A three-dimensional CT image reconstruction method and system for lumbar disease diagnosis

Abstract

The application provides a three-dimensional CT image reconstruction method and system for lumbar disease diagnosis, and belongs to the technical field of three-dimensional medical image reconstruction; a plurality of two-dimensional CT images are acquired to train a deep neural network; a two-dimensional CT image to be segmented is input into the trained deep neural network for segmentation, and the segmented image is used as a pre-reconstruction image; three-dimensional CT image reconstruction is performed on the pre-reconstruction image based on a VTK rendering method, so that the visualization of the three-dimensional CT image is realized. The application can accurately and efficiently simultaneously segment the vertebrae, intervertebral discs and nerve structures in the lumbar CT image, and realizes image visualization through three-dimensional reconstruction of the segmented CT image, thereby providing strong support for lumbar disease diagnosis, preoperative planning and intraoperative guidance.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of three-dimensional medical image reconstruction, and particularly relates to a three-dimensional CT image reconstruction method and system for lumbar disease diagnosis. BACKGROUND

[0002] The statements in this section merely provide background information related to the application and do not necessarily constitute prior art.

[0003] With the increasingly serious problem of population aging and the change of modern life and work mode, the incidence of spinal diseases has increased significantly, accounting for more than 40%; among them, lumbar disease is particularly prominent, which not only erodes the patient's nervous system function, but also greatly limits the daily activity ability, and thus reduces the overall quality of life. Common lumbar diseases include lumbar spinal stenosis (LSS), lumbar intervertebral disc herniation (IDH) and lumbar deformity, etc. LSS is the primary reason for patients over 65 years old to receive spinal surgery, mainly due to hyperosteosis or hyperplasia of fibrous tissue leading to spinal canal or nerve root canal stenosis, which stimulates or compresses the spinal nerve root and cauda equina, causing a series of clinical symptoms; IDH is due to the protrusion of intervertebral disc tissue, which compresses the adjacent nerve root, often manifested as low back pain and lower extremity radiating pain; and lumbar deformity is often accompanied by scoliosis or spondylolisthesis, which not only affects the appearance, but also may cause movement disorders, seriously damaging the patient's quality of life. Due to the complexity and comprehensiveness of lumbar disease, it often involves multiple lesions of the vertebral body, intervertebral disc and nervous system, so the diagnosis process becomes particularly complex.

[0004] The inventors found that the existing methods for diagnosing lumbar diseases have some technical problems, for example:

[0005] (1) In the actual clinical field, when diagnosing lumbar diseases, the current diagnosis is mainly based on the CT images taken, and the diagnosis method still completely relies on the personal experience of the doctor; however, the current medical equipment mainly provides two-dimensional images of the internal anatomical structure of the patient, which requires the doctor to construct the three-dimensional morphology, size and spatial relationship with the surrounding tissue of the lesion in the brain based on these two-dimensional image sequences, and then make a diagnosis. Therefore, the doctor can only diagnose the disease according to this single CT image, and it is difficult to detect the relatively subtle lesions in the CT image, and it is time-consuming and laborious.

[0006] (2) Some existing technologies capable of three-dimensional reconstruction of two-dimensional medical images can only directly map the original two-dimensional medical images to three-dimensional images, and cannot simultaneously segment multiple structures from the original two-dimensional medical images; therefore, even if mapped to three-dimensional medical images, it is still difficult to accurately identify each part represented in the image. SUMMARY

[0007] In order to overcome the above-mentioned deficiencies of the prior art, the present application provides a three-dimensional CT image reconstruction method and system for lumbar disease diagnosis, which can accurately and efficiently segment the vertebrae, intervertebral discs and nerve structures in the lumbar CT image at the same time, and realize image visualization through three-dimensional reconstruction of the segmented CT image, thereby providing strong support for lumbar disease diagnosis, preoperative planning and intraoperative guidance.

[0008] In order to achieve the above-mentioned purpose, one or more embodiments of the present application provide the following technical solutions:

[0009] The present application provides a three-dimensional CT image reconstruction method for lumbar disease diagnosis in the first aspect.

