WO2023241032A1

WO2023241032A1 - Deep learning-based method and system for intelligently identifying osteoarthritis

Info

Publication number: WO2023241032A1
Application number: PCT/CN2023/071148
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: 2022-06-15
Filing date: 2023-01-08
Publication date: 2023-12-21
Also published as: CN115131301A; CN115131301B

Abstract

The present application provides a deep learning-based method and system for intelligently identifying osteoarthritis, an electronic device, a readable storage medium, and a computer program product, applicable to the field of image processing. The method comprises: acquiring an image to be processed; determining a first contour area of a first bone and a second contour area of a second bone in the image to be processed; projecting multiple discrete points located on the first contour area of a distal end of the first bone to a second contour area of a proximal end of the second bone, to obtain multiple projection points corresponding one-to-one to the multiple discrete points; and determining a gap between the first bone and the second bone according to the distances between the multiple discrete points and the projection points corresponding to the multiple discrete points. The deep learning-based method and system for intelligently identifying osteoarthritis, the electronic device, the readable storage medium, and the computer program product provided by the present application may enhance the accuracy of determining an identified bone gap.

Description

Method and system for intelligent identification of osteoarthritis based on deep learning

Cross-references to related applications

This application claims priority to the Chinese patent application with application number 202210682179.8 submitted on June 15, 2022, titled "Method and System for Intelligent Identification of Osteoarthritis Based on Deep Learning", which is fully incorporated herein by reference. .

Technical field

This application relates to the field of image processing technology, and in particular to a method and system for intelligently identifying osteoarthritis based on deep learning.

Background technique

Bone gap refers to the distance between different types of bones. Accurately determining the gaps between bones is of great significance in the medical field.

In the existing technology, when determining the bone gap, the user usually evaluates the gap between the bones by taking medical images and observing with the naked eye.

However, the accuracy of the bone gaps obtained by the above method is not high.

Contents of the invention

This application provides a method and system for intelligently identifying osteoarthritis based on deep learning, which is used to solve the shortcoming of low accuracy in determining bone gaps in the existing technology and improve the accuracy of determined bone gaps.

This application provides a method for intelligently identifying osteoarthritis based on deep learning, including:

Get the image to be processed;

Determine the first outline area of the first bone and the second outline area of the second bone in the image to be processed;

A plurality of discrete points located on the first contour area at the distal end of the first bone are projected onto the second contour area at the proximal end of the second bone to obtain a multi-point coordinate system corresponding to the plurality of discrete points in a one-to-one manner. projection points;

The gap between the first bone and the second bone is determined according to the distance between the plurality of discrete points and respective corresponding projection points of the plurality of discrete points.

According to a method for intelligently identifying osteoarthritis based on deep learning provided by this application, the first step is determined based on the distance between the plurality of discrete points and the projection points corresponding to the plurality of discrete points. The gap between the bone and said second bone, including:

Determine the distance between each discrete point in the plurality of discrete points and the projection point corresponding to each discrete point;

The smallest distance among multiple distances is determined as the gap between the first bone and the second bone.

An average of the plurality of distances is determined as the gap between the first bone and the second bone.

According to a method for intelligently identifying osteoarthritis based on deep learning provided by this application, determining the first outline area of the first bone and the second outline area of the second bone in the image to be processed includes:

The image to be processed is input into a pre-trained bone segmentation model to obtain the area where the first bone is located and the area where the second bone is located, wherein the bone segmentation model utilizes multiple The sample images are obtained after training, wherein the plurality of sample images are images containing different types of bones;

Contour extraction is performed on the area where the first bone is located and the area where the second bone is located, respectively, to obtain a first outline area of the first bone and a second outline area of the second bone.

According to a method for intelligently identifying osteoarthritis based on deep learning provided by this application, a plurality of discrete points located on the first contour area at the distal end of the first bone are projected to the second bone Before obtaining a plurality of projection points corresponding to the plurality of discrete points on the proximal second contour area, the method further includes:

determining a medial edge boundary point of the distal end of the first bone and a lateral edge boundary point of the distal end of the first bone;

A plurality of discrete points are determined on the first contour area of the distal end of the first bone between the medial edge boundary point and the lateral edge boundary point.

