, the line ab of the bilateral teardrop points, and the central axis EF of the pelvis, and the acetabular cup eccentricity can be used to determine the patient's Eccentricity. Wherein, the line ab of the bilateral teardrop point is obtained after connecting the first teardrop point and the second teardrop point.

This application provides a method for calculating the eccentricity after total hip joint surgery based on deep learning, which identifies the corresponding key points in the hip image of patients after hip replacement surgery and calculates the acetabular cup eccentricity, which provides a basis for subsequent hip replacement surgery. Acetabular cup eccentricity evaluates the accuracy of the installation position of the femoral prosthesis, thereby laying the foundation for rapid and accurate evaluation of the patient's postoperative recovery.

Further, in one embodiment, step S36 may specifically include:

S361. Determine the seventh shortest distance between the first rotation center point and the ischial tuberosity line;

S362. Determine the eighth shortest distance between the second rotation center point and the ischial tuberosity line;

S363. Determine the ninth shortest distance between the first rotation center point and the central axis of the pelvis;

S364. Determine the tenth shortest distance between the second rotation center point and the central axis of the pelvis;

S365. Determine the acetabular cup eccentricity according to the difference between the seventh shortest distance and the eighth shortest distance and the difference between the ninth shortest distance and the tenth shortest distance.

Further, in one embodiment, step S37 may specifically include:

S371. Determine the eleventh shortest distance between the first rotation center point and the line connecting the teardrop points on both sides;

S372. Determine the twelfth shortest distance between the second rotation center point and the line connecting the teardrop points on both sides;

S373. Determine the thirteenth shortest distance between the first rotation center point and the central axis of the pelvis;

S374. Determine the fourteenth shortest distance between the second rotation center point and the central axis of the pelvis;

S375. Determine the acetabular cup eccentricity according to the difference between the eleventh shortest distance and the twelfth shortest distance and the difference between the thirteenth shortest distance and the fourteenth shortest distance.

Optionally, as shown in Figure 18, the acetabular cup eccentricity is determined according to the first rotation center point F1, the second rotation center point F2, the ischial tuberosity line CD and the pelvic central axis EF, specifically:

Make a vertical line from the first rotation center point F1 and the second rotation center point F2 to the ischial tuberosity line CD respectively, and obtain the line segment F1L1 and the line segment F2L2. The distance between the line segment F1L1 is the first rotation center point F1 and the ischial node The seventh shortest distance between the node lines CD, the distance between the line segment F2L2 is the eighth shortest distance between the second rotation center point F2 and the ischial tuberosity line CD.

And draw a vertical line from the first rotation center point F1 and the second rotation center point F2 to the pelvic central axis EF respectively, and obtain the line segment F1N1 and the line segment F2N2, and the distance between the line segment F1N1 is the first rotation center point F1 and the center of the pelvis. The ninth shortest distance between the axes EF, the distance between the line segments F2N2 is the twelve shortest distances between the second rotation center point F2 and the pelvic central axis EF.

By calculating the difference between the seventh shortest distance and the eighth shortest distance and the difference between the ninth shortest distance and the tenth shortest distance, and according to the absolute value of the difference between the seventh shortest distance and the eighth shortest distance value and the absolute value of the difference between the ninth and tenth shortest distances to determine the acetabular cup eccentricity, for example if the absolute value of the difference between the seventh and eighth shortest If the difference between the absolute value of the distance and the difference between the tenth shortest distance is within the preset threshold range, the accuracy of determining the installation position of the femoral prosthesis is relatively high.

It should be noted that the central axis EF of the pelvis is determined by making a vertical line perpendicular to the ischial tuberosity line CD along the pubic symphysis point G (see FIG. 17 ).

As shown in Figure 19, the acetabular cup eccentricity is determined according to the first rotation center point F1, the second rotation center point F2, the line connecting the bilateral teardrop points, and the central axis EF of the pelvis, specifically:

Make a vertical line from the first rotation center point F1 and the second rotation center point F2 to the line connecting the teardrop points on both sides to obtain the line segment F1P1 and the line segment F2P2. The distance between the line segments F1P1 is also the first rotation center point F1 The eleventh shortest distance between the line segment F2P2 and the line connecting the teardrop points on both sides is the twelfth shortest distance between the second rotation center point F2 and the line connecting the teardrop points on both sides.

