Two-stage liver CT segmentation method based on axial position and sagittal position and related device

Abstract

The invention discloses a two-stage liver CT segmentation method based on axial position and sagittal position and a related device, and the method comprises the steps: obtaining a liver CT image, carrying out the preprocessing of the liver CT image, and obtaining an axial position image containing a liver region; inputting the axial image containing the liver region into a U-Net improved network to obtain a liver segmentation mask region; and mapping the liver segmentation mask region to a sagittal view and then inputting the sagittal view into a deep learning network based on a diffusion model to obtain a liver segmentation result of the liver CT, and the method and the related device can perform accurate liver segmentation on the two-stage liver CT.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology and relates to a two-stage liver CT segmentation method and related device based on axial and sagittal planes. Background Technology

[0002] Liver diseases such as liver cancer, fatty liver, and cirrhosis have high incidence and mortality rates worldwide. Computed tomography (CT) scans, as a widely used imaging technique in clinical practice, play a crucial role in the detection, localization, and preoperative planning of liver lesions. With the rapid increase in medical image data, achieving high-precision, automated liver segmentation has become a key aspect of the development of computer-aided diagnostic systems and surgical navigation systems. Existing liver segmentation methods can be mainly divided into two categories: those based on traditional image processing techniques and those based on deep learning.

[0003] Traditional methods, such as region growing, thresholding, active contour models, or graph cut, while effective under certain conditions, generally suffer from sensitivity to image quality, strong parameter dependence, and poor generalization ability. They are particularly prone to inaccurate segmentation or missing regions when dealing with low contrast, high noise levels, or blurred boundaries between the liver and surrounding organs.

[0004] Deep learning-based methods have become mainstream in recent years, especially encoder-decoder structures like U-Net, which demonstrate excellent segmentation performance in 2D CT images due to their strong representation capabilities of small-sample features in medical images. However, most methods only train and predict on axial slices, which has the following limitations:

[0005] Lack of three-dimensional structural information: The lack of contextual relationships between cross-sectional sequences can easily lead to structural inconsistencies between slices.

[0006] Unclear boundaries: In the complex junction areas of organs, such as the blurred boundaries between the liver and stomach, or between the liver and kidney, inaccurate segmentation often occurs;

[0007] Single perspective limits understanding: Different perspectives in medical imaging contain complementary information, and relying solely on axial cross-sections is insufficient to fully reflect the morphology of the liver.

[0008] In addition, some studies have attempted to use 3D convolutional neural networks for liver segmentation to fuse volume information, but the amount of computational data is too large to meet clinical requirements. Summary of the Invention

[0009] The purpose of this invention is to overcome the shortcomings of the prior art and provide a two-stage liver CT segmentation method and related device based on axial and sagittal planes. This method and related device can accurately segment the liver using two-stage liver CT.

[0010] To achieve the above objectives, this invention discloses a two-stage liver CT segmentation method based on axial and sagittal planes, comprising:

[0011] Acquire liver CT images, and preprocess the liver CT images to obtain axial images containing the liver region;

[0012] The axial image containing the liver region is input into the improved U-Net network to obtain the liver segmentation mask region;

[0013] The liver segmentation mask region is mapped to the sagittal plane and then input into a deep learning network based on a diffusion model to obtain the liver segmentation result of liver CT.

[0014] The further improvement of the dual-stage liver CT segmentation method based on axial and sagittal planes described in this invention lies in:

[0015] Furthermore, the preprocessing process for the liver CT image is as follows:

[0016] The liver CT images were optimized by sequentially adjusting grayscale values ​​and histogram equalization.

[0017] Furthermore, the step of inputting the axial image containing the liver region into the U-Net improved network also includes:

[0018] Obtain the U-Net network, introduce residual connection structures into the encoder and decoder modules within the U-Net network, and introduce VSS modules into the skip connections of the U-Net network to obtain the improved U-Net network.

[0019] Furthermore, the step of inputting the axial image containing the liver region into the U-Net improved network also includes:

[0020] The improved U-Net network is trained by introducing cross-entropy loss.