[0010] The three-dimensional CT image reconstruction method for lumbar disease diagnosis comprises the following steps:

[0011] Obtaining a plurality of two-dimensional CT images to construct a training data set;

[0012] Training a deep neural network based on the training data set;

[0013] Inputting the two-dimensional CT image to be segmented into the trained deep neural network for two-dimensional CT image segmentation, and taking the segmented two-dimensional CT image as a pre-reconstruction image;

[0014] Performing three-dimensional CT image reconstruction on the pre-reconstruction image based on a VTK rendering method, specifically: building a user interface and embedding a VTK rendering window in the user interface; reading the pre-reconstruction image based on a reader provided by VTK, and converting the pre-reconstruction image into three-dimensional volume data based on an entity conversion method; performing three-dimensional reconstruction on the volume data using a ray transmission method, converting the volume data after three-dimensional reconstruction into a data format readable by an image pipeline based on a mapper, and performing data rendering based on the VTK rendering window; using an interactor to start a rendering cycle to display the three-dimensional CT image after rendering, thereby realizing visualization of the three-dimensional CT image.

[0015] Further, the deep neural network is a double network, that is, the deep neural network is composed of a Res-Net network and a V-Net network.

[0016] Further, the segmentation of the two-dimensional CT image using the deep neural network comprises the following steps: first, segmenting the two-dimensional CT image based on the Res-Net network to obtain a vertebra structure segmentation image; then, segmenting the two-dimensional CT image based on the V-Net network to obtain an intervertebral disc and nerve structure segmentation image; and then, integrating the vertebra structure segmentation image and the intervertebral disc and nerve structure segmentation image to obtain a pre-reconstruction image.

[0017] Further, the pre-reconstruction image is converted into three-dimensional volume data based on an entity conversion method, specifically: first, a Marching Cubes algorithm is called by using a vtkMarchingCubes class of VTK, each voxel in the volume data is traversed, and the position of an isosurface is determined according to the sign change of the vertex value in the voxel; subsequently, the generated isosurface data is mapped into a geometric representation by using a mapper of VTK, and the obtained geometric representation is added to a renderer by using an actor vtkActor; wherein the voxel is a three-dimensional pixel.

[0018] Further, a user interface is built, including: an initial user interface is built based on an interface tool Qt Designer, and controls are added to the initial user interface; after the user interface is built, the format of the user interface is converted, so as to be integrated with a VTK rendering window subsequently.

[0019] Further, the controls include buttons and labels; wherein the buttons include a single-click CT data import button and a single-click start segmentation and reconstruction button.

[0020] Further, after the volume data is three-dimensionally reconstructed, before VTK rendering, the volume data is sequentially subjected to a normal calculation, a smoothing processing and a polygon reduction operation, so as to ensure the quality of the volume data before rendering.

[0021] The second aspect of the application provides a three-dimensional CT image reconstruction system for lumbar disease diagnosis.

[0022] A three-dimensional CT image reconstruction system for lumbar disease diagnosis, comprising:

[0023] A data acquisition module configured to: acquire a plurality of two-dimensional CT images, and construct a training data set;

[0024] A model training module configured to: train a deep neural network based on the training data set;

[0025] A two-dimensional CT image segmentation module configured to: input a two-dimensional CT image to be segmented into the trained deep neural network for two-dimensional CT image segmentation, and take the segmented two-dimensional CT image as a pre-reconstruction image;

[0026] The three-dimensional CT image reconstruction module is configured to perform three-dimensional CT image reconstruction on the pre-reconstruction image based on a VTK rendering method, specifically: a user interface is built, and a VTK rendering window is embedded in the user interface; a pre-reconstruction image is read based on a reader provided by VTK, and the pre-reconstruction image is converted into three-dimensional volume data based on an entity conversion method; three-dimensional reconstruction is performed on the volume data by using a ray casting method, the volume data after three-dimensional reconstruction is converted into a data format readable by an image pipeline based on a mapper, and data rendering is performed based on the VTK rendering window; a renderer is started to display the three-dimensional CT image after rendering, and the visualization of the three-dimensional CT image is realized.

[0027] The third aspect of the present application provides a computer readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps of a three-dimensional CT image reconstruction method for lumbar disease diagnosis according to the first aspect of the present application.

[0028] The fourth aspect of the present application provides an electronic device comprising a memory, a processor, and a program stored on the memory and executable on the processor, wherein the processor implements the steps of a three-dimensional CT image reconstruction method for lumbar disease diagnosis according to the first aspect of the present application when executing the program.