And make a perpendicular line from the first rotation center point F1 and the second rotation center point F2 to the pelvic central axis EF respectively, to obtain the line segment F1Q1 and the line segment F2Q2, the distance between the line segment F1Q1 is the first rotation center point F1 and the center of the pelvis The thirteenth shortest distance between the axes EF, the distance between the line segment F2Q2 is the fourteenth shortest distance between the second rotation center point F2 and the pelvic central axis EF.

By calculating the difference between the eleventh shortest distance and the twelfth shortest distance and the difference between the thirteenth shortest distance and the fourteenth shortest distance, and based on the difference between the eleventh shortest distance and the twelfth shortest distance The absolute value of the difference between and the absolute value of the difference between the thirteenth shortest distance and the fourteenth shortest distance determines the acetabular cup eccentricity, for example, if the eleventh shortest distance and the twelfth shortest distance are between The difference between the absolute value of the difference and the absolute value of the difference between the thirteenth shortest distance and the fourteenth shortest distance is within the preset threshold range, the accuracy of determining the installation position of the femoral prosthesis is high .

This application provides a deep learning-based image evaluation method after hip replacement surgery, which identifies the corresponding key points in the hip image of patients after hip replacement surgery and calculates the acetabular cup eccentricity to evaluate the femoral prosthesis The accuracy of the installation position has laid the foundation for the rapid and accurate assessment of the patient's postoperative recovery.

Further, in one embodiment, step S3 may also specifically include:

S38. According to the two vertices of the outer diameter of the femoral prosthesis in the ball head area of the femoral prosthesis, the two junction points between the femoral prosthesis and the ball head area of the femoral prosthesis, and the ischial tubercle line, determine the anteversion angle and the external angle of the femoral prosthesis Spread angle;

S39. Determine the patient's femoral prosthesis index according to the anteversion angle and the abduction angle.

Optionally, the two outer diameter vertices of the femoral prosthesis in the femoral prosthesis ball head area and the two outer diameter vertices of the femoral prosthesis and the femoral prosthesis ball head area in the identified hip joint image of the patient after hip joint replacement surgery. A junction point (specifically as shown in Figure 20) carries out ellipse fitting, specifically:

As shown in Figure 21, the opening of the acetabular cup prosthesis (ie, the femoral prosthesis) is circular, and its projection on the medical image is an ellipse (hereinafter referred to as "acetabular ellipse"). According to the definition of anteversion angle, the acetabular ellipse is short The arcsine function of the ratio of the semi-axis to the semi-major axis is the image anteversion of the femoral prosthesis. The major axis of the acetabular ellipse is usually directly measured manually on medical images, but the apex of the minor axis of the acetabular ellipse is often blocked by the femoral prosthesis, so it is impossible to directly measure the semi-minor axis length on medical images. At present, the measurement of acetabular anteversion on medical images is calculated based on manual measurement data, and the occluded curve is supplemented by estimation, and the accuracy is very low.

In the embodiment of the present application, two intersecting arcs can be determined according to the four target key points determined by the deep learning model, and the least square method is used to perform ellipse fitting, and five parameters of the ellipse equation are obtained after fitting. Through these parameters, the semi-major axis and semi-minor axis of the ellipse can be obtained, and then the size of the forward tilt angle can be obtained according to the formula of the forward tilt angle. Wherein, the elliptic equation is mx2+nxy+oy2+px+qy+1=0, and m, n, o, p, and q are five elliptic equation parameters. Assuming that the semi-minor axis of the ellipse is K1 and the semi-major axis of the ellipse is K2, the anteversion angle of the femoral prosthesis can be determined as arcsin(K1/K2) according to K1 and K2.

The angle between the ischial tuberosity line CD and the apex of the outer diameter of the acetabular cup is taken as the abduction angle, as shown in Figure 22.

Then, according to the first rotation center point F1, the second rotation center point F2, the ischial tuberosity line CD and the pelvic central axis EF, determine the acetabular cup eccentricity; or according to the first rotation center point F1, the second rotation center point F2 , The line connecting the bilateral teardrop points and the pelvic axis EF determines the eccentricity of the acetabular cup.

This application provides a deep learning-based image evaluation method after hip replacement surgery, which identifies the corresponding key points in the hip joint images of patients after hip replacement surgery and calculates the femoral prosthesis index to evaluate the femoral prosthesis installation The accuracy of the position lays the foundation for the subsequent realization of a rapid and accurate assessment of the postoperative recovery of the patient.

The following is a description of the deep learning-based image evaluation system after hip arthroplasty provided by this application. The deep learning-based image evaluation system after hip arthroplasty described below is the same as the deep learning-based hip arthroplasty described above. The evaluation methods of post-images can be compared with each other.

