A Deep Learning-Based Diagnostic Method and System for Cervical Spinal Stenosis

Abstract

This application provides a method, system, device, and computer-readable storage medium for diagnosing cervical spinal stenosis based on deep learning. The method includes: acquiring cervical spine X-ray images of a patient; inputting the cervical spine X-ray images into a preset cervical spinal stenosis diagnostic model; and outputting a cervical spinal stenosis diagnostic result. The labels of the cervical spine X-ray images in the training dataset of the cervical spinal stenosis diagnostic model are determined based on the cervical spinal stenosis diagnostic result of the corresponding cervical spine MR images. According to the embodiments of this application, cervical spinal stenosis can be diagnosed more accurately based on X-ray images.

Description

Technical Field

[0001] This application belongs to the field of cervical spinal stenosis, and in particular relates to a deep learning-based diagnostic method, system, device, and computer-readable storage medium for cervical spinal stenosis. Background Technology

[0002] Cervical spinal stenosis is a common cervical spine disease. It refers to a narrowing of the diameter of the cervical spinal canal, which compresses the spinal cord and nerve roots within the canal, potentially causing a range of clinical symptoms. This stenosis can be congenital or acquired, such as due to cervical degenerative changes, bone hyperplasia, ossification of the ligamentum flavum, or herniated discs.

[0003] Symptoms of cervical spinal stenosis may include dizziness, headache, neck pain, arm pain, numbness and weakness in the upper limbs, and in severe cases may lead to weakness in all four limbs and bowel and bladder dysfunction.

[0004] Diagnosing cervical spinal stenosis usually requires combining the patient's symptoms, signs, and imaging examinations. MR images have unique advantages in diagnosing cervical spinal stenosis and spinal cord compression, and can clearly identify whether there are changes in spinal cord signal and nerve compression.

[0005] X-ray images have advantages in terms of cost-effectiveness, availability, and ease of use, but doctors cannot diagnose cervical spinal stenosis with the naked eye based on X-ray images.

[0006] Therefore, how to more accurately diagnose cervical spinal stenosis based on X-ray images is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0007] This application provides a method, system, device, and computer-readable storage medium for diagnosing cervical spinal stenosis based on deep learning, which can more accurately diagnose cervical spinal stenosis based on X-ray images.

[0008] In a first aspect, embodiments of this application provide a deep learning-based diagnostic method for cervical spinal stenosis, including:

[0009] Obtain X-ray images of the patient's cervical spine;

[0010] Input cervical spine X-ray images into a preset cervical spinal stenosis diagnostic model and output cervical spinal stenosis diagnostic results;

[0011] The labels of cervical spine X-ray images in the training dataset of the cervical spinal stenosis diagnostic model are determined based on the diagnostic results of cervical spinal stenosis from the corresponding cervical spine MR images.

[0012] Optionally, the network structure of the cervical spinal stenosis diagnostic model includes:

[0013] Feature extraction, parallax attention extraction, and feature reconstruction;

[0014] In the feature extraction module stage, a parameter-shared modified binarized feature fusion group is used to fuse different levels of features extracted by the multi-attention mechanism within a single view.

[0015] In the parallax attention extraction stage, a dual-channel parallax attention mechanism and a pyramid sampling mechanism are introduced to fuse local and global information between the two views;

[0016] The feature reconstruction stage continues the feature fusion group from the feature extraction module, and reconstructs the high-resolution left view image through the automatic calculation parameter unit.

[0017] Optionally, the feature fusion group is built on the multi-attention fusion module as the basic module and the modified binarization fusion framework as the feature fusion framework.

[0018] Optional, including:

[0019] The main branch of the multi-attention fusion module first extracts spatial features by stacking 3×3 convolutional layers to extract input features, and then using 1×1 convolution to reduce the number of channels to 1 / 4 of the original.

[0020] To increase the receptive field, a 7×7 convolution kernel with a stride of 2 is used to reduce the spatial dimension. Then, after the pooling layer, dilated convolution operations with dilation rates of 1 and 2 are performed respectively.