[0029] The above one or more technical solutions have the following beneficial effects:

[0030] (1) The two-dimensional CT image to be segmented is input into a deep neural network for two-dimensional CT image segmentation, and the segmented two-dimensional CT image is used as a pre-reconstruction image; the pre-reconstruction image is reconstructed into a three-dimensional CT image based on a VTK rendering method, so as to realize the visualization of the three-dimensional CT image. Therefore, the present application can perform three-dimensional reconstruction on the basis of two-dimensional CT images, not only reducing the work burden of spine surgeons, but also allowing doctors to freely switch perspectives and levels through three-dimensional visualization technology, deeply understanding the internal structure of patients, and accurately grasping the disease and anatomical details without relying on imagination, providing strong support for lumbar pain diagnosis, preoperative planning and intraoperative guidance.

[0031] (2) The present application performs two-dimensional CT image segmentation based on a deep neural network, that is, the vertebral structure and intervertebral disc and nerve structure of the two-dimensional CT image are segmented based on a Res-Net network and a V-Net network, respectively. Therefore, the present application can segment three spinal structures (intervertebral disc + vertebra + nerve structure) in one step, and on this basis, three-dimensional reconstruction is performed, so that each part in the three-dimensional CT image can be clearly distinguished, and the positioning accuracy of the pain source can be significantly improved.

[0032] Advantages of the present application's additional aspects will become apparent in the description that follows, and become apparent to those of ordinary skill in the art by following that description. BRIEF DESCRIPTION OF DRAWINGS

[0033] The accompanying drawings, which form a part of this specification, are included to provide a further understanding of the application, and are incorporated by reference in their entirety.

[0034] Figure 1 A flow chart of a three-dimensional CT image reconstruction method for lumbar disease diagnosis in the embodiment of the present application.

[0035] Figure 2 An example diagram of a two-dimensional CT image in the embodiment of the present application.

[0036] Figure 3 A structure diagram of a deep neural network in the embodiment of the present application.

[0037] Figure 4 A flow chart of a method of performing three-dimensional CT image reconstruction on a pre-reconstruction image in the embodiment of the present application.

[0038] Figure 5 A user interface schematic diagram before importing a two-dimensional CT image in the embodiment of the present application.

[0039] Figure 6 A user interface schematic diagram after importing a two-dimensional CT image in the embodiment of the present application.

[0040] Figure 7 A user interface schematic diagram after performing three-dimensional CT image reconstruction in the embodiment of the present application.

[0041] Figure 8 A data interaction relationship schematic diagram in the embodiment of the present application. DETAILED DESCRIPTION

[0042] It should be noted that the following detailed description is merely exemplary in nature and is intended to provide further description of the application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0043] It is to be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application.

[0044] In the case of no conflict, the embodiments in the present application and the features in the embodiments can be combined with each other.

[0045] Embodiment One

[0046] The embodiment discloses a three-dimensional CT image reconstruction method for lumbar disease diagnosis.

[0047] As shown in the figure, a three-dimensional CT image reconstruction method for lumbar disease diagnosis comprises the following steps: Figure 1

[0048] Step S1, a plurality of two-dimensional CT images are acquired to construct a training data set.

[0049] Step S2, a deep neural network is trained based on the training data set.

[0050] Step S3, a two-dimensional CT image to be segmented is input into the trained deep neural network for two-dimensional CT image segmentation, and the segmented two-dimensional CT image is taken as a pre-reconstruction image.

[0051] Step S4, a three-dimensional CT image is reconstructed based on a VTK rendering method, specifically: a user interface is built, and a VTK rendering window is embedded in the user interface; a pre-reconstruction image is read based on a reader provided by VTK, and the pre-reconstruction image is converted into three-dimensional volume data based on an entity conversion method; three-dimensional reconstruction is performed on the volume data by using a ray transmission method, the volume data after three-dimensional reconstruction is converted into a data format readable by an image pipeline based on a mapper, and data rendering is performed based on the VTK rendering window; a renderer is used to start a rendering cycle to display the three-dimensional CT image after rendering, and the visualization of the three-dimensional CT image is realized.

[0052] Based on the above process, the present application can accurately and efficiently segment the vertebrae, intervertebral discs and nerve structures in the lumbar CT image at the same time, and realize image visualization by three-dimensional reconstruction of the segmented CT image, thereby providing strong support for lumbar disease diagnosis, preoperative planning and intraoperative guidance. In order to facilitate the understanding of the technical scheme of the present application, the specific embodiments of each step in the technical scheme of the present application are further explained and described below.

[0053] Step S1, a plurality of two-dimensional CT images are acquired to construct a training data set.