[0021] Furthermore, the deep learning network based on the diffusion model includes a conditional encoder and a segmentation encoder, both of which incorporate an Inception module.

[0022] Furthermore, an Inception module is added between the SiLU layer and the convolutional layer in the segmentation encoder.

[0023] Furthermore, an Inception module is added between the ReLU layer and the batch normalization layer in the segment encoder.

[0024] This invention discloses a two-stage liver CT segmentation system based on axial and sagittal planes, comprising:

[0025] The preprocessing module is used to acquire liver CT images and preprocess the liver CT images to obtain axial images containing the liver region.

[0026] The first segmentation module is used to input the axial image containing the liver region into the U-Net improved network to obtain the liver segmentation mask region;

[0027] The second segmentation module is used to map the liver segmentation mask region to the sagittal plane and then input it into a deep learning network based on a diffusion model to obtain liver segmentation results based on CT images.

[0028] The present invention discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the two-stage liver CT segmentation method based on axial and sagittal planes.

[0029] The present invention discloses a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the two-stage liver CT segmentation method based on axial and sagittal planes.

[0030] The present invention has the following beneficial effects:

[0031] The dual-stage liver CT segmentation method and related device based on axial and sagittal planes described in this invention preprocesses the liver CT image to obtain an axial image containing the liver region, and then uses an improved U-Net network and a deep learning network based on a diffusion model to perform two-stage segmentation of the liver CT to obtain the liver segmentation result of the liver CT. The segmentation is relatively accurate and highly practical. Attached Figure Description

[0032] The accompanying drawings, which form part of this specification, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0033] Figure 1 This is a flowchart of the method of the present invention;

[0034] Figure 2a This is a schematic diagram of a liver CT image;

[0035] Figure 2b A schematic diagram of a liver CT image after preprocessing;

[0036] Figure 3 This is a flowchart of the preprocessing process in this invention;

[0037] Figure 4 An improved network structure diagram for U-Net;

[0038] Figure 5 This is a schematic diagram of the VSS module;

[0039] Figure 6 This is a schematic diagram of the SS2D module;

[0040] Figure 7 This is a structural diagram of a deep learning network based on a diffusion model.

[0041] Figure 8 This is a schematic diagram of the Inception module;

[0042] Figure 9a A diagram showing the location of the Inception module in the segment encoder;

[0043] Figure 9b This is a location diagram of the Inception module in the conditional encoder;

[0044] Figure 10 This is a visualization of the experimental results for a confirmatory experiment. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] In the description of this invention, it should be understood that the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0047] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0048] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Additionally, the character " / " in this invention generally indicates that the preceding and following objects have an "or" relationship.

[0049] It should be understood that although terms such as first, second, third, etc., may be used in the embodiments of the present invention to describe the preset range, these preset ranges should not be limited to these terms. These terms are only used to distinguish the preset ranges from one another. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0050] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0051] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. 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 claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0052] The accompanying drawings illustrate various structural schematic diagrams according to embodiments disclosed in this invention. These drawings are not to scale, and some details have been enlarged for clarity, and some details may have been omitted. The shapes of the various regions and layers shown in the drawings, as well as their relative sizes and positional relationships, are merely exemplary and may deviate from reality due to manufacturing tolerances or technical limitations. Furthermore, those skilled in the art can design regions / layers with different shapes, sizes, and relative positions as needed.

[0053] Example 1

[0054] refer to Figure 1 The dual-stage liver CT segmentation method based on axial and sagittal planes described in this invention includes the following steps:

[0055] 1) Acquire liver CT images, preprocess the liver CT images to obtain axial images containing the liver region;

[0056] The specific operation of step 1) is as follows:

[0057] To address the characteristics of 3D liver CT data, the original liver CT images and label data were first acquired to obtain axial images containing the liver region. Since CT images are 16-bit (0-65535), and the liver displays well under specific window widths and levels, the grayscale values ​​of the images were mapped to 8-bit (0-255) by adjusting the window width and level, significantly enhancing the visualization of the liver edges. To further enhance the image quality, histogram equalization was used to process the images and reduce background noise interference, such as... Figure 2a and Figure 2b As shown, after spatial cropping and thresholding, the target anatomical structures were accurately extracted, resulting in 19211 valid samples and their corresponding labels. The overall preprocessing workflow is as follows: Figure 3 As shown, to ensure the rigor of the experiment, the dataset was randomly divided into a training set (70%), a validation set (10%), and a test set (20%) for subsequent network training, in order to ensure the objectivity of the evaluation results.

