CN113724328A - Hip joint key point detection method and system - Google Patents
Hip joint key point detection method and system Download PDFInfo
- Publication number
- CN113724328A CN113724328A CN202111014674.3A CN202111014674A CN113724328A CN 113724328 A CN113724328 A CN 113724328A CN 202111014674 A CN202111014674 A CN 202111014674A CN 113724328 A CN113724328 A CN 113724328A
- Authority
- CN
- China
- Prior art keywords
- hip joint
- key point
- point detection
- training
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30008—Bone
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the application discloses a hip joint key point detection method and a hip joint key point detection system, wherein the method comprises the following steps: preprocessing the acquired hip joint image data; marking hip joint key points on the preprocessed hip joint image to generate a hip joint data set; determining parameters before training of a hip joint key point detection model; determining a target loss function of key point detection, minimizing the target loss function through a gradient descent method, and training a hip joint key point detection model; and inputting the preprocessed hip joint image into the hip joint key point detection model to obtain the key point coordinates of the total hip joint. The characteristics around the key points are fused, and the hip joint key points are detected efficiently, so that the error rate of subjective judgment of a doctor can be reduced, and the method has high clinical significance and value.
Description
Technical Field
The embodiment of the application relates to the technical field of deep learning, in particular to a hip joint key point detection method and system.
Background
Artificial hip replacement is an important method for treating degenerative hip diseases, and precise and detailed preoperative planning is an indispensable link. Generally, this is done by the surgeon manually measuring the X-ray images, which is time consuming, labor intensive, and lacking in repeatability. Therefore, it is very important to find a method for automatically measuring various indexes before the operation.
The anatomical landmark points of medical images are the prerequisite for various clinical applications such as image registration, segmentation, auxiliary diagnosis and the like. It is difficult to accurately detect the anatomical landmark points due to differences between patients and differences in photographing angles and apparatuses. Conventional image processing typically classifies pixels of an image to determine anatomical landmarks.
With the development of deep learning, some have detected a plurality of key points in the brain MR images by using two CNN models, the first CNN model is to detect the learning association between the local image blocks and the key points, and the second CNN model is to predict the learning association of the key points. A full convolution neural network is proposed, which combines regression and classification models to regress the displacement in the CTA sequence to a target point. However, the key points of these methods are independent and do not take into account information around the key points.
Disclosure of Invention
Therefore, the embodiment of the application provides a method and a system for detecting hip joint key points, features around the key points are fused, and the hip joint key points are detected efficiently, so that the error rate of subjective judgment of doctors can be reduced, and the method and the system have high clinical significance and value.
In order to achieve the above object, the embodiments of the present application provide the following technical solutions:
according to a first aspect of embodiments of the present application, there is provided a hip joint keypoint detection method, the method comprising:
preprocessing the acquired hip joint image data;
marking hip joint key points on the preprocessed hip joint image to generate a hip joint data set;
determining parameters before training of a hip joint key point detection model;
determining a target loss function of key point detection, minimizing the target loss function through a gradient descent method, and training a hip joint key point detection model;
and inputting the preprocessed hip joint image into the hip joint key point detection model to obtain the key point coordinates of the total hip joint.
Optionally, the hip key point labeling on the preprocessed hip image to generate a hip data set includes:
marking key points of a fixed number of hip joint images, and recording the meaning of each key point;
normalizing the marked key point coordinates to obtain a hip joint data set required by the training model;
and dividing the hip joint data set into a training set, a verification set and a test set according to a set proportion.
Optionally, the determining parameters of the hip joint key point detection model before training includes:
adding edge feature information between points;
randomly initializing parameters of a key point detection model, wherein the parameters comprise a learning rate and an attenuation coefficient;
and modifying the output parameters of the model, and changing the output category parameters into the number of the key points.
Optionally, the determining a target loss function for keypoint detection, minimizing the target loss function by a gradient descent method, and training a hip joint keypoint detection model, including:
determining a target loss function of key point detection; the target loss function comprises loss of key points and loss of side information; the target loss function takes the Euclidean distance between the output characteristic diagram of the neural network and the target characteristic diagram as model training error information to carry out back propagation;
outputting probability confidence maps of all key points of the total hip joint, and obtaining two-dimensional coordinates of hip joint points according to the positions of maximum probability pixel points in the probability confidence maps;
in the model training process, a gradient descent optimization algorithm is used, and end-to-end model training is carried out by using a training sample; meanwhile, the information of the side is used as a constraint condition; after multiple rounds of training, loss is converged and minimized, and the hip joint key point detection model is obtained.