[0054] In this embodiment, two-dimensional CT images are collected from the Spine data set, and the training data set is constructed by using the collected two-dimensional CT images. Further, the Spine data set used is collected from Qilu Hospital of Shandong University (Qingdao Branch), and the use of the Spine data set has been approved by the hospital data ethics committee; wherein, the Spine data set contains 200 lumbar CT plain scan data, and in order to better avoid the problem of homogeneity, the 200 lumbar CT plain scan data come from 200 different subjects.

[0055] ​In order to ensure the accuracy of subsequent model training, before inputting the training data set into the deep neural network to train the network model, the operations of uniforming the voxel spacing, image size, and slice number range, etc. need to be performed in sequence. Specifically, the voxel spacing of each CT image in the training data set is set to 0.3467*0.3467*0.999 mm 3 ; at the same time, for all lumbar CT plain scan data in the training data set, the image size is set to 512*512 pixels, and the slice number range is set to 216 to 326. It should be noted that patients may suffer from several spinal diseases / disorders (Note that patients may suffer from several spinal diseases / disorders), in order to better reflect the image content, all lumbar CT plain scan data in the training data set are first manually delineated by a primary expert; then, corrected by an experienced spine surgeon using ITK-SNAP1, and the delineation mask corrected by the spine surgeon is used as the true situation of the spine CT image; wherein ITK-SNAP1 is an open-source medical image analysis software. Therefore, as shown in Figure 2 , each subject will have a lumbar plain scan CT image and a mask corresponding to the lumbar plain scan CT image. In actual implementation, the lumbar plain scan CT image is used as the original image, and the mask image is used as the label to represent the region of interest for the deep neural network model, so as to make the deep neural network pay more attention to these regions of interest; specifically, each vertebra, IVD (intervertebral disc), and nerve in the mask image is assigned a unique label; thus, the training data set contains four different categories, i.e. background, nerve, intervertebral disc, and vertebral body. Through label design, the deep neural network model can better focus on the three main regions of vertebrae, IVD (intervertebral disc), and nerve.

[0056] Step S2, training the deep neural network based on the obtained training data set.

[0057] Before training the deep neural network, the training data set needs to be divided first. Specifically, 5-fold cross-validation is used to divide the training data set into a pre-training set and a test set, for example: if there are 200 data in the training data set, the 200 data are randomly divided into 5 groups of the same number according to a ratio of 4:1, each group containing data of 40 subjects. In the specific implementation process, four groups are selected as the pre-training set, and the remaining one group is used as the test set; first, the deep neural network is trained based on the obtained pre-training set; then, the model performance of the trained deep neural network is verified based on the obtained test set.

[0058] Furthermore, such as Figure 3 As shown, the deep neural network used in this embodiment is a dual-network, namely, a combination of a Res-Net network and a V-Net network. When training and validating the dual-network deep neural network based on the pre-training set and test set, the entire process is implemented iteratively. Specifically, an Adaptive Moment Estimation with Weight Decay (AdamW) is used as the optimizer. Before iterative training of the model, the number of epochs, the initial learning rate, and the weight decay value are pre-set. An epoch is a unit for measuring the model's training progress, referring to the process of the model traversing the entire training data once. In other words, when one epoch is completed, it means that the model has traversed every CT image data in the training set (historical dataset) at least once. When training the deep neural network, the historical dataset can be divided into multiple batches, each containing a certain number of CT images. In each iteration, the model processes samples from one batch, calculates the loss and gradient based on these samples, and then updates the model's weights. This process is repeated until all batches are processed, thus completing one epoch. Furthermore, since a single epoch may not be sufficient for the model to fully learn the features of the data, this embodiment chooses to train the model for multiple epochs. However, too many epochs can also lead to overfitting, meaning the model becomes too complex and begins to memorize the noise and details of individual samples in the historical dataset, rather than learning the general patterns of the data. Therefore, it is necessary to choose an appropriate number of epochs for training the deep neural network.

[0059] Furthermore, in this embodiment, the number of epochs is set to 30, with 200 iterations per epoch; simultaneously, the initial learning rate is set to 0.0001, and the weight decay value is set to 0.0001. Training is terminated if the model performance does not improve after 30 epochs.

[0060] Step S3: Input the two-dimensional CT image to be segmented into the trained deep neural network to perform two-dimensional CT image segmentation, and use the segmented two-dimensional CT image as the pre-reconstructed image.