According to a method for intelligently identifying osteoarthritis based on deep learning provided in this application, after determining the gap between the first bone and the second bone, the method further includes:

Match the gap with multiple preset gap ranges and determine the target gap range in which the gap is located;

According to the target gap range, the degree of disease of the joint where the first bone and the second bone are located is determined.

This application also provides a system for intelligent identification of osteoarthritis based on deep learning, including:

The acquisition module is configured to acquire the image to be processed;

a determination module configured to determine the first outline area of the first bone and the second outline area of the second bone in the image to be processed;

The projection module is configured to project a plurality of discrete points located on the first contour area at the distal end of the first bone to the second contour area at the proximal end of the second bone to obtain the results corresponding to the plurality of discrete points. Multiple projection points with one-to-one correspondence;

The determination module is further configured to determine the gap between the first bone and the second bone based on the distance between the plurality of discrete points and the corresponding projection points of the plurality of discrete points.

This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements any of the above deep learning-based methods. An intelligent method for identifying osteoarthritis.

This application also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the method for intelligently identifying osteoarthritis based on deep learning is implemented as described in any one of the above.

This application also provides a computer program product, which includes a computer program that, when executed by a processor, implements any one of the above deep learning-based intelligent identification methods for osteoarthritis.

The method and system for intelligent identification of osteoarthritis based on deep learning provided by this application, by acquiring the image to be processed, determine the first outline area of the first bone and the second outline area of the second bone in the image to be processed, and locate them A plurality of discrete points on the first contour area at the distal end of the first bone are projected onto the second contour area at the proximal end of the second bone. After obtaining a plurality of projection points corresponding to the plurality of discrete points one-to-one, based on the multiple The distance between the discrete points and the corresponding projection points of the plurality of discrete points determines the gap between the first bone and the second bone, since the plurality of discrete points located at the distal end of the first bone can be projected to the second contour area to determine the gap between the first bone and the second bone based on the distance between the discrete point and the projection point, thereby improving the accuracy of the determined bone gap. And, based on the determined gap between the first bone and the second bone, whether osteoarthritis exists between the first bone and the second bone can also be identified.

Description of the drawings

In order to explain the technical solutions in this application or the prior art more clearly, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are of the present invention. For some embodiments of the application, those of ordinary skill in the art can also obtain other drawings based on these drawings without exerting creative efforts.

Figure 1 is a schematic flowchart of a method for intelligently identifying osteoarthritis based on deep learning provided by an embodiment of the present application;

Figure 2 is a schematic structural diagram of the bone segmentation model;

Figure 3 is a schematic diagram of multiple discrete points and the projection of each discrete point provided by the embodiment of the present application;

Figure 4 is a schematic structural diagram of the Hourglass neural network;

Figure 5 is a schematic diagram of a system for intelligently identifying osteoarthritis based on deep learning provided by an embodiment of the present application;

Figure 6 illustrates a schematic diagram of the physical structure of an electronic device.

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 drawings in this application. Obviously, the described embodiments are part of the embodiments of this application. , not all examples. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of this application.

The method for intelligently identifying osteoarthritis based on deep learning provided in the embodiments of this application determines the contour areas corresponding to different types of bones in the image to be processed, and projects multiple discrete points on the contour area of one of the bones. onto another contour area to determine the distance between bones based on the distance between discrete points before and after projection. Through the above method, the determination of the distance in the three-dimensional space can be converted into the determination of the distance between two points in the two-dimensional space, thereby improving the accuracy of the determined bone gap.

The method for intelligently identifying osteoarthritis based on deep learning provided by embodiments of the present application is described below with reference to Figures 1-4.

Figure 1 is a schematic flowchart of a method for intelligently identifying osteoarthritis based on deep learning provided by an embodiment of the present application. The execution subject of the method for intelligently identifying osteoarthritis based on deep learning may be an electronic device. As shown in Figure 1, the method includes:

Step 101: Obtain the image to be processed.

In this step, the images to be processed include various types of medical images, such as computed tomography (CT) images, Magnetic Resonance Imaging (MRI) images, ultrasound images, X-ray images, and DynaCT images , positron emission computerized tomography (PET) image.