Optionally, the inputting the preprocessed hip joint image into the hip joint key point detection model to obtain key point coordinates of a total hip joint includes:
inputting the preprocessed hip joint image into the hip joint key point detection model, and carrying out coordinate reduction based on a normalization method according to a result output by the model to obtain the key point coordinates of the total hip joint.
According to a second aspect of embodiments of the present application, there is provided a hip joint keypoint detection system, the system comprising:
the preprocessing module is used for preprocessing the acquired hip joint image data;
the marking module is used for marking hip joint key points of the preprocessed hip joint image to generate a hip joint data set;
the parameter determination module is used for determining parameters before the hip joint key point detection model is trained;
the model training module is used for determining a target loss function of key point detection, minimizing the target loss function through a gradient descent method, and training a hip joint key point detection model;
and the key point detection module is used for inputting the preprocessed hip joint image into the hip joint key point detection model to obtain the key point coordinates of the total hip joint.
Optionally, the marking module is specifically configured to:
marking key points of a fixed number of hip joint images, and recording the meaning of each key point;
normalizing the marked key point coordinates to obtain a hip joint data set required by the training model;
and dividing the hip joint data set into a training set, a verification set and a test set according to a set proportion.
Optionally, the parameter determining module is specifically configured to:
adding edge feature information between points;
randomly initializing parameters of a key point detection model, wherein the parameters comprise a learning rate and an attenuation coefficient;
and modifying the output parameters of the model, and changing the output category parameters into the number of the key points.
Optionally, the model training module is specifically configured to:
determining a target loss function of key point detection; the target loss function comprises loss of key points and loss of side information; the target loss function takes the Euclidean distance between the output characteristic diagram of the neural network and the target characteristic diagram as model training error information to carry out back propagation;
outputting probability confidence maps of all key points of the total hip joint, and obtaining two-dimensional coordinates of hip joint points according to the positions of maximum probability pixel points in the probability confidence maps;
in the model training process, a gradient descent optimization algorithm is used, and end-to-end model training is carried out by using a training sample; meanwhile, the information of the side is used as a constraint condition; after multiple rounds of training, loss is converged and minimized, and the hip joint key point detection model is obtained.
Optionally, the key point detection module is specifically configured to:
inputting the preprocessed hip joint image into the hip joint key point detection model, and carrying out coordinate reduction based on a normalization method according to a result output by the model to obtain the key point coordinates of the total hip joint.
According to a third aspect of embodiments herein, there is provided an apparatus comprising: the device comprises a data acquisition device, a processor and a memory; the data acquisition device is used for acquiring data; the memory is to store one or more program instructions; the processor is configured to execute one or more program instructions to perform the method of any of the first aspect.
According to a fourth aspect of embodiments herein, there is provided a computer-readable storage medium having one or more program instructions embodied therein for performing the method of any of the first aspects.
In summary, the embodiment of the present application provides a method and a system for detecting hip joint key points, which preprocess acquired hip joint image data; marking hip joint key points on the preprocessed hip joint image to generate a hip joint data set; determining parameters before training of a hip joint key point detection model; determining a target loss function of key point detection, minimizing the target loss function through a gradient descent method, and training a hip joint key point detection model; and inputting the preprocessed hip joint image into the hip joint key point detection model to obtain the key point coordinates of the total hip joint. The characteristics around the key points are fused, and the hip joint key points are detected efficiently, so that the error rate of subjective judgment of a doctor can be reduced, and the method has high clinical significance and value.
Drawings
In order to more clearly illustrate the implementation of the embodiments of the present application or the technical solutions in the prior art, the drawings used in the description of the implementation or the prior art will be briefly described below. It should be apparent that the drawings in the following description are merely exemplary, and that other embodiments can be derived from the drawings provided by those of ordinary skill in the art without inventive effort.
The structures, ratios, sizes, and the like shown in the present specification are only used for matching with the contents disclosed in the present specification, so that those skilled in the art can understand and read the present disclosure, and do not limit the conditions that the embodiments of the present application can be implemented, so that the present disclosure has no technical significance, and any structural modifications, changes in the ratio relationship, or adjustments of the size should still fall within the scope that the technical contents disclosed in the embodiments of the present application can cover without affecting the efficacy and the achievable purpose that the embodiments of the present application can be implemented.