Fig. 23 is a schematic structural diagram of the deep learning-based image evaluation system after hip arthroplasty provided by this application, as shown in Fig. 23, including:

An acquisition module 2310, an identification module 2311, a determination module 2312 and an evaluation module 2313;

An acquisition module 2310 configured to acquire a hip joint image of a patient after hip joint replacement surgery;

The recognition module 2311 is configured as a target recognition network based on deep learning to recognize key point positions and target areas in the hip joint image;

The determination module 2312 is configured to determine the patient's leg length difference, eccentricity and femoral prosthesis index according to the key point position and the target area;

The evaluation module 2313 is configured to evaluate the installation accuracy of the patient's femoral prosthesis position according to the leg length difference between the two legs, the eccentric distance and the femoral prosthesis index;

The image evaluation system after hip replacement surgery based on deep learning provided by this application is based on the hip joint images of patients after hip joint replacement surgery, and calculates the difference in leg length, eccentricity and Femoral prosthesis metrics to enable accurate assessment of recovery in patients undergoing hip replacement surgery.

Further, in an embodiment, the identification module 2311 may also be specifically configured as:

Input the hip joint image into the target recognition network to determine the first inferior edge point, the second inferior edge point, the first teardrop point on both sides of the ischium region, and the bilateral lesser trochanter in the hip joint image. Second teardrop point, pubic symphysis point, femoral prosthesis ball head area, contralateral femoral head area, bilateral cortical bone area and ischium area;

The first lower edge point and the second lower edge point are determined as the first key point position, the first tear drop point and the second tear drop point are determined as the second key point position and the pubic symphysis point respectively Determined as the third key point position;

Determine the position of the key point according to the position of the first key point, the position of the second key point and the position of the third key point;

The area of the ball head of the femoral prosthesis, the area of the healthy femoral head, the area of the bilateral cortical bone, and the area of the ischium were determined as the target area;

Wherein, the target recognition network is obtained based on point recognition neural network and segmentation neural network training; or,

The image evaluation system after hip replacement surgery based on deep learning provided by this application combines deep learning methods to evaluate the accuracy of the installation position of the femoral prosthesis in patients undergoing hip replacement surgery, so as to realize the accuracy of hip replacement surgery. Rapid and accurate assessment of postoperative recovery in patients undergoing replacement surgery.

Further, in an embodiment, the determining module 2312 may also be specifically configured as:

According to the position of the first key point and the ischial tuberosity line, determine the leg length difference between the two legs; or

According to the position of the first key point and the line connecting the teardrop points on both sides, determine the leg length difference between the legs;

The image evaluation system after hip replacement surgery based on deep learning provided by this application uses deep learning methods to identify the corresponding key points and target areas in the hip joint images of patients after hip joint replacement surgery and calculate the difference in leg length. It laid the foundation for the subsequent rapid assessment of the postoperative recovery of patients undergoing hip replacement surgery based on the leg length difference.

Determine the first shortest distance between the first inferior border point and the ischial tuberosity line;

Determine the second shortest distance between the second inferior border point and the ischial tuberosity line;

According to the difference between the first shortest distance and the second shortest distance, the leg length difference between the two legs is determined.

The image evaluation system after hip replacement surgery based on deep learning provided by this application uses deep learning methods to analyze the corresponding key points in the hip joint image of patients after hip joint replacement surgery (the first lower edge point and the second edge point) two lower edge points) and the target area to identify and calculate the leg length difference, so as to realize the rapid assessment of postoperative recovery of patients undergoing total hip replacement surgery.

Determine the third shortest distance between the first lower edge point and the line between the bilateral teardrop points;

Determine the fourth shortest distance between the second lower edge point and the line connecting the bilateral teardrop points;

Based on the difference between the third shortest distance and the fourth shortest distance, the leg length difference between the two legs is determined.

The image evaluation system after hip joint replacement based on deep learning provided by this application uses the deep learning method to analyze the corresponding key points in the hip joint image of the patient's hip joint after hip joint replacement surgery (the first lower edge point, the second The two lower edge points, the first teardrop point and the second teardrop point) are identified and the difference in leg length is calculated, so as to realize the rapid assessment of postoperative recovery of patients undergoing total hip replacement surgery.