[0021] After upsampling to the original input feature dimension, the spatial feature tensors of the left and right view images are output through a 1×1 convolution kernel Sigmoid normalization operation.

[0022] Optionally, after acquiring the patient's cervical spine X-ray images, the following may be included:

[0023] Image preprocessing of cervical spine X-ray images;

[0024] Image preprocessing includes: filtering and denoising, contrast enhancement, and image normalization.

[0025] Optionally, the cervical spine X-ray images may be filtered for noise reduction, contrast enhancement, and image normalization, including:

[0026] Use a median filter or a Gaussian filter to filter and denoise cervical spine X-ray images;

[0027] For the filtered and denoised cervical spine X-ray images, histogram equalization is used to adjust the contrast and brightness of the images, thereby enhancing the visibility of lesion areas in the images.

[0028] The contrast-enhanced cervical spine X-ray image is normalized to ensure that the pixel values ​​are distributed within a preset range.

[0029] Optionally, during the training process of the cervical spinal stenosis diagnostic model, the batch size is set to 32.

[0030] The initial learning rate is set to 1e-4, with an additional learning rate decay strategy. Every 5000 iterations, the learning rate decays to 0.9 of the previous learning rate.

[0031] Set the optimizer to the Adam optimizer;

[0032] Set the loss function to DICEloss;

[0033] The training and validation sets are validated every 1000 iterations. The early stopping method is used to determine the network training stopping time and obtain the cervical spinal stenosis diagnostic model.

[0034] Secondly, embodiments of this application provide a deep learning-based diagnostic system for cervical spinal stenosis, comprising:

[0035] The cervical spine X-ray image acquisition module is used to acquire X-ray images of the patient's cervical spine.

[0036] The cervical spinal stenosis diagnosis module is used to input cervical spine X-ray images into a preset cervical spinal stenosis diagnosis model and output the cervical spinal stenosis diagnosis results.

[0037] The labels of cervical spine X-ray images in the training dataset of the cervical spinal stenosis diagnostic model are determined based on the diagnostic results of cervical spinal stenosis from the corresponding cervical spine MR images.

[0038] Thirdly, embodiments of this application provide an electronic device, which includes: a processor and a memory storing computer program instructions;

[0039] When the processor executes the computer program instructions, it implements a deep learning-based diagnostic method for cervical spinal stenosis.

[0040] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement a deep learning-based diagnostic method for cervical spinal stenosis.

[0041] The deep learning-based diagnostic method, system, device, and computer-readable storage medium of this application can more accurately diagnose cervical spinal stenosis based on X-ray images.

[0042] This deep learning-based diagnostic method for cervical spinal stenosis includes:

[0043] Obtain X-ray images of the patient's cervical spine;

[0044] Input cervical spine X-ray images into a preset cervical spinal stenosis diagnostic model and output cervical spinal stenosis diagnostic results;

[0045] The labels of cervical spine X-ray images in the training dataset of the cervical spinal stenosis diagnostic model are determined based on the diagnostic results of cervical spinal stenosis from the corresponding cervical spine MR images. Attached Figure Description

[0046] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0047] Figure 1 This is a flowchart illustrating a deep learning-based diagnostic method for cervical spinal stenosis provided in one embodiment of this application.

[0048] Figure 2 This is a schematic diagram of the network structure of a cervical spinal stenosis diagnostic model provided in one embodiment of this application;

[0049] Figure 3 This is a schematic diagram of the network structure of a feature fusion group provided in one embodiment of this application;

[0050] Figure 4 This is a schematic diagram of the network structure of a multi-attention fusion module provided in one embodiment of this application;

[0051] Figure 5 This is a schematic diagram of the structure of a deep learning-based cervical spinal stenosis diagnostic system provided in one embodiment of this application;

[0052] Figure 6 This is a schematic diagram of the structure of an electronic device provided in one embodiment of this application. Detailed Implementation

[0053] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0054] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0055] To address the problems of existing technologies, embodiments of this application provide a method, system, device, and computer-readable storage medium for diagnosing cervical spinal stenosis based on deep learning. The method for diagnosing cervical spinal stenosis based on deep learning provided in this application embodiment will be described below first.