Fig. 1 is a schematic flow chart of a hip joint key point detection method according to an embodiment of the present disclosure;
FIG. 2 is a schematic diagram of a keypoint model provided in an embodiment of the present application;
FIG. 3 is a diagram illustrating test results of a model provided in an embodiment of the present application;
fig. 4 is a block diagram of a hip joint key point detection system provided in an embodiment of the present application.
Detailed Description
Other advantages and features of the embodiments of the present application will become apparent to those skilled in the art from the following description, wherein it is to be understood that the embodiments of the present application are described in connection with the particular illustrative embodiments thereof, and that the embodiments of the present application are not limited to the particular embodiments disclosed herein. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present application without any creative effort belong to the protection scope of the embodiments in the present application.
Fig. 1 shows a hip joint key point detection method provided by the embodiment of the present application, which fuses features around key points and is used for detecting hip joint key points, and specifically includes the following steps:
step 101: preprocessing the acquired hip joint image data;
step 102: marking hip joint key points on the preprocessed hip joint image to generate a hip joint data set;
step 103: determining parameters before training of a hip joint key point detection model;
step 104: determining a target loss function of key point detection, minimizing the target loss function through a gradient descent method, and training a hip joint key point detection model;
step 105: and inputting the preprocessed hip joint image into the hip joint key point detection model to obtain the key point coordinates of the total hip joint.
In one possible embodiment, in step 102, the hip key point labeling on the preprocessed hip image to generate a hip data set includes: marking key points of a fixed number of hip joint images, and recording the meaning of each key point; normalizing the marked key point coordinates to obtain a hip joint data set required by the training model; and dividing the hip joint data set into a training set, a verification set and a test set according to a set proportion.
In one possible embodiment, in step 103, the determining parameters of the hip joint key point detection model before training includes: adding edge feature information between points; randomly initializing parameters of a key point detection model, wherein the parameters comprise a learning rate and an attenuation coefficient; and modifying the output parameters of the model, and changing the output category parameters into the number of the key points.
In a possible implementation manner, in step 104, the determining a target loss function for keypoint detection, minimizing the target loss function by a gradient descent method, and training a hip joint keypoint detection model, including: determining a target loss function of key point detection; the target loss function comprises loss of key points and loss of side information; the target loss function takes the Euclidean distance between the output characteristic diagram of the neural network and the target characteristic diagram as model training error information to carry out back propagation; outputting probability confidence maps of all key points of the total hip joint, and obtaining two-dimensional coordinates of hip joint points according to the positions of maximum probability pixel points in the probability confidence maps; in the model training process, a gradient descent optimization algorithm is used, and end-to-end model training is carried out by using a training sample; meanwhile, the information of the side is used as a constraint condition; after multiple rounds of training, loss is converged and minimized, and the hip joint key point detection model is obtained.
In a possible implementation manner, in step 105, the inputting the preprocessed hip image into the hip key point detection model to obtain key point coordinates of the total hip joint includes: inputting the preprocessed hip joint image into the hip joint key point detection model, and carrying out coordinate reduction based on a normalization method according to a result output by the model to obtain the key point coordinates of the total hip joint.
A first object of an embodiment of the present application is to acquire hip image data and pre-process the image. The technical scheme for realizing the first purpose of the embodiment of the application is as follows: image preprocessing of hip joint X-ray film: and adjusting the window level of the hip joint X-ray dicom original image, and converting the window level into an image JPG format to obtain a hip joint image.
A second object of an embodiment of the present application is: and marking hip joint key points to generate a data set required by the model. The method specifically comprises the following steps:
s21: and marking the acquired jpg image with key points of the total hip joint, wherein the number of the key points needs to be fixed in the marking process, otherwise, the model cannot train the variable number of the key points, and simultaneously recording the specific meaning represented by each point.
S22: sample image normalization: and normalizing the marked key point coordinates, wherein the normalization needs to be processed into 512 by 512 size, and a data set required by the model is obtained.
S23: data set partitioning: the data set is divided into a training set, a verification set and a test set according to the ratio of 8:1: 1.