According to the bilateral cortical bone area, determine the centerline of the first femoral medullary canal on the same side as the ball head area of the femoral prosthesis and the second femoral medullary canal centerline on the same side as the femoral head area on the healthy side;

Determine the fifth shortest distance between the first center of rotation of the femoral prosthesis ball head area and the centerline of the first femoral medullary canal;

Determine the femoral eccentricity according to the difference between the fifth shortest distance and the sixth shortest distance;

Wherein, the eccentricity includes the femoral eccentricity.

The image evaluation system after hip joint replacement based on deep learning provided by this application can identify the corresponding key points in the hip joint image of patients after hip joint replacement surgery and calculate the femoral eccentricity, so as to realize the evaluation of total hip joint Rapid assessment of postoperative recovery in patients undergoing replacement surgery.

Determine the acetabular cup offset based on the first center of rotation in the region of the femoral prosthesis ball, the second center of rotation in the region of the uninjured femoral head, the ischial tuberosity line, and the midline of the pelvis; or

Determine the acetabular cup eccentricity according to the first rotation center point, the second rotation center point, the line connecting the bilateral teardrop points and the central axis of the pelvis;

Eccentricity includes acetabular cup eccentricity.

The image evaluation system after hip replacement surgery based on deep learning provided by this application identifies the corresponding key points in the hip image of patients after hip replacement surgery and calculates the acetabular cup eccentricity, which is used for subsequent acetabular-based Cup eccentricity evaluates the accuracy of the installation position of the femoral prosthesis, thereby laying the foundation for rapid and accurate evaluation of the patient's postoperative recovery.

determining the seventh shortest distance between the first center of rotation and the ischial tuberosity line;

determining a tenth shortest distance between the second center of rotation and the central axis of the pelvis;

The acetabular cup offset is determined based on the difference between the seventh shortest distance and the eighth shortest distance and the difference between the ninth shortest distance and the tenth shortest distance.

Determine the eleventh shortest distance between the first rotation center point and the line connecting the teardrop points on both sides;

Determine the twelfth shortest distance between the second rotation center point and the line connecting the bilateral teardrop points;

determining a thirteenth shortest distance between the first center of rotation and the central axis of the pelvis;

determining a fourteenth shortest distance between the second center of rotation and the central axis of the pelvis;

The acetabular cup offset was determined from the difference between the eleventh and twelfth shortest distances and the difference between the thirteenth and fourteenth shortest distances.

The image evaluation system after hip replacement surgery based on deep learning provided by this application identifies the corresponding key points in the hip image of patients after hip replacement surgery and calculates the acetabular cup eccentricity to evaluate the femoral prosthesis The accuracy of the installation position has laid the foundation for the rapid and accurate assessment of the patient's postoperative recovery.

Determine the anteversion and abduction angles of the femoral component based on the two vertices of the outer diameter of the femoral component in the area of the femoral component ball, the two junction points of the femoral component and the area of the femoral component ball, and the ischial tuberosity line ;

According to the anteversion angle and abduction angle, determine the patient's femoral prosthesis index.

The image evaluation system after hip replacement surgery based on deep learning provided by this application identifies the corresponding key points in the hip joint images of patients after hip replacement surgery and calculates the femoral prosthesis index to evaluate the femoral prosthesis installation The accuracy of the position lays the foundation for the subsequent realization of a rapid and accurate assessment of the postoperative recovery of the patient.

Fig. 24 is a schematic diagram of the physical structure of an electronic device provided by the present application. As shown in Fig. 24, the electronic device may include: a processor (processor) 2410, a communication interface (communication interface) 2411, a memory (memory) 2412 and a bus (bus) 2413, wherein, the processor 2410, the communication interface 2411, and the memory 2412 complete mutual communication through the bus 2413. Processor 2410 may invoke logic instructions in memory 2412 to perform the following methods:

Obtaining hip images of patients after hip replacement surgery;

Target recognition network based on deep learning to identify key point positions and target areas in hip joint images;

According to the leg length difference, eccentric distance and femoral prosthesis index, the accuracy of the patient's femoral prosthesis installation is evaluated;

In addition, the above logic instructions in the memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including Several instructions are used to make a computer power panel (which may be a personal computer, a server, or a network power panel, etc.) execute all or part of the steps of the methods described in various embodiments of the present application. The aforementioned storage medium includes: various media that can store program codes such as U disk, mobile hard disk, read-only memory (ROM, Read-only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk, etc. .