In addition, the image to be processed may include a two-dimensional image or a three-dimensional image.

Step 102: Determine the first outline area of the first bone and the second outline area of the second bone in the image to be processed.

In this step, the first bone and the second bone may be different types of bones. For example, the first bone may be the femur and the second bone may be the tibia. Of course, the first bone and the second bone can also be other bones, such as fibula, patella, etc. In the embodiments of this application, the first bone is the femur and the second bone is the tibia. The method of determining the outline areas of other types of bones is similar to the way of determining the outline areas of the femur and tibia, and will not be used here. Again.

Optionally, the image to be processed can be input into a pre-trained bone segmentation model to obtain the area where the first bone is located and the area where the second bone is located, and the area where the first bone is located and the area where the second bone is located are respectively processed. Contour extraction, obtaining the first contour area of the first bone and the second contour area of the second bone.

Among them, the bone segmentation model is obtained by training the initial bone segmentation model using multiple sample images, where the multiple sample images are images containing different types of bones.

Specifically, the image to be processed is input into a pre-trained bone segmentation model, and different types of bones included in the image to be processed can be divided or segmented, thereby segmenting the area where the first bone is located and the area where the second bone is located. . After the segmentation is completed, contour extraction is performed on the area where the first bone is located and the area where the second bone is located, so that the first outline area of the first bone and the second outline area of the second bone can be obtained.

Exemplarily, Figure 2 is a schematic structural diagram of a bone segmentation model. As shown in Figure 2, the image segmentation algorithm in the embodiment of this application adopts a deeplabv3+ network structure, which includes an encoder (Encoder) and a decoder (Decoder). Among them, the first module connected in the Encoder is Deep Convolutional Neural Networks (DCNN), which represents the backbone network used to extract image features. After the image to be processed is input to the encoder, two effective feature layers are generated through DCNN, namely the shallow feature layer and the deep feature layer. The height and width of the shallow feature layer will be larger, while the downsampling of the deep feature layer will There are more, so the height and width will be smaller.

On the right side of DCNN is an atrous spatial convolutional pyramid pooling (ASPP) network. The ASPP network includes a 1*1 convolution, three 3*3 atrous convolutions and a global pooling to complete the backbone network. The output is processed. Among them, three 3*3 atrous convolutions with different sampling rates can be used for feature extraction. For example, the sampling rates of three 3*3 atrous convolutions are 6, 12 and 18 respectively. The ASPP network samples the input deep feature layer in parallel with atrous convolutions with different sampling rates, which can better capture the contextual information of the image. After that, the results are connected and a 1*1 convolution is used to reduce the number of channels to obtain a feature layer with a reduced number of channels.

The shallow feature layer generated by DCNN enters the decoder (Decoder), and after 1x1 convolution, it is transformed into the same shape as the feature layer with the number of channels reduced. The feature layer with high semantic information and reduced channel number generated by the encoder enters the Decoder for upsampling, and then is feature fused with the result obtained by 1x1 convolution of the shallow feature layer, and then through 3x3 convolution for feature fusion. Extract, and use the results of feature extraction to achieve regional segmentation of different types of bones to obtain feature maps. In addition, Decoder is an upsampling oversymmetry, that is, a process of feature restoration, which can restore the feature map to be consistent with the size of the input image to be processed.

The deeplabv3+ image segmentation algorithm can be used to segment the tibia, femur and fibula areas in the image to be processed.

In addition, the sample images required for training the bone segmentation model can be a medical image set. The medical image set includes multiple images containing different types of bones. In these images, areas of different types of bones are manually marked. , as label information. In order to improve the accuracy of the trained model, the above medical image set can also be divided into a training set and a test set, and the annotated medical images can be used as images in the training set to train the initial bone segmentation model. As images in the test set, the trained model is tested.

Input the sample image into the initial bone segmentation model, and the segmentation results of different types of bones will be output. Compare the segmentation result with the label information to obtain the loss information, adjust the network parameters of the initial bone segmentation model based on the loss information, and repeat the above training process until the loss information is minimal or the obtained bone segmentation model converges. The training process ends, and the final model is determined as the bone segmentation model.