[0058] 2) Input the axial image sequence containing the liver region into the improved U-Net network to obtain the liver segmentation mask;

[0059] It should be noted that this invention introduces residual connection structures into the encoder and decoder modules of the U-Net network to improve the model's ability to recognize liver regions, especially boundary regions. Simultaneously, a VSS module is introduced into the skip connections of the U-Net network to enhance multi-scale feature fusion and the ability to express contextual information. The structure of the improved U-Net network is referenced below. Figure 4 .

[0060] It should be noted that the U-Net network is prone to gradient vanishing or gradient exploding problems, and its ability to recognize blurred boundaries or low-contrast regions decreases as the network deepens. To address this issue, this invention introduces residual connection structures in the paths of both the encoder and decoder modules within the U-Net network. Specifically, each encoder or decoder module consists of two convolutional layers and a residual branch. The residual branch uses a 1×1 convolution to adjust the input dimension and then directly adds it to the output of the main branch; the activation function is ReLU to maintain non-linear expressiveness.

[0061] Furthermore, to fully utilize the information transfer between the encoder and decoder modules, this invention optimizes the skip connections in the U-Net network by introducing a VSS module. Specifically, the input features are divided into two branches through a linear embedding layer. One branch is first processed by depthwise convolution and the SiLU activation function, then fed into the SS2D module. After layer normalization, it is merged with the other branch through SiLU activation. Figure 5 As shown.

[0062] The data transfer process of the SS2D module is as follows: Figure 6 As shown, the process includes cross-scan, S6-based selective scanning, and cross-merge. Specifically, the input image patch is first expanded into four sequences, each traversing four different paths. Each sequence is processed in parallel by an independent S6 module. The resulting sequences are then reshaped and merged to form the final output image. The SS2D module combines information from each pixel in the image with information from other pixels. This information propagates from different directions, helping the network establish a global receptive field in 2D space.

[0063] In addition, this invention introduces cross-entropy loss during the training process of the U-Net improved network to enhance the network's modeling accuracy of the liver region's boundaries and regional coherence.

[0064] 3) After mapping the liver segmentation mask to the sagittal plane, input it into the trained deep learning network based on the diffusion model to obtain the liver segmentation result of the two-stage liver CT.

[0065] Specifically, the image of the liver segmentation mask is mapped to a sagittal view to obtain a resampled liver mask, thereby acquiring richer spatial information. This resampled sagittal liver mask is used as conditional input to provide preliminary region information, guiding the diffusion-based deep learning network to perform more refined boundary refinement. The main architecture of the diffusion-based deep learning network is based on the ResNet encoder and U-Net decoder, and its network structure is as follows: Figure 7 As shown, in the forward inference phase of the model, Gaussian noise is first gradually added to the segmentation label x0 to simulate the data degradation process; while in the reverse generation phase, noise is gradually removed through reverse restoration operations to reconstruct data that approximates the original label, i.e.:

[0066]

[0067] Where, p θ (x t-1 |x t ) indicates that the condition is x t At that time, x t-1 The probability distribution of p;θ (x 0:T-1 |x T ) indicates that the condition is x T When x0 to x T-1 The probability distribution, where θ is the parameter for the denoising process, is expressed by the Gaussian noise formula as:

[0068]

[0069] Where I represents the identity matrix, with a size of n×n, and N(x T ;0,I n×n ) represents the sample x at time step T. T Following a Gaussian distribution with mean 0 and covariance matrix I, p is denoised after a denoising process. θ (=x T Restore to p θ (x0), to obtain the final segmentation result, using the original image as a priori condition, and setting the step size estimation function ∈ as:

[0070]

[0071] in, Conditional feature embedding for the original image, This is the feature embedding of the segmentation result at the current step t.