[0061] First, the vertebral structure is segmented from the 2D CT image using the Res-Net network; then, the intervertebral disc and nerve structure are segmented from the 2D CT image using the V-Net network; finally, the resulting segmented vertebral structure image and the segmented images of the intervertebral disc and nerve structure are integrated and recorded as the pre-reconstructed image.

[0062] Further, the Res-Net network is a deep convolutional neural network with residual connection, and based on the Res-Net network, the segmentation of the vertebral structure of the two-dimensional CT image is carried out, specifically: the Res-Net network with the number of layers of ResNet101 is selected as the basic network for segmenting the vertebral structure, and a full convolution layer is added on the basis of the Res-Net network to perform pixel-level classification; wherein, the network of U-Net or similar structure can be considered to use Res-Net as the encoder part to extract image features; the decoder part can use upsampling and deconvolution operation to gradually restore the spatial resolution of the image and generate the vertebral segmentation result.

[0063] Further, the V-Net network adopts an encoder-decoder structure, has multiple three-dimensional convolution layers, pooling layers, upsampling layers and skip connections; wherein, the encoder part is used to gradually reduce the resolution of the CT image and extract high-level features; the decoder part is used to gradually restore the resolution of the CT image and generate the segmentation result. In a more specific implementation process, first, replace the three-dimensional convolution layer in the V-Net network with a two-dimensional convolution layer to adapt to the processing of the two-dimensional CT image; and segment the intervertebral disc and nerve structure (nerve root) in the CT image through the set pooling layer, upsampling layer and skip connection.

[0064] Step S4, based on the VTK rendering method, the pre-reconstruction image is reconstructed into a three-dimensional CT image, specifically: a user interface is built, and a VTK rendering window is embedded in the user interface; the pre-reconstruction image is read based on the reader provided by VTK, and the pre-reconstruction image is converted into three-dimensional volume data based on the entity conversion method; the volume data is reconstructed into a three-dimensional CT image using the ray transmission method, and the volume data after three-dimensional reconstruction is converted into a data format readable by the image pipeline based on the mapper and rendered based on the VTK rendering window; the renderer is started to display the rendered three-dimensional CT image, and the visualization of the three-dimensional CT image is realized.

[0065] As shown in Figure 4 The reconstruction process of the three-dimensional CT image needs to go through the processes of volume data preprocessing, surface reconstruction and rendering, reconstructed surface object and three-dimensional reconstruction, which are completed based on the cooperation of PyQt5 and VTK rendering. The basic idea is: process the segmented image data and extract information to build a three-dimensional model; then, perform surface rendering and projection to realize multi-view display. In VTK, the geometric shape of the CT image is processed by a drawing, and the drawn graphics are displayed. Further, the several basic object elements involved in the drawing process of the embodiment include:

[0066] ①vtkMapper: A mapper that serves as a bridge between the visualization pipeline and the graphics model, vtkMapper defines the mapping between the basic graphics elements and the rendering data. Its derived classes map through LookupTable and generate corresponding vtkActor graphics elements, and each vtkMapper can be associated with one or more vtkActor.

[0067] ②vtkActor: Represents the graphics entity to be drawn, analogous to an actor on stage; the spatial position, rendering effect, and other attributes of vtkActor can be flexibly customized by adjusting parameters.

[0068] ③vtkRenderer: As the core of scene management, it is similar to a stage, responsible for integrating the properties of drawing objects, light sources, cameras, and other elements, and embedding them into the renderer window vtkRenderWindow to display the complete scene.

[0069] ④vtkRenderWindow: The renderer window, which is the core component for controlling the display window, is like a theater stage, supporting the combined display of multiple vtkRenderWindow, providing functions such as stereoscopic visual effect settings, window size adjustment, etc.

[0070] ⑤vtkProperty: Focuses on enhancing the realism of visual objects, sets the physical properties of objects (such as material, transparency, etc.) through vtkProperty, making three-dimensional data present more realistic visual effects.

[0071] ⑥vtkCamera: Plays the role of observer in the visualization process, manages parameters such as view position, focus, and supports view manipulation such as rotation, translation, and scrolling to optimize user experience.

[0072] ⑦vtkLight: Responsible for scene lighting, vtkLight class can fine-tune light source status (such as on / off), color, intensity, etc., to enhance the level of rendering effects and realism; the created light source can be directly added to vtkRenderer for use.