Further, the present application discloses a computer program product, the computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program includes program instructions, and when the program instructions are executed by a computer The computer can execute the deep learning-based image evaluation method after hip arthroplasty provided by the above method embodiments, for example including:

Obtaining hip images of patients after hip replacement surgery;

On the other hand, the present application also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it is implemented to perform the hip joint replacement based on deep learning provided by the above-mentioned embodiments. Evaluation methods for postoperative imaging include, for example:

Obtaining hip images of patients after hip replacement surgery;

The system embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in One place, or it can be distributed to multiple network elements. Part or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. It can be understood and implemented by those skilled in the art without any creative effort.

Through the above description of the implementations, those skilled in the art can clearly understand that each implementation can be implemented by means of software plus a necessary general-purpose hardware platform, and of course also by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the prior art can be embodied in the form of software products, and the computer software products can be stored in computer-readable storage media, such as ROM/RAM, magnetic Disc, CD, etc., including several instructions to make a computer power panel (which can be a personal computer, a server, or a network power panel, etc.) execute the methods described in various embodiments or some parts of the embodiments.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, rather than limiting them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: it can still Modifications are made to the technical solutions described in the foregoing embodiments, or equivalent replacements are made to some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present application.

Claims

A method for evaluating images after hip replacement surgery based on deep learning, including:

Obtaining hip images of patients after hip replacement surgery;

A target recognition network based on deep learning, identifying key point positions and target areas in the hip joint image;

According to the position of the key point and the target area, determine the patient's leg length difference, eccentricity and femoral prosthesis index;

According to the leg length difference of the two legs, the eccentric distance and the femoral prosthesis index, the accuracy of the installation of the patient's femoral prosthesis position is evaluated;

Wherein, the installation accuracy of the position of the femoral prosthesis is used to evaluate the postoperative recovery of the patient.
The image evaluation method after hip joint replacement based on deep learning according to claim 1, wherein the target recognition network is trained based on a point recognition neural network and a segmentation neural network; or,

It is trained based on the preset neural network model including stacked hourglass network structure, segmented Segment-Head network and keypoint Keypoint-Head network.
The image evaluation method after hip arthroplasty based on deep learning according to claim 1, wherein the target recognition network based on deep learning identifies key point positions and target areas in the hip joint image, including:

Input the hip joint image into the target recognition network to determine the first lower edge point, the second lower edge point, and the first tear on both sides of the ischium region corresponding to the bilateral lesser trochanter in the hip joint image Drop point, second tear drop point, pubic symphysis point, femoral prosthesis ball head area, healthy side femoral head area, bilateral cortical bone area and ischium area;

Determining the first lower edge point and the second lower edge point as the first key point position, and determining the first teardrop point and the second teardrop point as the second key point respectively The position and the pubic symphysis point are determined as the third key point position;

determining the first key point position, the second key point position and the third key point position as the key point position;

The area of the ball head of the femoral prosthesis, the area of the femoral head on the healthy side, the area of the bilateral cortical bone, and the area of the ischium are determined as the target area.
The image evaluation method after hip joint replacement based on deep learning according to claim 3, wherein, according to the position of the key point and the target area, determining the leg length difference of the patient's legs includes:

Determine the leg length difference between the two legs according to the position of the first key point and the ischial tuberosity line; or,

According to the position of the first key point and the line connecting the teardrop points on both sides, determine the leg length difference between the two legs;

Wherein, the ischial tuberosity line is determined according to the bilateral first and second lowest points of the ischial region;

The line connecting the bilateral teardrop points is determined according to the position of the second key point.
The image evaluation method after hip joint replacement based on deep learning according to claim 4, wherein, according to the position of the first key point and the ischial tuberosity line, determining the leg length difference between the legs comprises:

determining a first shortest distance between the first inferior border point and the ischial tuberosity line;

determining a second shortest distance between the second inferior border point and the ischial tuberosity line;

The leg length difference between the two legs is determined according to the difference between the first shortest distance and the second shortest distance.
The image evaluation method after hip joint replacement based on deep learning according to claim 4, wherein, according to the position of the first key point and the line connecting the teardrop points on both sides, the length of the legs is determined poor, including:

determining the third shortest distance between the first lower edge point and the line connecting the bilateral teardrop points;

determining the fourth shortest distance between the second lower edge point and the line connecting the bilateral teardrop points;

The leg length difference between the two legs is determined according to the difference between the third shortest distance and the fourth shortest distance.
The image evaluation method after hip arthroplasty based on deep learning according to claim 3, wherein said determining the femoral offset of the patient according to the position of the key point and the target area includes:

According to the bilateral cortical bone area, determine the first femoral medullary canal centerline on the same side as the femoral prosthesis ball head area and the second femoral medullary canal centerline on the same side as the healthy side femoral head area;

determining the fifth shortest distance between the first center of rotation of the femoral prosthesis ball head area and the centerline of the first femoral canal;

Determine the sixth shortest distance between the second center of rotation of the femoral head region on the healthy side and the centerline of the second femoral canal;

determining the femoral offset according to the difference between the fifth shortest distance and the sixth shortest distance;

Wherein, the eccentricity includes the femoral eccentricity.
The image evaluation method after hip arthroplasty based on deep learning according to claim 3, wherein said determining the eccentricity of the patient according to the position of the key point and the target area further comprises:

Determine the acetabular cup eccentricity according to the first center of rotation of the femoral prosthesis ball head area, the second center of rotation of the healthy side femoral head area, the ischial tuberosity line and the pelvic central axis; or,

Determine the eccentricity of the acetabular cup according to the first center of rotation, the second center of rotation, the line connecting the bilateral teardrop points and the central axis of the pelvis;

Wherein, the central axis of the pelvis is determined according to the position of the third key point and the ischial tuberosity line;

The eccentricity includes the acetabular cup eccentricity.
The method for evaluating images after hip arthroplasty based on deep learning according to claim 8, wherein, according to the first rotation center point of the ball head area of the femoral prosthesis and the second center point of the femoral head area of the healthy side 2 Rotation center point, ischial tuberosity line and pelvic central axis, determine acetabular cup eccentricity, including:

determining a seventh shortest distance between the first center of rotation and the ischial tuberosity line;

determining an eighth shortest distance between the second center of rotation and the ischial tuberosity line;

determining a ninth shortest distance between the first center of rotation and the central axis of the pelvis;

determining a tenth shortest distance between the second center of rotation and the central pelvic axis;

The acetabular cup eccentricity is determined according to the difference between the seventh shortest distance and the eighth shortest distance and the difference between the ninth shortest distance and the tenth shortest distance.
The image evaluation method after hip joint replacement based on deep learning according to claim 8, wherein, according to the first rotation center point, the second rotation center point, and the line connecting the teardrop points on both sides and the central axis of the pelvis to determine the eccentricity of the acetabular cup, including:

determining the eleventh shortest distance between the first rotation center point and the line connecting the bilateral teardrop points;

determining the twelfth shortest distance between the second rotation center point and the line connecting the bilateral teardrop points;

determining a thirteenth shortest distance between the first center of rotation and the central pelvic axis;

determining a fourteenth shortest distance between the second center of rotation and the central pelvic axis;

Determining the acetabular cup eccentricity based on the difference between the eleventh shortest distance and the twelfth shortest distance and the difference between the thirteenth shortest distance and the fourteenth shortest distance distance.
The method for evaluating images after hip arthroplasty based on deep learning according to claim 3, wherein said determining the patient's femoral prosthesis index according to said key point position and said target area includes:

According to the two outer diameter vertices of the femoral prosthesis in the ball head area of the femoral prosthesis, the two junction points of the femoral prosthesis and the ball head area of the femoral prosthesis and the ischial tuberosity line, determine the femoral prosthesis Body anteversion and abduction angles;

According to the anteversion angle and the abduction angle, the femoral prosthesis index of the patient is determined.
A deep learning-based image evaluation system after hip replacement surgery, including: an acquisition module, an identification module, a determination module, and an evaluation module;

The acquiring module is configured to acquire hip joint images of patients after hip joint replacement surgery;

The identification module is configured as a target recognition network based on deep learning to identify key point positions and target areas in the hip joint image;

The determination module is configured to determine the patient's leg length difference, eccentricity and femoral prosthesis index according to the key point position and the target area;

The evaluation module is configured to evaluate the installation accuracy of the patient's femoral prosthesis position according to the leg length difference between the two legs, the eccentric distance and the femoral prosthesis index;

Wherein, the installation accuracy of the position of the femoral prosthesis is used to evaluate the postoperative recovery of the patient.
An electronic device, comprising a processor and a memory storing a computer program, when the processor executes the computer program, the method for evaluating images after hip joint replacement based on deep learning according to any one of claims 1 to 11 is implemented .
A non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the deep learning-based postoperative imaging of hip joint replacement according to any one of claims 1 to 11 is realized. assessment method.
A computer program product, comprising a computer program, when the computer program is executed by a processor, the deep learning-based image evaluation method after hip joint replacement according to any one of claims 1 to 11 is realized.