[0072] Compare the original input image I with the label x under noise perturbation. t The features are fed into two independent encoders for feature extraction. Each encoder consists of multiple stacked residual modules, each containing two convolutional sub-units. Each convolutional sub-unit comprises a group normalization layer, a SiLU activation function (Sigmoid-weighted Linear Unit), and a convolutional layer, sequentially. During encoding, the features output from the first three layers of the two encoders are denoted as E. I and After feature fusion, the two are input to the last layer of the encoder to further integrate image information and label noise features. Then, the fused deep features are passed to the decoder, which performs progressive upsampling and restoration of the fused features to finally generate a refined segmentation prediction result.

[0073] This invention incorporates Inception modules into both the second and third layers of the conditional encoder and the segment encoder, with the following structure: Figure 8As shown, the introduction of the Inception module enhances multi-scale information in the deep feature extraction stage of the encoder, improving the model's ability to understand lesions of different sizes and anatomical structures. The network structures of the conditional encoder and segmentation encoder, and the insertion position of the Inception module are respectively as follows: Figure 9a and Figure 9b As shown.

[0074] Through the refinement in the second stage, the boundary clarity of the liver segmentation results is significantly improved, and the boundary between the liver and adjacent structures (such as the stomach and kidneys) can be accurately distinguished.

[0075] Confirmatory Experiment

[0076] All experiments were built on the PyTorch framework and run on a single NVIDIA RTX 3080 GPU. All images were uniformly resized to 256×256 pixels. The network was trained end-to-end using the standard Adam optimizer, with 100 diffusion steps for inference. The initial learning rate was set to 1×10⁻⁶. -4 The iteration count is set to 200, the batch size to 10, and an early stopping condition is set. The loss function is the Tversky function.

[0077]

[0078] Where α is 0.7, β is 0.3, and V seg V represents the segmentation result obtained by the computer. gt For accurate segmentation results.

[0079] The DSC and HD95 evaluation results of this invention are 92.02% and 24.52 mm, respectively, indicating that this invention has better segmentation ability in terms of liver segmentation results. Its visualization results are as follows: Figure 10 As shown.

[0080] Example 2

[0081] The dual-stage liver CT segmentation system based on axial and sagittal planes described in this invention includes:

[0082] The preprocessing module is used to acquire liver CT images and preprocess the liver CT images to obtain axial images containing the liver region.

[0083] The first segmentation module is used to input the axial image containing the liver region into the U-Net improved network to obtain the liver segmentation mask region;

[0084] The second segmentation module is used to map the liver segmentation mask region to the sagittal plane and then input it into a deep learning network based on a diffusion model to obtain the liver segmentation result of liver CT.

[0085] In this embodiment, the preprocessing process for the liver CT image is as follows:

[0086] The axial liver CT images were optimized by sequentially adjusting grayscale values ​​and histogram equalization.

[0087] In this embodiment, the step of inputting the axial image containing the liver region into the U-Net improved network further includes:

[0088] Obtain the U-Net network, introduce residual connection structures into the encoder and decoder modules within the U-Net network, and introduce VSS modules into the skip connections of the U-Net network to obtain the improved U-Net network.

[0089] In this embodiment, the step of inputting the axial image containing the liver region into the U-Net improved network further includes:

[0090] The improved U-Net network is trained by introducing cross-entropy loss.

[0091] In this embodiment, the deep learning network based on the diffusion model includes a conditional encoder and a segmentation encoder, and both the conditional encoder and the segmentation encoder incorporate an Inception module.

[0092] In this embodiment, an Inception module is added between the SiLU layer and the convolutional layer in the segment encoder.

[0093] In this embodiment, an Inception module is added between the ReLU layer and the batch normalization layer in the segment encoder.