[0073] ⑧vtkRenderWindowInteractor: Enhances the interactive experience of the rendering window, such as adjusting the vtkCamera view, selecting Actor objects, and processing user-defined messages, providing an intuitive operation interface for users. Specifically, the following steps can be implemented:

[0074] Step S4-1, build a user interface and embed a VTK rendering window in the user interface.

[0075] An initial user interface is built based on the interface tool Qt Designer, and controls are added to the initial user interface; after the user interface is built, the format of the user interface is converted, in preparation for subsequent integration with a VTK rendering window.

[0076] As shown in Figure 5 , the user interface before importing the two-dimensional CT image only has basic designs such as buttons and labels for data interaction with the user; among them, the buttons in the user interface designed in this embodiment include “single-click import CT data” and “single-click start segmentation reconstruction” buttons; the labels can be used to display names or data, for example, the segmentation accuracy, segmentation reconstruction time, etc. can be displayed.

[0077] Step S4-2, reading the pre-reconstruction image based on the reader vtkReader provided by VTK, wherein the user interface after reading the two-dimensional CT image is as shown in Figure 6 ; the pre-reconstruction image is converted into three-dimensional volume data based on the entity conversion method.

[0078] Further, the pre-reconstruction image is converted into three-dimensional volume data based on the entity conversion method, including: first, using the vtkMarchingCubes class of VTK to call the Marching Cubes algorithm, traversing each voxel in the volume data, and determining the position of the isosurface according to the sign change of the vertex value in the voxel; then, using the mapper vtkMapper of VTK to map the generated isosurface data to a geometric representation, and using the actor vtkActor to add the obtained geometric representation to the renderer; wherein the voxel is a three-dimensional pixel.

[0079] Step S4-3, three-dimensional reconstruction of the volume data is performed using the ray casting method, the volume data after three-dimensional reconstruction is converted into a data format readable by the image pipeline based on the vtkMapper mapper, and data rendering is performed based on the VTK rendering window; the renderer is started using the interactor to display the three-dimensional CT image after rendering, realizing the visualization of the three-dimensional CT image, and the user interface after three-dimensional CT image reconstruction is as shown in Figure 7 .

[0080] After the volume data is three-dimensionally reconstructed, the volume data needs to be sequentially subjected to normal calculation, smoothing processing and polygon reduction operation before VTK rendering, so as to ensure the quality of the volume data before rendering.

[0081] First, normal calculation is performed. The normal calculation is to calculate the normal of each triangle in the three-dimensional volume data through the points in the neighborhood, that is:

[0082] n=(P2-P1)(P3-P1);

[0083] Where n represents the normal value formed by the three vertices of the triangle, and P1, P2, and P3 represent the first, second, and third vertices of the triangle, respectively. Then, smoothing is performed. Smoothing is achieved using the Laplace smoothing method, i.e.:

[0084]

[0085] Among them, P j Let n represent the i-th vertex in the triangle, and n represent the distance from vertex P. j The number of adjacent nodes in an adjacent connection. Represent the i-th vertex of the triangle after Laplace smoothing. Finally, reduce the number of polygon faces using the Decimation method.

[0086] Step S4-4: Synchronize the segmentation results and 3D reconstruction results to the user interface for display. For example... Figure 8 As shown, the data interaction in this embodiment uses a three-tier (bottom, middle, and top) architecture design, namely: the human-computer interaction design is implemented through a presentation layer, a business logic layer, and a bottom-level algorithm library. Medical personnel issue user requests through the human-computer interaction interface; the business logic layer receives the user request, executes the corresponding task by calling the relevant bottom-level algorithm library and trained models, and then displays the response result to the presentation layer.

[0087] The upper layer is the presentation layer, providing visualization of spinal CT data. Users interact with the interface and submit requests; this layer displays the system's response, enabling user interaction, CT image display, and data selection. The middle layer is the business logic layer, providing the system's core functionality. This layer is primarily responsible for data processing and, upon receiving user requests, executing corresponding tasks. It then provides the response to the presentation layer for display, enabling functions such as medical image reading, image segmentation and reconstruction, and display. The lower layer contains the various algorithm libraries, underlying frameworks, and trained network models used to complete the functionalities; the middle layer calls these to fulfill the business logic.