In this embodiment, through the pre-trained bone segmentation model, the first bone and the second bone in the image to be processed can be segmented, and then the first outline area of the first bone and the second outline area of the second bone can be extracted. Contour regions, improve the efficiency and accuracy of different types of bone segmentation.

Step 103: Project multiple discrete points located on the first contour area at the distal end of the first bone onto the second contour area at the proximal end of the second bone to obtain multiple projection points that correspond one-to-one to the multiple discrete points. .

In this step, after segmenting the first outline area of the first bone and the second outline area of the second bone, multiple discrete points may be determined on the first outline area at the distal end of the first bone.

Optionally, when determining multiple discrete points, the inner edge boundary point of the distal end of the first bone and the lateral edge boundary point of the distal end of the first bone can be determined first, between the inner edge boundary point and the lateral edge boundary point. During the period, a plurality of discrete points are determined on the first contour area of the distal end of the first bone.

Specifically, Figure 3 is a schematic diagram of multiple discrete points and the projection of discrete points provided by the embodiment of the present application. As shown in Figure 3, the first outline area of the first bone is the outline area of the femur, and the second outline area of the second bone is Taking the outline area of the tibia as an example, after the outline area is determined, the medial edge boundary point P1 and the lateral edge boundary point Pn will be determined in the outline area of the distal femur.

In a possible implementation, for femur key point detection, the Hourglass neural network key point detection algorithm is used in this application. Figure 4 is a schematic structural diagram of the Hourglass neural network. As shown in Figure 4, the network is an hourglass. structure that can output pixel-level predictions. The network consists of a convolution layer (C1-C7) and a pooling layer. The feature map (C1a-C4a) in the middle part is a copy layer of the convolution layer. By adding the copy layer and the upsampling of the corresponding layer in the convolution layer, New feature information can be obtained to achieve the effect of feature fusion, which is the C1b-C4b part in the figure. The entire Hourglass is symmetrical. For every network layer in the process of obtaining low-resolution features, there will be a corresponding network layer in the upsampling process.

In this way, after superimposing the feature layers, a large feature layer, namely C1b, is obtained. This layer retains the information of all layers and is equal to the size of the input original image. Afterwards, a heatmap representing the probability of key points is generated through 1x1 convolution. , take the point with the maximum probability value in the heat map as the feature point, and the location of this feature point is the predicted femur boundary point location.

In another possible implementation, the curvature of each point in the contour area of the distal femur can also be calculated, and the two points with the largest curvatures are determined as the medial edge boundary point P1 and the lateral edge boundary point Pn.

Further, multiple discrete points can be determined on the first contour area between the inner edge boundary point P1 and the outer edge boundary point Pn, such as P1, P2, P3, P4...Pn, wherein among the multiple discrete points , the distance between any two adjacent discrete points can be the same or different. It can be understood that the smaller the distance between adjacent discrete points, that is, the greater the number of extracted discrete points, the more accurate the determined gap between the first bone and the second bone.

In this embodiment, by determining the medial edge boundary point of the distal end of the first bone and the lateral edge boundary point of the distal end of the first bone, and between the medial edge boundary point and the lateral edge boundary point, the first bone A plurality of discrete points are determined on the first contour area at the distal end, so that the distance between the first bone and the second bone can be determined through specific discrete points on the contour area of the first bone that are close to the contour area of the second bone. , thereby improving the accuracy of the determined gap between the first bone and the second bone.

After determining the plurality of discrete points, the plurality of discrete points on the first contour area can be projected onto the second contour area at the proximal end of the second bone to obtain multiple projection points. As shown in Figure 3, discrete points P1, P2, P3, P4...Pn are vertically projected onto the second contour area of the proximal tibia to obtain multiple projection points P1', P2', P3', P4'... Pn'.

Step 104: Determine the gap between the first bone and the second bone based on the distance between the plurality of discrete points and the projection points corresponding to the plurality of discrete points.

In this step, after vertically projecting multiple discrete points, the distance between each discrete point and its corresponding projection point can be calculated, thereby determining the gap between the first bone and the second bone.