[0094] The module division in this embodiment is illustrative and represents only one logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional modules in each embodiment of this application can be integrated into a single processor, exist as separate physical entities, or be integrated into a single module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0095] Example 3

[0096] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a two-stage liver CT segmentation method based on axial and sagittal planes. For example, the method includes: acquiring a liver CT image; preprocessing the liver CT image to obtain an axial image containing the liver region; inputting the axial image containing the liver region into an improved U-Net network to obtain a liver segmentation mask region; mapping the corresponding region of the liver segmentation mask to a sagittal plane and then inputting it into a deep learning network based on a diffusion model to obtain the liver segmentation result from the liver CT scan. The memory may include main memory, such as high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device. The processor, network interface, and memory are interconnected via an internal bus, which can be an industry-standard architecture bus, a peripheral component interconnection standard bus, an extended industry-standard architecture bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. The memory is used to store the program; specifically, the program may include program code, which includes computer operation instructions. Memory can include main memory and non-volatile memory, and provides instructions and data to the processor.

[0097] Example 4

[0098] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of a two-stage liver CT segmentation method based on axial and sagittal planes. For example, the method includes: preprocessing a liver CT image to obtain an axial image containing the liver region; inputting the axial image containing the liver region into an improved U-Net network to obtain a liver segmentation mask region; mapping the liver segmentation mask region to a sagittal plane and then inputting it into a deep learning network based on a diffusion model to obtain the liver segmentation result from the liver CT scan. Specifically, the computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. The volatile memory may include random access memory (RAM) and / or cache memory, etc. The non-volatile memory may include read-only memory (ROM), hard disk, flash memory, optical disk, magnetic disk, etc.

[0099] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0100] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations 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, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0101] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0102] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0103] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and disclosure of the invention. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.

[0104] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

[0105] The above description is merely a preferred embodiment of the present invention and does not constitute any limitation on the present invention. Any simple modifications, alterations, or equivalent structural changes made to the above embodiments based on the technical essence of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A two-stage liver CT segmentation method based on axial and sagittal views, characterized in that, include: Acquire liver CT images, and preprocess the liver CT images to obtain axial images containing the liver region; The axial image containing the liver region is input into the improved U-Net network to obtain the liver segmentation mask region; The liver segmentation mask region is mapped to the sagittal plane and then input into a deep learning network based on a diffusion model to obtain the liver segmentation result of liver CT.

2. The dual-stage liver CT segmentation method based on axial and sagittal planes according to claim 1, characterized in that, The preprocessing process for the liver CT images is as follows: The liver CT images were optimized by sequentially adjusting grayscale values ​​and histogram equalization.

3. The dual-stage liver CT segmentation method based on axial and sagittal planes according to claim 1, characterized in that, The process of inputting the axial image containing the liver region into the improved U-Net network also includes: Obtain the U-Net network, introduce residual connection structures into the encoder and decoder modules within the U-Net network, and introduce VSS modules into the skip connections of the U-Net network to obtain the improved U-Net network.

4. The dual-stage liver CT segmentation method based on axial and sagittal planes according to claim 1, characterized in that, The process of inputting the axial image containing the liver region into the improved U-Net network also includes: The improved U-Net network is trained by introducing cross-entropy loss.

5. The liver CT segmentation method based on axial and sagittal planes according to claim 1, characterized in that, The deep learning network based on the diffusion model includes a conditional encoder and a segmentation encoder, both of which incorporate an Inception module.

6. The dual-stage liver CT segmentation method based on axial and sagittal planes according to claim 5, characterized in that, An Inception module is added between the SiLU layer and the convolutional layer in the segment encoder.

7. The dual-stage liver CT segmentation method based on axial and sagittal planes according to claim 6, characterized in that, An Inception module is added between the ReLU layer and the batch normalization layer in the segment encoder.

8. A two-stage liver CT segmentation system based on axial and sagittal planes, characterized in that, include: The preprocessing module preprocesses the axial CT image of the liver to obtain an image containing the liver region; The first segmentation module is used to input the axial sequence containing the liver region into the U-Net improved network to obtain the liver segmentation mask region; The second segmentation module is used to map the liver segmentation mask region to the sagittal plane and then input it into a deep learning network based on a diffusion model to obtain the liver segmentation result of liver CT.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the two-stage liver CT segmentation method based on axial and sagittal planes as described in any one of claims 1-7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the two-stage liver CT segmentation method based on axial and sagittal planes as described in any one of claims 1-7.

Images