[0088] To verify the effectiveness of this invention, this embodiment uses a unified experimental framework to comprehensively evaluate the performance differences between the algorithm of this invention and five mainstream deep learning models (GraphCut, FCN, SegNet, U-Net, and UNetR). This experimental framework strictly follows the same baseline configuration, and all experiments are conducted on PyTorch using an Intel(R) Core(TM) CPU and an NVIDIA GeForce RTX 3090 GPU with 24GB of memory. The experimental results are shown in Table 1.

[0089] Table 1 Comparison of segmentation experiment results

[0090]

[0091] The experimental results show that the segmentation performance of the segmentation network is obviously better than that of other classical methods. Therefore, on the one hand, the three-dimensional reconstruction can be performed on the basis of the two-dimensional CT image, not only reducing the work burden of the spine surgeon, but also through the three-dimensional visualization technology, the doctor can freely switch the perspective and level, deeply understand the internal structure of the patient, and accurately grasp the disease and anatomical details without relying on imagination, thereby providing strong support for lumbar pain diagnosis, preoperative planning and intraoperative guidance. On the other hand, the three spinal structures (disc + vertebra + nerve structure) can be segmented in one step, and on the basis of the three-dimensional reconstruction, each part in the three-dimensional CT image can be clearly distinguished, and the positioning accuracy of the pain source can be significantly improved.

[0092] Embodiment two

[0093] The embodiment discloses a three-dimensional CT image reconstruction system for lumbar disease diagnosis.

[0094] The three-dimensional CT image reconstruction system for lumbar disease diagnosis comprises:

[0095] The data acquisition module is configured to acquire a plurality of two-dimensional CT images and construct a training data set.

[0096] The model training module is configured to train a deep neural network based on the training data set.

[0097] The two-dimensional CT image segmentation module is configured to input a two-dimensional CT image to be segmented into the trained deep neural network for two-dimensional CT image segmentation, and use the segmented two-dimensional CT image as a pre-reconstruction image.

[0098] The three-dimensional CT image reconstruction module is configured to perform three-dimensional CT image reconstruction on the pre-reconstruction image based on a VTK rendering method, specifically: building a user interface and embedding a VTK rendering window in the user interface; reading the pre-reconstruction image based on a reader provided by VTK, and converting the pre-reconstruction image into three-dimensional volume data based on an entity conversion method; performing three-dimensional reconstruction on the volume data by using a ray transmission method, converting the volume data after three-dimensional reconstruction into a data format readable by an image pipeline based on a mapper, and performing data rendering based on the VTK rendering window; using an interactor to start a rendering cycle to display the three-dimensional CT image after rendering, and realizing the visualization of the three-dimensional CT image.

[0099] Embodiment three

[0100] The purpose of the embodiment is to provide a computer readable storage medium.

[0101] A computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the three-dimensional CT image reconstruction method for lumbar disease diagnosis according to the first embodiment of the present disclosure.

[0102] Embodiment Four

[0103] An object of the present embodiment is to provide an electronic device.

[0104] An electronic device, comprising a memory, a processor, and a program stored in the memory and executable on the processor, wherein the processor implements the steps of the three-dimensional CT image reconstruction method for lumbar disease diagnosis according to the first embodiment of the present disclosure when executing the program.

[0105] The steps involved in the devices of the above embodiments two, three and four correspond to the method of the first embodiment, and the specific implementation can be seen from the relevant description of the first embodiment. The term "computer readable storage medium" should be understood to include a single medium or multiple media of one or more instruction sets; it should also be understood to include any medium capable of storing, encoding or carrying a set of instructions for execution by a processor and causing the processor to perform any of the methods of the present disclosure.

[0106] Those skilled in the art should understand that each module or step of the present disclosure described above can be realized by a general computer device, alternatively, they can be realized by program codes executable by a computing device, so that they can be stored in a storage device for execution by a computing device, or they can be respectively manufactured into individual integrated circuit modules, or a plurality of modules or steps among them can be manufactured into a single integrated circuit module to realize. The present disclosure is not limited to any specific combination of hardware and software.

[0107] Although the specific embodiments of the present disclosure are described above in combination with the accompanying drawings, it is not a limitation on the scope of protection of the present disclosure, and those skilled in the art should understand that various modifications or changes made by those skilled in the art on the basis of the technical solutions of the present disclosure without creative labor are still within the scope of protection of the present disclosure.

Claims

1. A three-dimensional CT image reconstruction method for lumbar disease diagnosis, characterized by, The method comprises the following steps: acquiring a plurality of two-dimensional CT images to construct a training data set; training a deep neural network based on the training data set; the deep neural network is a double network, that is, the deep neural network is composed of a Res-Net network and a V-Net network; inputting a two-dimensional CT image to be segmented into the trained deep neural network to segment the two-dimensional CT image, and taking the segmented two-dimensional CT image as a pre-reconstruction image; segmenting the two-dimensional CT image using the deep neural network, which comprises the following steps: first, segmenting the vertebral structure of the two-dimensional CT image based on the Res-Net network; then, segmenting the intervertebral disc and nerve structure of the two-dimensional CT image based on the V-Net network; then, integrating the segmented vertebral structure image and the segmented intervertebral disc and nerve structure image to obtain a pre-reconstruction image; reconstructing a three-dimensional CT image based on a VTK rendering method, specifically: building a user interface and embedding a VTK rendering window in the user interface; reading the pre-reconstruction image based on a reader provided by VTK, converting the pre-reconstruction image into three-dimensional volume data based on a solid conversion method; performing three-dimensional reconstruction on the volume data using a ray transmission method, converting the three-dimensionally reconstructed volume data into a data format readable by an image pipeline based on a mapper, and performing data rendering based on the VTK rendering window; starting a rendering cycle using an interactor to display the three-dimensionally reconstructed CT image, thereby realizing the visualization of the three-dimensional CT image; after the volume data is three-dimensionally reconstructed and before VTK rendering, sequentially performing normal calculation, smoothing processing and polygon reduction operations on the volume data to ensure the quality of the volume data before rendering.

2. The three-dimensional CT image reconstruction method for lumbar disease diagnosis according to claim 1, wherein, converting the pre-reconstruction image into three-dimensional volume data based on a solid conversion method, specifically: first, using the vtkMarchingCubes class of VTK to call the Marching Cubes algorithm, traversing each voxel in the volume data, and determining the position of the isosurface according to the sign change of the vertex value in the voxel; then, using the mapper of VTK to map the generated isosurface data into a geometric representation, and using the actor vtkActor to add the obtained geometric representation to the renderer; wherein the voxel is a three-dimensional pixel.

3. The three-dimensional CT image reconstruction method for lumbar disease diagnosis according to claim 1, wherein, building a user interface, which comprises: building an initial user interface based on the interface tool Qt Designer, adding controls to the initial user interface; after the user interface is built, converting the format of the user interface for subsequent integration with the VTK rendering window.

4. The three-dimensional CT image reconstruction method for lumbar disease diagnosis according to claim 3, wherein, the controls include buttons and labels; wherein the buttons include a single-click CT data import button and a single-click start segmentation and reconstruction button.

5. A three-dimensional CT image reconstruction system for lumbar disease diagnosis, which adopts the three-dimensional CT image reconstruction method for lumbar disease diagnosis according to any one of claims 1 to 4, characterized in that, The method comprises the following steps: a data acquisition module configured to acquire a plurality of two-dimensional CT images to construct a training data set; a model training module configured to train a deep neural network based on the training data set; a two-dimensional CT image segmentation module configured to input a two-dimensional CT image to be segmented into the trained deep neural network to segment the two-dimensional CT image, and take the segmented two-dimensional CT image as a pre-reconstruction image; The three-dimensional CT image reconstruction module is configured to perform three-dimensional CT image reconstruction on the pre-reconstruction image based on a VTK rendering method, specifically: a user interface is built, and a VTK rendering window is embedded in the user interface; a pre-reconstruction image is read based on a reader provided by VTK, and the pre-reconstruction image is converted into three-dimensional volume data based on a solid conversion method; The volume data is reconstructed in three dimensions by using a ray transmission method, the volume data after three-dimensional reconstruction is converted into a data format readable by an image pipeline based on a mapper, and data rendering is performed based on the VTK rendering window; a rendering cycle is started by using an interactor, so as to display the three-dimensional CT image after rendering, and the visualization of the three-dimensional CT image is realized.

6. A computer-readable storage medium having stored thereon a program, characterized in that, The program is executed by the processor to implement the steps in the three-dimensional CT image reconstruction method for lumbar disease diagnosis according to any one of claims 1-4.

7. An electronic device comprising a memory, a processor, and a program stored in the memory and capable of running on the processor, characterized by The processor executes the program to implement the steps in the three-dimensional CT image reconstruction method for lumbar disease diagnosis according to any one of claims 1-4.

Images