A third object of an embodiment of the present application is: and designing parameters before model training. The technical scheme for realizing the third purpose of the embodiment of the application comprises the following steps:
s31: adding edge information between points and thus can be expressed for edge features as
Where vi is a 128-dimensional feature vector.
S32: the parameters of the keypoint detection model are randomly initialized. If the learning rate is 0.001, the batch _ size is 32, and the attenuation coefficient is 0.0001.
S33: and modifying the output parameters of the model, and changing the output category into the number of the key points.
S34: model design: and (4) taking the residual error network as a foundation stone, and connecting the sub-networks with multiple resolutions in parallel to realize multi-scale fusion.
A fourth object of an embodiment of the present application is: and designing a loss function of key point detection, minimizing the loss function by a gradient descent method, and obtaining a model of key point detection capability. As shown in fig. 2, the technical solution for achieving the fourth object of the embodiment of the present application includes the following steps:
s41: designing a loss function of key point detection: l ═ Lkeypoints+λ*Ledges。
Wherein: the loss function consists of two parts, one is the loss of keypoints and the other is the loss of side information. And the target loss function takes the Euclidean distance between the output characteristic diagram of the neural network and the target characteristic diagram as model training error information to carry out back propagation.
S42: and (3) outputting a model: and outputting the probability confidence maps of all key points of the total hip joint, and obtaining the two-dimensional coordinates of the hip joint points according to the positions of the maximum probability pixel points in the probability confidence maps.
S43: training process: and performing end-to-end model training by using the training sample by using a gradient descent optimization algorithm. Meanwhile, the information of the side is used as a constraint condition. And finally, after multiple rounds of training, the loss is converged and minimized, and a final key point model is obtained. At this point, the model has the capability of keypoint detection.
A fifth object of an embodiment of the present application is: and automatically acquiring key points of the total hip joint. The technical scheme for realizing the fifth purpose of the embodiment of the application comprises the following steps:
s51: from the model obtained at S43, taking the image preprocessed at S1 as input, and obtaining a final output result through the model at S43;
s52: based on the result of S51, the coordinates are restored by the normalization method of S22 to the original image position, and at this time, the key point coordinates of the hip joint can be obtained.
In the specific implementation mode, 300 cases are tested, and 92.25% is finally obtained, as shown in fig. 3, so that the embodiment of the application is effective in detecting most hip joint key points, automatic detection can be realized, and the working efficiency of doctors is greatly improved.
The embodiment of the application provides a hip joint key point detection method. The method can detect the regional characteristics of the key points and simultaneously increase the edge information between the points. By means of the edge information, the key points of the deviation can be corrected through the edge information, and therefore accuracy is improved. The output hip joint key points can be used for preoperative measured value calculation, such as neck shaft angle, medullary cavity flicker index, femoral head rotation center and diameter and the like. The method provided by the embodiment of the application provides a theoretical basis for realizing automatic preoperative measurement and prosthesis planning.
In summary, the embodiment of the present application provides a method for detecting hip joint key points, which includes preprocessing acquired hip joint image data; marking hip joint key points on the preprocessed hip joint image to generate a hip joint data set; determining parameters before training of a hip joint key point detection model; determining a target loss function of key point detection, minimizing the target loss function through a gradient descent method, and training a hip joint key point detection model; and inputting the preprocessed hip joint image into the hip joint key point detection model to obtain the key point coordinates of the total hip joint. The characteristics around the key points are fused, and the hip joint key points are detected efficiently, so that the error rate of subjective judgment of a doctor can be reduced, and the method has high clinical significance and value.
Based on the same technical concept, the embodiment of the present application further provides a hip joint key point detection system, as shown in fig. 4, the system includes:
a preprocessing module 401, configured to preprocess the acquired hip joint image data;
a marking module 402, configured to mark hip key points on the preprocessed hip image to generate a hip data set;
a parameter determining module 403, configured to determine parameters before training of the hip joint key point detection model;
a model training module 404, configured to determine a target loss function for key point detection, minimize the target loss function by using a gradient descent method, and train a hip joint key point detection model;
and a key point detection module 405, configured to input the preprocessed hip joint image into the hip joint key point detection model, so as to obtain a key point coordinate of the total hip joint.
In a possible implementation, the marking module 402 is specifically configured to: marking key points of a fixed number of hip joint images, and recording the meaning of each key point; normalizing the marked key point coordinates to obtain a hip joint data set required by the training model; and dividing the hip joint data set into a training set, a verification set and a test set according to a set proportion.