In a possible implementation, after determining the distance between each discrete point among the plurality of discrete points and the projection point corresponding to each discrete point, the smallest distance among the plurality of distances is determined as the first The gap between the bone and the second bone.

Specifically, as shown in Figure 3, the projection point corresponding to the discrete point P1 is P1', the projection point corresponding to the discrete point P2 is P2',..., the projection point corresponding to the discrete point Pn is Pn', the discrete point P1 can be calculated and the distance L1 between the projection point P1', the distance L2 between the discrete point P2 and the projection point P2',..., the distance Ln between the discrete point Pn and the projection point Pn'. After determining the distance between each discrete point and the corresponding projection point, the smallest distance Lmin among all distances is determined as the gap between the first bone and the second bone. For example, the distance between the discrete point Pn and the projection point Pn' can be determined as the gap between the first bone and the second bone.

In this embodiment, after determining the distance between each discrete point and its corresponding projection point, the minimum distance is determined as the gap between the first bone and the second bone, which can be used for subsequent determination based on the bone gap. The degree of joint disease provides a favorable basis.

In another possible implementation, after determining the distance between each discrete point among the multiple discrete points and the projection point corresponding to each discrete point, the average value of the multiple distances is determined as the first bone and The gap between the second bones.

Specifically, as shown in Figure 3, when calculating the distance L1 between the discrete point P1 and the projection point P1', the distance L2 between the discrete point P2 and the projection point P2',..., the discrete point Pn and the projection point After being the distance Ln between Pn', the average of all distances is calculated and this average is determined as the gap between the first bone and the second bone.

In this embodiment, after determining the distance between each discrete point and its corresponding projection point, the average value of multiple distances is determined as the gap between the first bone and the second bone, thereby improving the determination The accuracy of the gap.

Optionally, after determining the gap between the first bone and the second bone, the gap can also be matched with multiple preset gap ranges, a target gap range in which the gap is located is determined, and based on the target gap range, the The degree of disease at the joint where the first and second bones are located.

Specifically, in one application scenario, after determining the gap between the first bone and the second bone, the gap is matched with multiple preset gap ranges to determine the target gap range in which the gap is located. . For example, assuming that the gap between the first bone and the second bone is 0.8cm, the preset multiple gap ranges include 0.3cm-0.5cm, 0.5cm-0.7cm, 0.7cm-0.9cm, 0.9cm-1.1 cm, after matching the determined gap with multiple preset gap ranges, it is determined that the gap 0.8cm between the first bone and the second bone is within the target gap range of 0.7cm-0.9cm.

In addition, each preset gap range corresponds to a degree of disease. For example, 0.3cm-0.5cm corresponds to grade four, which means that there are a large number of osteophytes in the articular cartilage joints, severe narrowing of the joint space, and obvious subchondral bone sclerosis and deformity. 0.5 cm-0.7cm corresponds to grade three, indicating moderate narrowing of the articular cartilage joint space and sclerosis of the subchondral bone. 0.7cm-0.9cm corresponds to grade two, indicating the presence of obvious osteophytes in the articular cartilage and mild narrowing of the joint space. 0.9 cm-1.1cm corresponds to the first level, indicating suspicious narrowing of the joint space and possible labial hyperplasia. Therefore, after determining the target gap range where the gap is located, the degree of disease at the joint where the first bone and the second bone are located can be further determined. For example, the degree of disease can be determined to be level two, that is to say, the degree of disease at the joint between the first bone and the second bone can be determined. Osteoarthritis occurs between bones.

The method for intelligently identifying osteoarthritis based on deep learning provided by embodiments of the present application obtains the image to be processed, determines the first outline area of the first bone and the second outline area of the second bone in the image to be processed, and locates them. A plurality of discrete points on the first contour area at the distal end of the first bone are projected onto the second contour area at the proximal end of the second bone. After obtaining a plurality of projection points corresponding to the plurality of discrete points one-to-one, based on the multiple The distance between the discrete points and the corresponding projection points of the plurality of discrete points determines the gap between the first bone and the second bone, since the plurality of discrete points located at the distal end of the first bone can be projected to the second contour area to determine the gap between the first bone and the second bone based on the distance between the discrete point and the projection point, thereby improving the accuracy of the determined bone gap. And, based on the determined gap between the first bone and the second bone, it can also be determined whether osteoarthritis exists between the first bone and the second bone.

