Method and system for measuring rear foot force line angle through X rays based on deep learning

Abstract

The invention discloses a deep learning-based X-ray rear foot force line angle measurement method and system, and relates to the technical field of medical detection. The method comprises the following steps: acquiring and processing a rear foot force line position sample image to construct a training data set; constructing a convolutional neural network model, and training the convolutional neural network model based on the training data set to obtain a final convolutional neural network model; obtaining and inputting a posterior foot force line position X-ray film of the patient into the final convolutional neural network model for segmentation processing, and outputting a segmentation result; and fitting an anatomical axis according to a segmentation result, and calculating a rear foot force line angle according to the anatomical axis. According to the method, rapid and accurate diagnosis of the posterior foot deformity based on the X-ray image and artificial intelligence auxiliary diagnosis can be realized.

Description

Technical Field

[0001] The present invention relates to the field of medical detection technology, and in particular to a method and system for measuring hindfoot force line angle by X-ray based on deep learning. Background Art

[0002] Hindfoot deformity is one of the most common causes of foot pain. The hindfoot force line angle is one of the important indicators for evaluating patients with hindfoot deformity. Its measurement method and accuracy are of great significance for clinical diagnosis and treatment. Imaging examination is one of the main methods for evaluating the hindfoot force line angle. Through imaging methods such as X-ray and CT scanning, the bone structure of the ankle and hindfoot can be observed in detail, so as to measure the hindfoot force line angle. For example, X-rays can show the bone structure of the ankle and hindfoot, while CT and its three-dimensional reconstruction technology can provide more detailed anatomical information, which helps to more accurately evaluate the hindfoot force line angle. In addition, improved foot force line shooting and measurement methods, such as the modified THA method, have improved the accuracy and repeatability of the measurement, providing a more reliable reference for clinical practice.

[0003] Imaging is an important means of diagnosing foot deformities. Although it plays a certain role, it is currently generally combined with manual measurement of the hindfoot force line angle, and its consistency is poor. Summary of the invention

[0004] In order to overcome the above problems or at least partially solve the above problems, the present invention provides a method and system for measuring the hindfoot force angle by X-ray based on deep learning, which can realize rapid and accurate diagnosis of hindfoot deformity based on X-ray images and artificial intelligence-assisted diagnosis.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0006] In a first aspect, the present invention provides a method for measuring the hindfoot force line angle by X-ray based on deep learning, comprising the following steps:

[0007] Acquire and process sample images of hindfoot force line positions to construct a training data set;

[0008] Construct a convolutional neural network model and train the convolutional neural network model based on the training data set to obtain the final convolutional neural network model;

[0009] Obtain and input the patient's hindfoot force line X-ray film into the final convolutional neural network model for segmentation processing and output the segmentation result;

[0010] The anatomical axis was fitted according to the segmentation results, and the hindfoot force line angle was calculated based on the anatomical axis.

[0011] This method acquires and processes a large number of sample images to construct a reasonable data set and an effective convolutional neural network model, so as to perform accurate and efficient image segmentation in the future. The tibia and calcaneus are segmented in the hindfoot force line X-ray film, and the anatomical axes are fitted respectively. Finally, the hindfoot force line angle is calculated according to the fitting results, so as to realize rapid and accurate diagnosis of hindfoot deformity based on X-ray images and artificial intelligence-assisted diagnosis.

[0012] Based on the first aspect, further, the method for acquiring and processing the rear foot force line position sample images to construct a training data set includes the following steps:

[0013] Obtain N rear foot force line position sample images, where: N is a natural number greater than or equal to 1;

[0014] Mark I region of interest contours Ri for each rear foot force line position sample image, and generate a corresponding mask image, where: 1≤i≤I, and I is a natural number; the mask image is composed of I binary mask rectangles that assign different representative values ​​i to the coordinate points in each region of interest contour Ri and have the same size as the original rear foot force line position sample image;

[0015] The N mask images are paired with N hindfoot force line position sample images to form a training data set.

[0016] Based on the first aspect, further, the method for constructing a convolutional neural network model comprises the following steps:

[0017] Construct a convolutional neural network model and set the hyperparameters of the convolutional neural network model, where the input channel of the convolutional neural network model is 1, the output channel is 1×(I+1), and the output corresponds to the detection probability map of the contours of I regions of interest and the detection probability map of the background.

[0018] Based on the first aspect, further, the method for training the convolutional neural network model based on the training data set to obtain the final convolutional neural network model includes the following steps:

[0019] Divide the training data set into training set, validation set and test set;

[0020] The training set is input into the convolutional neural network model in batches for training. The difference between the output image and the mask image is calculated using the loss function as the training error and fed back to the convolutional neural network model. The model parameters are updated through the learning algorithm.

[0021] After each batch of training is completed, the validation set is used to perform performance testing on the convolutional neural network model until the performance test index reaches the preset stable value. The convolutional neural network model training is then considered complete and saved as the final convolutional neural network model.

[0022] Based on the first aspect, further, the method of inputting the patient's hindfoot force line X-ray film into the final convolutional neural network model for segmentation processing includes the following steps:

[0023] The patient's hindfoot force line X-ray film is input into the final convolutional neural network model, and the region of interest of the hindfoot force line X-ray film is segmented based on the convolutional neural network model, and the segmentation result with a dimension of 1×(I+1) is output. The segmentation result includes a probability map corresponding to I regions of interest in one input channel and a background probability map.

[0024] Based on the first aspect, further, the method for measuring the hindfoot force line angle by X-ray based on deep learning further includes the following steps:

[0025] Binarizing the probability map according to a preset probability threshold to obtain a binary mask image;

[0026] For the binary mask image obtained for the i-th region of interest except the background probability map, retain the first two connected domains with the largest area, and ensure that the two connected domains are located on both sides of the vertical bisector of the output probability map as the segmentation result of the i-th region of interest.

[0027] Based on the first aspect, further, the method for fitting the anatomical axis according to the segmentation result comprises the following steps:

[0028] Based on the segmentation results of I regions of interest, the anatomical axes are calculated for the first two connected domains respectively to obtain the fitting axis of the first connected domain and the fitting axis of the second connected domain.

[0029] In a second aspect, the present invention provides a system for measuring hindfoot force line angle by X-ray based on deep learning, comprising a training set construction module, a model training module, an image segmentation module and a hindfoot force line angle calculation module, wherein:

[0030] A training set construction module is used to obtain and process sample images of hindfoot force line positions to construct a training data set;

[0031] The model training module is used to build a convolutional neural network model and train the convolutional neural network model based on the training data set to obtain the final convolutional neural network model;

[0032] An image segmentation module is used to obtain and input the patient's hindfoot force line X-ray film into the final convolutional neural network model for segmentation processing and output the segmentation result;

[0033] The hindfoot force line angle calculation module is used to fit the anatomical axis according to the segmentation results and calculate the hindfoot force line angle according to the anatomical axis.

[0034] This system is based on the cooperation of multiple modules such as the training set construction module, model training module, image segmentation module and hindfoot force line angle calculation module. By acquiring and processing a large number of sample images, a reasonable data set is constructed, and an effective convolutional neural network model is constructed to facilitate subsequent accurate and efficient image segmentation. The hindfoot force line X-ray film is segmented into the tibia and calcaneus, and the anatomical axes are fitted respectively. Finally, the hindfoot force line angle is calculated according to the fitting results, so as to realize rapid and accurate diagnosis of hindfoot deformity based on X-ray images and artificial intelligence-assisted diagnosis.

[0035] In a third aspect, the present application provides an electronic device, comprising a memory for storing one or more programs; a processor; when the one or more programs are executed by the processor, the method as described in any one of the first aspects above is implemented.

[0036] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method as described in any one of the first aspects above.

[0037] The present invention has at least the following advantages or beneficial effects:

[0038] The present invention provides a method and system for measuring the hindfoot force line angle by X-ray based on deep learning. A large number of sample images are acquired and processed to construct a reasonable data set, and an effective convolutional neural network model is constructed to facilitate subsequent accurate and efficient image segmentation. The hindfoot force line X-ray film is segmented into the tibia and calcaneus, and the anatomical axes are fitted respectively. Finally, the hindfoot force line angle is calculated according to the fitting results, so as to realize rapid and accurate diagnosis of hindfoot deformity based on X-ray images and artificial intelligence-assisted diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0040] Figure 1 This is a flow chart of a method for measuring hindfoot force line angle by X-ray based on deep learning according to an embodiment of the present invention;

[0041] Figure 2 This is a principle block diagram of a system for measuring hindfoot force angle by X-ray based on deep learning according to an embodiment of the present invention;

[0042] Figure 3 It is a schematic diagram of the anatomical axis of the foot force line X-ray film based on the anatomical definition method in an embodiment of the present invention;

[0043] Figure 4 It is a schematic diagram of anatomical axis fitting of the foot force line X-ray film based on the algorithm fitting method in an embodiment of the present invention;

[0044] Figure 5 A structural block diagram of an electronic device provided by an embodiment of the present invention.

[0045] Explanation of the accompanying drawings: 100, training set construction module; 200, model training module; 300, image segmentation module; 400, rear foot force line angle calculation module; 101, memory; 102, processor; 103, communication interface. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0047] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0049] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the term "comprise" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0050] In the description of the embodiments of the present invention, "plurality" means at least 2.

[0051] Example:

[0052] like Figure 1 As shown, in a first aspect, an embodiment of the present invention provides a method for measuring the hindfoot force line angle by X-ray based on deep learning, comprising the following steps:

[0053] S1, obtaining and processing sample images of hindfoot force line positions to construct a training data set;

[0054] Further, it includes: obtaining N hindfoot force line position sample images, wherein: N is a natural number greater than or equal to 1; marking I region of interest contours Ri for each hindfoot force line position sample image, and generating a corresponding mask image, wherein: 1≤i≤I, and I is a natural number; the mask image is composed of I binary mask rectangles that assign different representative values ​​i to the coordinate points in each region of interest contour Ri and have the same size as the original hindfoot force line position sample image; pairing the N mask images with the N hindfoot force line position sample images to form a training data set.

[0055] In some embodiments of the present invention, for a given N sample images, labeling software is used to mark I region of interest contours Ri (1≤i≤I, and I is a natural number) for each sample image, and generate its corresponding mask image. The mask image here is composed of I binary mask rectangles with the same size as the original sample image, which assign different representative values ​​i to the coordinate points in each region of interest contour Ri. In this way, N mask images are paired with N sample images to form a training data set.

[0056] S2. Construct a convolutional neural network model, and train the convolutional neural network model based on the training data set to obtain the final convolutional neural network model;

[0057] Furthermore, it includes: constructing a convolutional neural network model and setting the hyperparameters of the convolutional neural network model, wherein: the input channel of the convolutional neural network model is 1, the output channel is 1×(I+1), and the outputs correspond to the detection probability maps of the contours of I regions of interest and the detection probability maps of the background, respectively.

[0058] Furthermore, it includes: dividing the training data set into a training set, a validation set and a test set; inputting the training set into the convolutional neural network model in batches for training, using the loss function to calculate the difference between the output image and the mask image as the training error to feed back to the convolutional neural network model, and updating the model parameters through the learning algorithm; after each batch of training, using the validation set to perform a performance test on the convolutional neural network model until the performance test index reaches a preset stable value, then the convolutional neural network model training is deemed complete and saved as the final convolutional neural network model.

[0059] In some embodiments of the present invention, a convolutional neural network model is constructed, and the hyperparameters of the convolutional neural network model are set, wherein the input channel of the network is 1, and the output channel is 1×(I+1), which correspond to the detection probability maps of the contours of I regions of interest and the detection probability maps of the background, respectively; the training data set is divided into a training set X1, a validation set X2, and a test set X3, the training set, the validation set, and the test set are independent of each other, and their numbers are n1, n2, and n3, respectively, which are natural numbers, and n1+n2+n3=N, n1≥1 / 2N. The training set X1 is used to train the convolutional neural network model, the validation set X2 is used to evaluate the current performance of the model, and the test set X3 is used to test the generalization performance of the model; during the training process, the training set will be divided into multiple batches and repeatedly input into the convolutional neural network model for multiple rounds of training. At the same time, the difference between the output image and the mask image is calculated using the loss function as the training error and fed back to the convolutional neural network model, and the model parameters are updated through the learning algorithm; the difference between the output detection probability map and the mask image is calculated using the loss function as the training error and performance test index, and optionally, the loss function includes but is not limited to the set similarity measurement function (Dice loss) or the focus loss function (Focol loss). Through the predefined loss function, the difference between the predicted result and the actual result is calculated, and finally the model parameters are adjusted according to the difference to minimize the loss. Through the optimization algorithm, the fitting process of minimizing the error during the training of the convolutional neural network model is realized. Optionally, commonly used optimization algorithms include but are not limited to: stochastic gradient descent (SGD), adaptive moment estimation optimization algorithm (Adam), and momentum algorithm (Momentum). After each batch of training is completed, the validation set is used to perform a performance test on the convolutional neural network model. When the performance test index training tends to be stable, the training of the convolutional neural network model is considered to be completed, and the trained network model is saved.

[0060] S3, obtaining and inputting the patient's hindfoot force line X-ray film into the final convolutional neural network model for segmentation processing, and outputting the segmentation result;

[0061] Furthermore, it includes: inputting the patient's hindfoot force line X-ray film into the final convolutional neural network model, performing region of interest segmentation on the hindfoot force line X-ray film based on the convolutional neural network model, and outputting a segmentation result with a dimension of 1×(I+1), wherein the segmentation result includes a probability map corresponding to I regions of interest in one input channel and a background probability map.

[0062] Furthermore, it includes: binarizing the probability map according to a preset probability threshold to obtain a binary mask image; for the binary mask image obtained for the i-th region of interest excluding the background probability map, retaining the first two connected domains with the largest areas, and ensuring that the two connected domains are respectively located on both sides of the perpendicular bisector of the output probability map, as the segmentation result of the i-th region of interest output.

[0063] In some embodiments of the present invention, the patient's hindfoot force line X-ray film is used as the original image, and necessary preprocessing is performed, including but not limited to pixel standardization, voxel normalization, etc., and the preprocessed image is used as the input of the network model. In this embodiment, the hindfoot force line X-ray film is taken as the original image. The input image is automatically segmented into regions of interest using a pre-trained network, and the output dimension of the network model is 1×(I+1). This output represents a probability map corresponding to I regions of interest in one input channel and a background probability map, respectively, wherein the pixel value in the probability map is the probability of the corresponding key point or background. In this embodiment, I=2. In this embodiment, there are two regions of interest, namely the left and right femurs and the left and right calcaneus in the hindfoot force line X-ray film.

[0064] For each of the probability maps, the following steps can be used to obtain the region of interest:

[0065] ① Binarize the probability map according to the preset probability threshold to obtain a binary mask image;

[0066] ② For the binary mask image obtained for the i-th region of interest except the background probability map, retain the first two connected domains with the largest area, and ensure that the two connected domains are located on both sides of the vertical bisector of the output probability map as the segmentation result of the i-th region of interest.

[0067] S4. Perform anatomical axis fitting based on the segmentation results, and calculate the hindfoot force line angle based on the anatomical axis.

[0068] Further, it includes: based on the segmentation result of the I region of interest, respectively calculating the anatomical axis of the first two connected domains to obtain the first connected domain fitting axis and the second connected domain fitting axis. Based on the first connected domain fitting axis and the second connected domain fitting axis, the final hindfoot force line angle is measured to obtain the angle formed by the first connected domain fitting axis and the second connected domain fitting axis.

[0069] In some embodiments of the present invention, there are multiple predefined algorithms for the calculation method and rules of the anatomical axis of a certain connected domain. In this embodiment, two methods, anatomical definition method and algorithm fitting method, are provided. The calculation results are as follows: Figure 3 and Figure 4 As shown:

[0070] 1. Anatomical definition method:

[0071] 1-1 Determination of the tibial axis: The left and right tibia in the hindfoot force line X-ray film are regarded as two independent connected domains. For each connected domain, two parallel lines A and B are taken with a distance of 30 mm between the mid- and distal ends. The line E connecting the midpoints of the two line segments is the central axis of the connected domain.

[0072] 1-2 Determination of the calcaneal axis: The left and right calcaneus in the hindfoot force line X-ray film are regarded as two independent connected domains. For each connected domain, take the line segment D parallel to the horizontal plane at 7mm from the lowest point, take the point d at 40% of the distance from the outside, and then make a line segment C parallel to the horizontal plane 20mm away from the lowest point of the connected domain, take the point c in it, and the line F connecting point c and d is the central axis of the connected domain.

[0073] 2. Algorithm fitting method:

[0074] 2-1 Generate distance field function: The left and right tibia or left and right calcaneus in the hindfoot force line X-ray are regarded as independent connected domains, and the binary region contour is obtained by edge detection algorithm. The distance from each pixel point in the connected domain to the boundary contour is calculated to generate the distance field function;

[0075] 2-2 Extracting the medial skeleton of the connected domain: Based on the generated distance field function, the medial skeleton line of the independent connected domain is extracted through a series of mathematical operations or morphological operations. The extraction methods that can be used include but are not limited to: local minimum detection algorithm, morphological processing of expansion and erosion, etc.;

[0076] 2-3 Central axis optimization and smoothing: The extracted central axis skeleton is optimized and smoothed to eliminate noise, enhance continuity, and make the connected domain axis fitted more accurate.

[0077] In some embodiments of the present invention, the hindfoot force line angle to be measured is defined as the angle between the ipsilateral tibial anatomical axis and the calcaneal anatomical axis. Therefore, the image matrix is ​​divided into left and right sides with the perpendicular bisector of the hindfoot force line X-ray as the boundary. On the left and right sides, based on the tibia and calcaneal axes fitted in the previous steps, the angle between the two axes is calculated. When calculating the angle, the positive direction of the y-axis of the hindfoot force line X-ray image matrix is ​​taken as the positive direction of the tibia and calcaneal axes. If the angle α between the two is <90°, the positive direction of the scalar α is given, so the final force line angle is vector +α; if the angle α between the two is >90°, the negative direction of the scalar α is given, so the final force line angle is vector -(180°α).

[0078] This method acquires and processes a large number of sample images to construct a reasonable data set and an effective convolutional neural network model, so as to perform accurate and efficient image segmentation in the future. The tibia and calcaneus are segmented in the hindfoot force line X-ray film, and the anatomical axes are fitted respectively. Finally, the hindfoot force line angle is calculated according to the fitting results, so as to realize rapid and accurate diagnosis of hindfoot deformity based on X-ray images and artificial intelligence-assisted diagnosis.

[0079] like Figure 2 As shown, in a second aspect, an embodiment of the present invention provides a system for measuring hindfoot force line angle by X-ray based on deep learning, comprising a training set construction module 100, a model training module 200, an image segmentation module 300 and a hindfoot force line angle calculation module 400, wherein:

[0080] The training set construction module 100 is used to obtain and process the sample images of the rear foot force line position to construct a training data set;

[0081] The model training module 200 is used to construct a convolutional neural network model and train the convolutional neural network model based on the training data set to obtain a final convolutional neural network model;

[0082] The image segmentation module 300 is used to obtain and input the patient's hindfoot force line X-ray film into the final convolutional neural network model for segmentation processing and output the segmentation result;

[0083] The hindfoot force line angle calculation module 400 is used to perform anatomical axis fitting according to the segmentation results and calculate the hindfoot force line angle according to the anatomical axis.

[0084] This system is based on the cooperation of multiple modules such as the training set construction module 100, the model training module 200, the image segmentation module 300 and the hindfoot force line angle calculation module 400. By acquiring and processing a large number of sample images, a reasonable data set is constructed, and an effective convolutional neural network model is constructed to facilitate subsequent accurate and efficient image segmentation. The hindfoot force line X-ray film is segmented into the tibia and calcaneus, and the anatomical axes are fitted respectively. Finally, the hindfoot force line angle is calculated according to the fitting results, so as to achieve rapid and accurate diagnosis of hindfoot deformity based on X-ray images and artificial intelligence-assisted diagnosis.

[0085] like Figure 5 As shown, in a third aspect, an embodiment of the present application provides an electronic device, which includes a memory 101 for storing one or more programs and a processor 102. When the one or more programs are executed by the processor 102, any method in the first aspect described above is implemented.

[0086] The memory 101, the processor 102 and the communication interface 103 are directly or indirectly electrically connected to each other to achieve data transmission or interaction. For example, these elements can be electrically connected to each other through one or more communication buses or signal lines. The memory 101 can be used to store software programs and modules, and the processor 102 executes various functional applications and data processing by executing the software programs and modules stored in the memory 101. The communication interface 103 can be used to communicate signaling or data with other node devices.

[0087] Among them, the memory 101 can be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable read-only memory (EEPROM), etc.

[0088] The processor 102 may be an integrated circuit chip with signal processing capability. The processor 102 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0089] In the embodiments provided in the present application, it should be understood that the disclosed method and system can also be implemented in other ways. The method and system embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the method and system, method and computer program product according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or the flowchart, and the combination of boxes in the block diagram and / or the flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0090] In addition, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0091] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by the processor 102, a method as in any one of the first aspects described above is implemented. If the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or part of the contribution to the prior art or 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 for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a disk or an optical disk.

[0092] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0093] It will be apparent to those skilled in the art that the present application is not limited to the details of the exemplary embodiments described above, and that the present application can be implemented in other specific forms without departing from the spirit or essential features of the present application. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the present application is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims be included in the present application. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A method for measuring the hindfoot force line angle by X-ray based on deep learning, characterized in that: The following steps are involved: Acquire and process sample images of hindfoot force line positions to construct a training data set; Construct a convolutional neural network model and train the convolutional neural network model based on the training data set to obtain the final convolutional neural network model; Obtain and input the patient's hindfoot force line X-ray film into the final convolutional neural network model for segmentation processing and output the segmentation result; The anatomical axis was fitted according to the segmentation results, and the hindfoot force line angle was calculated based on the anatomical axis.

2. The method for measuring the hindfoot force line angle by X-ray based on deep learning according to claim 1, characterized in that: The method of acquiring and processing the rear foot force line position sample images to construct a training data set comprises the following steps: Obtain N rear foot force line position sample images, where: N is a natural number greater than or equal to 1; Mark I region of interest contours Ri for each rear foot force line position sample image, and generate a corresponding mask image, where: 1≤i≤I, and I is a natural number; the mask image is composed of I binary mask rectangles that assign different representative values ​​i to the coordinate points in each region of interest contour Ri and have the same size as the original rear foot force line position sample image; The N mask images are paired with N hindfoot force line position sample images to form a training data set.

3. The method for measuring the hindfoot force line angle by X-ray based on deep learning according to claim 1, characterized in that: The method for constructing a convolutional neural network model comprises the following steps: Construct a convolutional neural network model and set the hyperparameters of the convolutional neural network model, where the input channel of the convolutional neural network model is 1, the output channel is 1×(I+1), and the output corresponds to the detection probability map of the contours of I regions of interest and the detection probability map of the background.

4. The method for measuring the hindfoot force line angle by X-ray based on deep learning according to claim 1, characterized in that: The method of training the convolutional neural network model based on the training data set to obtain the final convolutional neural network model comprises the following steps: Divide the training data set into training set, validation set and test set; The training set is input into the convolutional neural network model in batches for training. The difference between the output image and the mask image is calculated using the loss function as the training error and fed back to the convolutional neural network model. The model parameters are updated through the learning algorithm. After each batch of training is completed, the validation set is used to perform performance testing on the convolutional neural network model until the performance test index reaches the preset stable value. The convolutional neural network model training is then considered complete and saved as the final convolutional neural network model.

5. The method for measuring the hindfoot force line angle by X-ray based on deep learning according to claim 1, characterized in that: The method of inputting the patient's hindfoot force line X-ray film into the final convolutional neural network model for segmentation processing comprises the following steps: The patient's hindfoot force line X-ray film is input into the final convolutional neural network model, and the region of interest of the hindfoot force line X-ray film is segmented based on the convolutional neural network model, and the segmentation result with a dimension of 1×(I+1) is output. The segmentation result includes a probability map corresponding to I regions of interest in one input channel and a background probability map.

6. The method for measuring the hindfoot force line angle by X-ray based on deep learning according to claim 5, characterized in that: The following steps are also included: Binarizing the probability map according to a preset probability threshold to obtain a binary mask image; For the binary mask image obtained for the i-th region of interest except the background probability map, retain the first two connected domains with the largest area, and ensure that the two connected domains are located on both sides of the vertical bisector of the output probability map as the segmentation result of the i-th region of interest.

7. The method for measuring the hindfoot force line angle by X-ray based on deep learning according to claim 6, characterized in that: The method for fitting the anatomical axis according to the segmentation result comprises the following steps: Based on the segmentation results of I regions of interest, the anatomical axes are calculated for the first two connected domains respectively to obtain the fitting axis of the first connected domain and the fitting axis of the second connected domain.

8. A system for measuring hindfoot force line angle by X-ray based on deep learning, characterized in that: It includes a training set construction module, a model training module, an image segmentation module, and a rear foot force line angle calculation module, wherein: A training set construction module is used to obtain and process sample images of hindfoot force line positions to construct a training data set; The model training module is used to build a convolutional neural network model and train the convolutional neural network model based on the training data set to obtain the final convolutional neural network model; An image segmentation module is used to obtain and input the patient's hindfoot force line X-ray film into the final convolutional neural network model for segmentation processing and output the segmentation result; The hindfoot force line angle calculation module is used to fit the anatomical axis according to the segmentation results and calculate the hindfoot force line angle according to the anatomical axis.

9. An electronic device, characterized in that: include: A memory for storing one or more programs; processor; When the one or more programs are executed by the processor, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images