In a possible implementation manner, the parameter determining module 403 is specifically configured to: adding edge feature information between points; randomly initializing parameters of a key point detection model, wherein the parameters comprise a learning rate and an attenuation coefficient; and modifying the output parameters of the model, and changing the output category parameters into the number of the key points.
In a possible implementation manner, the model training module 404 is specifically configured to: determining a target loss function of key point detection; the target loss function comprises loss of key points and loss of side information; the target loss function takes the Euclidean distance between the output characteristic diagram of the neural network and the target characteristic diagram as model training error information to carry out back propagation; outputting probability confidence maps of all key points of the total hip joint, and obtaining two-dimensional coordinates of hip joint points according to the positions of maximum probability pixel points in the probability confidence maps; in the model training process, a gradient descent optimization algorithm is used, and end-to-end model training is carried out by using a training sample; meanwhile, the information of the side is used as a constraint condition; after multiple rounds of training, loss is converged and minimized, and the hip joint key point detection model is obtained.
In a possible implementation manner, the key point detecting module 405 is specifically configured to: inputting the preprocessed hip joint image into the hip joint key point detection model, and carrying out coordinate reduction based on a normalization method according to a result output by the model to obtain the key point coordinates of the total hip joint.
Based on the same technical concept, an embodiment of the present application further provides an apparatus, including: the device comprises a data acquisition device, a processor and a memory; the data acquisition device is used for acquiring data; the memory is to store one or more program instructions; the processor is configured to execute one or more program instructions to perform the method.
Based on the same technical concept, the embodiment of the present application also provides a computer-readable storage medium, wherein the computer-readable storage medium contains one or more program instructions, and the one or more program instructions are used for executing the method.
In the present specification, each embodiment of the method is described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. Reference is made to the description of the method embodiments.
It should be noted that although the operations of the methods of the embodiments of the present application are depicted in the drawings in a particular order, this does not require or imply that these operations must be performed in this particular order, or that all of the illustrated operations must be performed, to achieve desirable results. Additionally or alternatively, certain steps may be omitted, multiple steps combined into one step execution, and/or one step broken down into multiple step executions.
Although the present application provides method steps as in embodiments or flowcharts, additional or fewer steps may be included based on conventional or non-inventive approaches. The order of steps recited in the embodiments is merely one manner of performing the steps in a multitude of orders and does not represent the only order of execution. When an apparatus or client product in practice executes, it may execute sequentially or in parallel (e.g., in a parallel processor or multithreaded processing environment, or even in a distributed data processing environment) according to the embodiments or methods shown in the figures. The terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, the presence of additional identical or equivalent elements in a process, method, article, or apparatus that comprises the recited elements is not excluded.
The units, devices, modules, etc. set forth in the above embodiments may be implemented by a computer chip or an entity, or by a product with certain functions. For convenience of description, the above devices are described as being divided into various modules by functions, and are described separately. Of course, in implementing the present application, the functions of each module may be implemented in one or more software and/or hardware, or a module implementing the same function may be implemented by a combination of a plurality of sub-modules or sub-units, and the like. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
Those skilled in the art will also appreciate that, in addition to implementing the controller as pure computer readable program code, the same functionality can be implemented by logically programming method steps such that the controller is in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers and the like. Such a controller may therefore be considered as a hardware component, and the means included therein for performing the various functions may also be considered as a structure within the hardware component. Or even means for performing the functions may be regarded as being both a software module for performing the method and a structure within a hardware component.
The application may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, classes, etc. that perform particular tasks or implement particular abstract data types. The application may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.
From the above description of the embodiments, it is clear to those skilled in the art that the present application can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solutions of the present application may be embodied in the form of a software product, which may be stored in a storage medium, such as a ROM/RAM, a magnetic disk, an optical disk, or the like, and includes several instructions for enabling a computer device (which may be a personal computer, a mobile terminal, a server, or a network device) to execute the method according to the embodiments or some parts of the embodiments of the present application.
The embodiments in the present specification are described in a progressive manner, and the same or similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. The application is operational with numerous general purpose or special purpose computing system environments or configurations. For example: personal computers, server computers, hand-held or portable devices, tablet-type devices, multiprocessor systems, microprocessor-based systems, set top boxes, programmable electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments that include any of the above systems or devices, and the like.