The following is a description of the system for intelligently identifying osteoarthritis based on deep learning provided by the embodiments of the present application. The system for intelligently identifying osteoarthritis based on deep learning described below is different from the system for intelligently identifying osteoarthritis based on deep learning described above. Methods can be compared to each other.

Figure 5 is a schematic diagram of a system for intelligently identifying osteoarthritis based on deep learning provided by an embodiment of the present application. As shown in Figure 5, the device includes:

The acquisition module 11 is configured to acquire the image to be processed;

The determination module 12 is configured to determine the first outline area of the first bone and the second outline area of the second bone in the image to be processed;

The projection module 13 is configured to project a plurality of discrete points located on the first contour area at the distal end of the first bone onto the second contour area at the proximal end of the second bone, to obtain the same as the plurality of discrete points. Multiple projection points corresponding to discrete points one-to-one;

The determination module 12 is further configured to determine the gap between the first bone and the second bone based on the distance between the plurality of discrete points and the projection points corresponding to the plurality of discrete points. .

Optionally, the determination module 12 is specifically configured as:

Optionally, the determination module 12 is also configured to:

The device of this embodiment can be used to execute the method of any of the foregoing electronic device side method embodiments. Its specific implementation process and technical effects are similar to those in the electronic device side method embodiments. For details, please refer to Electronic Device Side Method Implementation The detailed introduction in the example will not be repeated here.

Figure 6 illustrates a schematic diagram of the physical structure of an electronic device. As shown in Figure 6, the electronic device may include: a processor (processor) 410, a communications interface (Communications Interface) 420, a memory (memory) 430 and a communication bus 440. Among them, the processor 410, the communication interface 420, and the memory 430 complete communication with each other through the communication bus 440. The processor 410 can call logical instructions in the memory 430 to perform a method for intelligently identifying osteoarthritis based on deep learning. The method includes: acquiring an image to be processed; and determining a first outline area of a first bone in the image to be processed. and the second contour area of the second bone; project a plurality of discrete points located on the first contour area at the distal end of the first bone onto the second contour area at the proximal end of the second bone to obtain the corresponding A plurality of projection points corresponding to the plurality of discrete points in a one-to-one manner; determining the first skeleton and the second skeleton according to the distance between the plurality of discrete points and the projection points corresponding to the plurality of discrete points. the gap between.

In addition, the above-mentioned logical instructions in the memory 430 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 existing technology or the part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including Several instructions are used to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program code. .

On the other hand, the present application also provides a computer program product. The computer program product includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can Execute the method for intelligently identifying osteoarthritis based on deep learning provided by each of the above methods. The method includes: acquiring an image to be processed; determining the first contour area of the first bone and the second contour area of the second bone in the image to be processed. Contour area: project a plurality of discrete points located on the first contour area at the distal end of the first bone onto the second contour area at the proximal end of the second bone to obtain a one-to-one relationship with the plurality of discrete points. Corresponding plurality of projection points; determine the gap between the first bone and the second bone according to the distance between the plurality of discrete points and the corresponding projection points of the plurality of discrete points.

On the other hand, the present application also provides a non-transitory computer-readable storage medium on which a computer program is stored. The computer program is implemented when executed by the processor to perform the intelligent identification of bone joints based on deep learning provided by the above methods. inflammatory method, the method includes: acquiring an image to be processed; determining a first outline area of a first bone and a second outline area of a second bone in the image to be processed; placing the first outline area located at the distal end of the first bone A plurality of discrete points on the contour area are projected onto the second contour area at the proximal end of the second bone to obtain a plurality of projection points corresponding to the plurality of discrete points; according to the plurality of discrete points and The distance between the corresponding projection points of the plurality of discrete points determines the gap between the first bone and the second bone.

The device embodiments described above are only illustrative. The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in One location, or it can be distributed across multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Persons of ordinary skill in the art can understand and implement the method without any creative effort.

Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the part of the above technical solution that essentially contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., including a number of instructions to cause a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various embodiments or certain parts of the embodiments.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present application, but not to limit it; 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 be Modifications are made to the technical solutions described in the foregoing embodiments, or equivalent substitutions are made to some of the technical features; however, these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions in the embodiments of the present application.

Claims

A method for intelligent identification of osteoarthritis based on deep learning, including:

Get the image to be processed;

Determine the first outline area of the first bone and the second outline area of the second bone in the image to be processed;

A plurality of discrete points located on the first contour area at the distal end of the first bone are projected onto the second contour area at the proximal end of the second bone to obtain a multi-point coordinate system corresponding to the plurality of discrete points in a one-to-one manner. projection points;

The gap between the first bone and the second bone is determined according to the distance between the plurality of discrete points and respective corresponding projection points of the plurality of discrete points.
The method for intelligently identifying osteoarthritis based on deep learning according to claim 1, wherein the determination of the said plurality of discrete points is based on the distance between the plurality of discrete points and the projection points corresponding to the plurality of discrete points. The gap between the first bone and the second bone includes:

Determine the distance between each discrete point in the plurality of discrete points and the projection point corresponding to each discrete point;

The smallest distance among multiple distances is determined as the gap between the first bone and the second bone.
The method for intelligently identifying osteoarthritis based on deep learning according to claim 1, wherein the determination of the said plurality of discrete points is based on the distance between the plurality of discrete points and the projection points corresponding to the plurality of discrete points. The gap between the first bone and the second bone includes:

Determine the distance between each discrete point in the plurality of discrete points and the projection point corresponding to each discrete point;

An average of the plurality of distances is determined as the gap between the first bone and the second bone.
The method for intelligently identifying osteoarthritis based on deep learning according to any one of claims 1-3, wherein the determining the first contour area of the first bone and the second contour area of the second bone in the image to be processed Contour areas, including:

The image to be processed is input into a pre-trained bone segmentation model to obtain the area where the first bone is located and the area where the second bone is located, wherein the bone segmentation model utilizes multiple The sample images are obtained after training, wherein the plurality of sample images are images containing different types of bones;

Contour extraction is performed on the area where the first bone is located and the area where the second bone is located, respectively, to obtain a first outline area of the first bone and a second outline area of the second bone.
The method for intelligently identifying osteoarthritis based on deep learning according to any one of claims 1-3, wherein the plurality of discrete points will be located on the first contour area of the distal end of the first bone, Before projecting onto the second contour area of the proximal end of the second bone to obtain a plurality of projection points corresponding to the plurality of discrete points, the method further includes:

determining a medial edge boundary point of the distal end of the first bone and a lateral edge boundary point of the distal end of the first bone;

A plurality of discrete points are determined on the first contour area of the distal end of the first bone between the medial edge boundary point and the lateral edge boundary point.
The method for intelligently identifying osteoarthritis based on deep learning according to claim 1, wherein after determining the gap between the first bone and the second bone, the method further includes:

Match the gap with multiple preset gap ranges and determine the target gap range in which the gap is located;

According to the target gap range, the degree of disease of the joint where the first bone and the second bone are located is determined.
A system for intelligent identification of osteoarthritis based on deep learning, including:

The acquisition module is configured to acquire the image to be processed;

a determination module configured to determine the first outline area of the first bone and the second outline area of the second bone in the image to be processed;

The projection module is configured to project a plurality of discrete points located on the first contour area at the distal end of the first bone to the second contour area at the proximal end of the second bone to obtain the results corresponding to the plurality of discrete points. Multiple projection points with one-to-one correspondence;

The determination module is further configured to determine the gap between the first bone and the second bone based on the distance between the plurality of discrete points and the corresponding projection points of the plurality of discrete points.
An electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the computer program as claimed in any one of claims 1 to 7 is implemented. Describes a method for intelligent identification of osteoarthritis based on deep learning.
A non-transitory computer-readable storage medium with a computer program stored thereon. When the computer program is executed by a processor, the intelligent identification of osteoarthritis based on deep learning as described in any one of claims 1 to 7 is realized. method.
A computer program product, comprising a computer program that, when executed by a processor, implements the method for intelligently identifying osteoarthritis based on deep learning as described in any one of claims 1 to 7.