[0056] Figure 1 This diagram illustrates a flowchart of a deep learning-based diagnostic method for cervical spinal stenosis provided in one embodiment of this application. Figure 1 As shown, this deep learning-based diagnostic method for cervical spinal stenosis includes:

[0057] S101. Obtain X-ray images of the patient's cervical spine;

[0058] S102. Input the cervical spine X-ray image into the preset cervical spinal stenosis diagnosis model and output the cervical spinal stenosis diagnosis result; wherein, the labels of the cervical spine X-ray images in the training dataset of the cervical spinal stenosis diagnosis model are determined based on the cervical spinal stenosis diagnosis result of the corresponding cervical spine MR image.

[0059] Figure 2 This is a schematic diagram of the network structure of a cervical spinal stenosis diagnostic model provided in one embodiment of this application;

[0060] In one embodiment, the network structure of the cervical spinal stenosis diagnostic model includes:

[0061] Feature extraction, parallax attention extraction, and feature reconstruction;

[0062] In the feature extraction module stage, a parameter-shared modified binarized feature fusion group is used to fuse different levels of features extracted by the multi-attention mechanism within a single view.

[0063] In the parallax attention extraction stage, a dual-channel parallax attention mechanism and a pyramid sampling mechanism are introduced to fuse local and global information between the two views;

[0064] The feature reconstruction stage continues the feature fusion group from the feature extraction module, and reconstructs the high-resolution left view image through the automatic calculation parameter unit.

[0065] Figure 3 This is a schematic diagram of the network structure of a feature fusion group provided in one embodiment of this application;

[0066] In one embodiment, the feature fusion group is built on a multi-attention fusion module as the base module and a modified binarization fusion framework as the feature fusion framework.

[0067] Figure 4 This is a schematic diagram of the network structure of a multi-attention fusion module provided in one embodiment of this application;

[0068] In one embodiment, it includes:

[0069] The main branch of the multi-attention fusion module first extracts spatial features by stacking 3×3 convolutional layers to extract input features, and then using 1×1 convolution to reduce the number of channels to 1 / 4 of the original.

[0070] To increase the receptive field, a 7×7 convolution kernel with a stride of 2 is used to reduce the spatial dimension. Then, after the pooling layer, dilated convolution operations with dilation rates of 1 and 2 are performed respectively.

[0071] After upsampling to the original input feature dimension, the spatial feature tensors of the left and right view images are output through a 1×1 convolution kernel Sigmoid normalization operation.

[0072] In one embodiment, after acquiring X-ray images of the patient's cervical spine, the following steps are included:

[0073] Image preprocessing of cervical spine X-ray images;

[0074] Image preprocessing includes: filtering and denoising, contrast enhancement, and image normalization.

[0075] In one embodiment, filtering, denoising, contrast enhancement, and image normalization of cervical spine X-ray images include:

[0076] Use a median filter or a Gaussian filter to filter and denoise cervical spine X-ray images;

[0077] For the filtered and denoised cervical spine X-ray images, histogram equalization is used to adjust the contrast and brightness of the images, thereby enhancing the visibility of lesion areas in the images.

[0078] The contrast-enhanced cervical spine X-ray image is normalized to ensure that the pixel values ​​are distributed within a preset range.

[0079] In one embodiment, during the training process of the cervical spinal stenosis diagnostic model, the batch size is set to 32.

[0080] The initial learning rate is set to 1e-4, with an additional learning rate decay strategy. Every 5000 iterations, the learning rate decays to 0.9 of the previous learning rate.

[0081] Set the optimizer to the Adam optimizer;

[0082] Set the loss function to DICEloss;

[0083] The training and validation sets are validated every 1000 iterations. The early stopping method is used to determine the network training stopping time and obtain the cervical spinal stenosis diagnostic model.

[0084] Figure 5 This is a schematic diagram of the structure of a deep learning-based cervical spinal stenosis diagnostic system provided in one embodiment of this application;

[0085] This deep learning-based diagnostic system for cervical spinal stenosis includes:

[0086] Cervical spine X-ray image acquisition module 501, used to acquire cervical spine X-ray images of patients;

[0087] The cervical spinal stenosis diagnosis module 502 is used to input cervical spine X-ray images into a preset cervical spinal stenosis diagnosis model and output cervical spinal stenosis diagnosis results.

[0088] The labels of cervical spine X-ray images in the training dataset of the cervical spinal stenosis diagnostic model are determined based on the diagnostic results of cervical spinal stenosis from the corresponding cervical spine MR images.

[0089] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of this application is shown.

[0090] The electronic device may include a processor 601 and a memory 602 storing computer program instructions.

[0091] Specifically, the processor 601 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0092] Memory 602 may include mass storage for data or instructions. For example, and not limitingly, memory 602 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where suitable, memory 602 may include removable or non-removable (or fixed) media. Where suitable, memory 602 may be internal or external to an electronic device. In a particular embodiment, memory 602 may be a non-volatile solid-state memory.

[0093] In one embodiment, memory 602 may be read-only memory (ROM). In one embodiment, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0094] The processor 601 reads and executes computer program instructions stored in the memory 602 to implement any of the deep learning-based cervical spinal stenosis diagnosis methods in the above embodiments.

[0095] In one example, the electronic device may also include a communication interface 603 and a bus 610. For example, Figure 6 As shown, the processor 601, memory 602, and communication interface 603 are connected through bus 610 and complete communication with each other.

[0096] The communication interface 603 is mainly used to realize communication between various modules, systems, units and / or devices in the embodiments of this application.

[0097] Bus 610 includes hardware, software, or both, that couples components of an electronic device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 610 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0098] Furthermore, in conjunction with the deep learning-based cervical spinal stenosis diagnosis method in the above embodiments, this application embodiment can provide a computer-readable storage medium for implementation. This computer-readable storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the deep learning-based cervical spinal stenosis diagnosis methods in the above embodiments.

[0099] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0100] The functional modules shown in the above-described block diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0101] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or systems. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0102] The foregoing flowcharts and / or block diagrams of methods, systems, and computer program products according to embodiments of this application have described various aspects of the present application. It should be understood that each block in the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing system to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing system, enable the implementation of the functions / actions specified in one or more blocks of the flowcharts and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0103] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A deep learning-based diagnostic method for cervical spinal stenosis, characterized in that, include: Obtain X-ray images of the patient's cervical spine; Input cervical spine X-ray images into a preset cervical spinal stenosis diagnostic model and output cervical spinal stenosis diagnostic results; The labels of cervical spine X-ray images in the training dataset of the cervical spinal stenosis diagnostic model are determined based on the diagnostic results of cervical spinal stenosis in the corresponding cervical spine MR images. The network structure of the cervical spinal stenosis diagnostic model includes: feature extraction, disparity attention extraction, and feature reconstruction. In the feature extraction stage, a parameter-shared modified binarized feature fusion group is used to fuse different levels of features extracted by the multi-attention mechanism within a single view. In the disparity attention extraction stage, a dual-channel disparity attention mechanism and a pyramid sampling mechanism are introduced to fuse local and global information between the two views. In the feature reconstruction stage, the feature fusion group of the feature extraction module is continued, and a high-resolution left view image is reconstructed through an automatic parameter calculation unit. The feature fusion group is built upon the multi-attention fusion module as its basic module and a modified binarization fusion framework as its feature fusion framework. The main branch of the multi-attention fusion module first extracts spatial features by stacking 3×3 convolutional layers to extract input features, and then reducing the number of channels to 1 / 4 of the original by 1×1 convolution. To increase the receptive field, a 7×7 convolutional kernel with a stride of 2 is used to reduce the spatial dimension. After passing through the pooling layer, dilated convolution operations with dilation rates of 1 and 2 are performed respectively. After upsampling to the original input feature dimension, the spatial feature tensors of the left and right view images are output through a 1×1 convolutional kernel sigmoid normalization operation.

2. The deep learning-based diagnostic method for cervical spinal stenosis according to claim 1, characterized in that, After obtaining the patient's cervical spine X-ray images, the following steps are taken: Image preprocessing of cervical spine X-ray images; Image preprocessing includes: filtering and denoising, contrast enhancement, and image normalization.

3. The deep learning-based diagnostic method for cervical spinal stenosis according to claim 2, characterized in that, The cervical spine X-ray images were filtered for noise reduction, contrast enhancement, and image normalization, including: Use a median filter or a Gaussian filter to filter and denoise cervical spine X-ray images; For the filtered and denoised cervical spine X-ray images, histogram equalization is used to adjust the contrast and brightness of the images, thereby enhancing the visibility of lesion areas in the images. The contrast-enhanced cervical spine X-ray image is normalized to ensure that the pixel values ​​are distributed within a preset range.

4. The deep learning-based diagnostic method for cervical spinal stenosis according to claim 1, characterized in that, During the training process of the cervical spinal stenosis diagnostic model, the batch size is set to 32. The initial learning rate is set to 1e-4, with an additional learning rate decay strategy. Every 5000 iterations, the learning rate decays to 0.9 of the previous learning rate. Set the optimizer to the Adam optimizer; Set the loss function to DICEloss; The training and validation sets are validated every 1000 iterations. The early stopping method is used to determine the network training stopping time and obtain the cervical spinal stenosis diagnostic model.

5. A deep learning-based diagnostic system for cervical spinal stenosis, characterized in that, The system includes: The cervical spine X-ray image acquisition module is used to acquire X-ray images of the patient's cervical spine. The cervical spinal stenosis diagnosis module is used to input cervical spine X-ray images into a preset cervical spinal stenosis diagnosis model and output the cervical spinal stenosis diagnosis results. The labels of cervical spine X-ray images in the training dataset of the cervical spinal stenosis diagnostic model are determined based on the diagnostic results of cervical spinal stenosis in the corresponding cervical spine MR images. The network structure of the cervical spinal stenosis diagnostic model includes: feature extraction, disparity attention extraction, and feature reconstruction. In the feature extraction stage, a parameter-shared modified binarized feature fusion group is used to fuse different levels of features extracted by the multi-attention mechanism within a single view. In the disparity attention extraction stage, a dual-channel disparity attention mechanism and a pyramid sampling mechanism are introduced to fuse local and global information between the two views. In the feature reconstruction stage, the feature fusion group of the feature extraction module is continued, and a high-resolution left view image is reconstructed through an automatic parameter calculation unit. The feature fusion group is built upon the multi-attention fusion module as its basic module and a modified binarization fusion framework as its feature fusion framework. The main branch of the multi-attention fusion module first extracts spatial features by stacking 3×3 convolutional layers to extract input features, and then reducing the number of channels to 1 / 4 of the original by 1×1 convolution. To increase the receptive field, a 7×7 convolutional kernel with a stride of 2 is used to reduce the spatial dimension. After passing through the pooling layer, dilated convolution operations with dilation rates of 1 and 2 are performed respectively. After upsampling to the original input feature dimension, the spatial feature tensors of the left and right view images are output through a 1×1 convolutional kernel sigmoid normalization operation.

6. An electronic device, characterized in that, The electronic device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the deep learning-based diagnostic method for cervical spinal stenosis as described in any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the deep learning-based diagnostic method for cervical spinal stenosis as described in any one of claims 1-4.

Images