The above-mentioned embodiments are further described in detail for the purpose of illustrating the invention, and it should be understood that the above-mentioned embodiments are only illustrative of the present invention and are not intended to limit the scope of the present invention, and any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (10)
1. A method of hip key detection, the method comprising:
preprocessing the acquired hip joint image data;
marking hip joint key points on the preprocessed hip joint image to generate a hip joint data set;
determining parameters before training of a hip joint key point detection model;
determining a target loss function of key point detection, minimizing the target loss function through a gradient descent method, and training a hip joint key point detection model;
and inputting the preprocessed hip joint image into the hip joint key point detection model to obtain the key point coordinates of the total hip joint.
2. The method of claim 1, wherein the hip keypoint labeling of the pre-processed hip image to generate a hip data set comprises:
marking key points of a fixed number of hip joint images, and recording the meaning of each key point;
normalizing the marked key point coordinates to obtain a hip joint data set required by the training model;
and dividing the hip joint data set into a training set, a verification set and a test set according to a set proportion.
3. The method of claim 1, wherein determining parameters of the hip keypoint detection model prior to training comprises:
adding edge feature information between points;
randomly initializing parameters of a key point detection model, wherein the parameters comprise a learning rate and an attenuation coefficient;
and modifying the output parameters of the model, and changing the output category parameters into the number of the key points.
4. The method of claim 1, wherein determining the target loss function for keypoint detection, minimizing the target loss function by a gradient descent method, and training a hip keypoint detection model comprises:
determining a target loss function of key point detection; the target loss function comprises loss of key points and loss of side information; the target loss function takes the Euclidean distance between the output characteristic diagram of the neural network and the target characteristic diagram as model training error information to carry out back propagation;
outputting probability confidence maps of all key points of the total hip joint, and obtaining two-dimensional coordinates of hip joint points according to the positions of maximum probability pixel points in the probability confidence maps;
in the model training process, a gradient descent optimization algorithm is used, and end-to-end model training is carried out by using a training sample; meanwhile, the information of the side is used as a constraint condition; after multiple rounds of training, loss is converged and minimized, and the hip joint key point detection model is obtained.
5. The method of claim 1, wherein inputting the preprocessed hip image into the hip keypoint detection model to obtain keypoint coordinates of the total hip joint comprises:
inputting the preprocessed hip joint image into the hip joint key point detection model, and carrying out coordinate reduction based on a normalization method according to a result output by the model to obtain the key point coordinates of the total hip joint.
6. A hip joint keypoint detection system, the system comprising:
the preprocessing module is used for preprocessing the acquired hip joint image data;
the marking module is used for marking hip joint key points of the preprocessed hip joint image to generate a hip joint data set;
the parameter determination module is used for determining parameters before the hip joint key point detection model is trained;
the model training module is used for determining a target loss function of key point detection, minimizing the target loss function through a gradient descent method, and training a hip joint key point detection model;
and the key point detection module is used for inputting the preprocessed hip joint image into the hip joint key point detection model to obtain the key point coordinates of the total hip joint.
7. The system of claim 6, wherein the tagging module is specifically configured to:
marking key points of a fixed number of hip joint images, and recording the meaning of each key point;
normalizing the marked key point coordinates to obtain a hip joint data set required by the training model;
and dividing the hip joint data set into a training set, a verification set and a test set according to a set proportion.
8. The system of claim 6, wherein the parameter determination module is specifically configured to:
adding edge feature information between points;
randomly initializing parameters of a key point detection model, wherein the parameters comprise a learning rate and an attenuation coefficient;
and modifying the output parameters of the model, and changing the output category parameters into the number of the key points.
9. The system of claim 6, wherein the model training module is specifically configured to:
determining a target loss function of key point detection; the target loss function comprises loss of key points and loss of side information; the target loss function takes the Euclidean distance between the output characteristic diagram of the neural network and the target characteristic diagram as model training error information to carry out back propagation;
outputting probability confidence maps of all key points of the total hip joint, and obtaining two-dimensional coordinates of hip joint points according to the positions of maximum probability pixel points in the probability confidence maps;
in the model training process, a gradient descent optimization algorithm is used, and end-to-end model training is carried out by using a training sample; meanwhile, the information of the side is used as a constraint condition; after multiple rounds of training, loss is converged and minimized, and the hip joint key point detection model is obtained.
10. The system of claim 6, wherein the keypoint detection module is specifically configured to:
inputting the preprocessed hip joint image into the hip joint key point detection model, and carrying out coordinate reduction based on a normalization method according to a result output by the model to obtain the key point coordinates of the total hip joint.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111014674.3A CN113724328A (en) | 2021-08-31 | 2021-08-31 | Hip joint key point detection method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111014674.3A CN113724328A (en) | 2021-08-31 | 2021-08-31 | Hip joint key point detection method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113724328A true CN113724328A (en) | 2021-11-30 |
Family
ID=78679988
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111014674.3A Pending CN113724328A (en) | 2021-08-31 | 2021-08-31 | Hip joint key point detection method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113724328A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114663363A (en) * | 2022-03-03 | 2022-06-24 | 四川大学 | Hip joint medical image processing method and device based on deep learning |
| CN115239720A (en) * | 2022-09-22 | 2022-10-25 | 安徽省儿童医院(安徽省新华医院、安徽省儿科医学研究所、复旦大学附属儿科医院安徽医院) | Classical Graf-based DDH ultrasonic image artificial intelligence diagnosis system and method |
-
2021
- 2021-08-31 CN CN202111014674.3A patent/CN113724328A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114663363A (en) * | 2022-03-03 | 2022-06-24 | 四川大学 | Hip joint medical image processing method and device based on deep learning |
| CN114663363B (en) * | 2022-03-03 | 2023-11-17 | 四川大学 | Deep learning-based hip joint medical image processing method and device |
| CN115239720A (en) * | 2022-09-22 | 2022-10-25 | 安徽省儿童医院(安徽省新华医院、安徽省儿科医学研究所、复旦大学附属儿科医院安徽医院) | Classical Graf-based DDH ultrasonic image artificial intelligence diagnosis system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Forsberg et al. | Detection and labeling of vertebrae in MR images using deep learning with clinical annotations as training data | |
| US11589928B2 (en) | Artificial intelligence intra-operative surgical guidance system and method of use | |
| Wang et al. | Automatic analysis of lateral cephalograms based on multiresolution decision tree regression voting | |
| CN111160367B (en) | Image classification method, apparatus, computer device, and readable storage medium | |
| Zhou et al. | Automatic lumbar MRI detection and identification based on deep learning | |
| CN113724328A (en) | Hip joint key point detection method and system | |
| CN111488872B (en) | Image detection method, image detection device, computer equipment and storage medium | |
| CN113592820A (en) | Method and system for detecting femoral region and key points | |
| KR102228087B1 (en) | Method and apparatus for segmentation of specific cartilage in medical image | |
| JP7346553B2 (en) | Determining the growth rate of objects in a 3D dataset using deep learning | |
| US20230005138A1 (en) | Lumbar spine annatomical annotation based on magnetic resonance images using artificial intelligence | |
| Zhang et al. | Learning-based coronal spine alignment prediction using smartphone-acquired scoliosis radiograph images | |
| Zhao et al. | Automatic Cobb angle measurement method based on vertebra segmentation by deep learning | |
| JP5704909B2 (en) | Attention area detection method, attention area detection apparatus, and program | |
| KR102127151B1 (en) | Lower extremities bone image processing apparatus and method | |
| CN110473193A (en) | Detection method, detection device, storage medium and the processor of acetabular bone defect | |
| CN115634044A (en) | Operation planning and model training method and device and electronic equipment | |
| Song et al. | Semi-automatic cephalometric landmark detection on x-ray images using deep learning method | |
| CN113962957A (en) | Medical image processing method, bone image processing method, device and equipment | |
| CN113724319A (en) | Artificial hip joint prosthesis model matching method and system | |
| CN113392895A (en) | Knee joint cartilage damage detection method and system | |
| Zhang et al. | Intelligent measurement of spinal curvature using cascade gentle AdaBoost classifier and region-based DRLSE | |
| KR102567497B1 (en) | spinal pelvic balance measurement apparatus based on artificial neural network and method therefor | |
| Neeraja et al. | A Review on Automatic Cephalometric Landmark Identification Using Artificial Intelligence Techniques | |
| Zeng et al. | A novel solder joints inspection method using curvature and geometry features in high-density flexible IC substrates surface mount technology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |