Aortic dissection segmentation apparatus, storage medium, and electronic device

Abstract

Embodiments of the present application provide an aortic dissection segmentation device, a computer readable storage medium and an electronic device, comprising: a preprocessing module configured to preprocess first medical image data containing aortic information to obtain preprocessed second medical image data; a positioning module configured to detect the second medical image data using a region detection model, determine an aortic region in the second medical image data, and determine lesion position information of an aortic dissection in the aortic region using a positioning model; a region determination module configured to determine a first region of interest containing the aortic dissection in the second medical image data according to the second medical image data and the lesion position information; and a segmentation module configured to perform multi-class segmentation on the first region of interest using a segmentation model to obtain segmentation results corresponding to different pathological characteristics of the aortic dissection. Thus, the aortic dissection can be automatically and accurately positioned and multi-class segmented.

Description

Technical Field

[0001] This invention relates to the field of medical image processing technology, and in particular to an aortic dissection segmentation device, a computer-readable storage medium, and an electronic device. Background Technology

[0002] Aortic dissection is a serious cardiovascular disease characterized by a tear in the intima of the aorta, allowing blood to seep into the aortic wall's media, creating a true lumen and a false lumen. Aortic dissections can be classified into different types based on their extent and location, such as Stanford type A and Stanford type B. Type B aortic dissection primarily refers to a dissection originating in the descending aorta and potentially extending to the abdominal aorta. However, type B aortic dissections are often concentrated in a specific segment of the aorta, rather than the entire aorta, and this is relatively common, accounting for approximately 30% to 40% of all aortic dissection cases. Therefore, locating the dissection in this type is more challenging.

[0003] Currently, clinical diagnosis mainly relies on imaging techniques, such as CT (Computed Tomography). Based on the obtained CT image data, doctors manually locate and segment the aortic dissection. However, this process is time-consuming and prone to errors, especially in the acute phase when a treatment plan needs to be quickly formulated. It is difficult to accurately locate the aortic dissection and perform multiple segments, including the true lumen, false lumen, and tear site, which affects the formulation of the treatment plan and consequently the treatment outcome.

[0004] Therefore, how to automatically and accurately locate and segment aortic dissection has become an urgent problem to be solved in this field. Summary of the Invention

[0005] In view of the above, embodiments of this application provide an aortic dissection segmentation device, a computer-readable storage medium, and an electronic device to solve at least one problem existing in the prior art.

[0006] In a first aspect, embodiments of this application provide an aortic dissection segmentation device, comprising:

[0007] The preprocessing module is used to preprocess the first medical image data containing aortic information to obtain the preprocessed second medical image data.

[0008] The positioning module is used to detect the second medical image data using a region detection model, determine the aortic region in the second medical image data, and determine the lesion location information of aortic dissection in the aortic region using the positioning model.

[0009] The region determination module is used to determine a first region of interest containing the aortic dissection in the second medical image data based on the second medical image data and the lesion location information;

[0010] The segmentation module uses a segmentation model to perform multi-class segmentation on the first region of interest to obtain segmentation results corresponding to different pathological features of the aortic dissection.

[0011] In conjunction with the first aspect of this application, in an optional embodiment, the positioning module is further configured to: after determining the aortic region in the second medical image data, use a region growing algorithm to identify the aortic wall in the aortic region;

[0012] The positioning module is specifically used for:

[0013] Based at least on the identified aortic wall, the localization model is used to determine the lesion location information of the aortic dissection in the aortic region.

[0014] In conjunction with the first aspect of this application, in an optional embodiment, the lesion location information includes: the center coordinate information of the aortic dissection and the size information of the aortic dissection;

[0015] The region determination module is specifically used for:

[0016] The center coordinates of the first region of interest are determined based on the center coordinate information of the aortic dissection, and the boundary of the first region of interest is determined based on the size information of the aortic dissection, so as to determine the first region of interest containing the aortic dissection in the second medical image.

[0017] In conjunction with the first aspect of this application, in an optional embodiment, the segmentation module is further configured to perform at least one of the following:

[0018] Output a 2D slice view based on the segmentation results;

[0019] A 3D reconstructed view is constructed based on the 2D slice views corresponding to multiple primary medical image data.

[0020] In conjunction with the first aspect of this application, in an optional embodiment, the apparatus further includes: a first model training module, configured to:

[0021] The fourth medical image data, obtained by preprocessing the third medical image data through the preprocessing module, is used as the input to the localization model, and the predicted lesion location information output by the localization model is obtained.

[0022] Based on the predicted lesion location information and the actual lesion location information of the third medical image data, a first loss function is used to train the localization model; wherein, the first loss function includes mean squared error (MSE) and Dice loss.

[0023] In conjunction with the first aspect of this application, in an optional embodiment, the preprocessing module is further configured to: normalize the actual lesion location information to normalize the lesion coordinates indicated by the actual lesion location information.

[0024] In conjunction with the first aspect of this application, in an optional embodiment, the apparatus further includes: a second model training module, configured to:

[0025] The second region of interest of the fifth medical image data determined by the region determination module is used as the input of the segmentation model, and the predicted segmentation result output by the segmentation model is obtained.

[0026] Based on the predicted segmentation result and the actual segmentation result of the fifth medical image data, a second loss function is used to train the segmentation model; wherein, the second loss function includes multi-class cross-entropy loss and Dice loss.

[0027] In conjunction with the first aspect of this application, in an optional embodiment, the apparatus further includes: a second model training module, configured to:

[0028] Obtain the second region of interest of the fifth medical image data determined by the region determination module;

[0029] Image deformation processing is performed on the second region of interest;

[0030] The second region of interest, after the image deformation processing is completed, is used as the input to the segmentation model, and the predicted segmentation result of the segmentation model is obtained.

[0031] Based on the predicted segmentation result and the actual segmentation result of the fifth medical image data, a second loss function is used to train the segmentation model; wherein, the second loss function includes multi-class cross-entropy loss and Dice loss.

[0032] Secondly, embodiments of this application provide a computer-readable storage medium storing instructions that, when executed by a processor of an electronic device, enable the electronic device to control the aortic dissection segmentation device described in any of the first aspects above.

[0033] Thirdly, embodiments of this application provide an electronic device, the electronic device comprising:

[0034] processor;

[0035] Memory used to store executable instructions for a computer;

[0036] The processor is configured to execute the computer-executable instructions to control the aortic dissection segmentation device according to any one of the first aspects described above.

[0037] The aortic dissection segmentation device, computer-readable storage medium, and electronic device provided in this application preprocess the first medical image data containing aortic information through a preprocessing module to obtain preprocessed second medical image data. Then, a localization module uses a region detection model to detect the second medical image data, determine the aortic region within the second medical image data, and uses the localization model to determine the lesion location information of aortic dissection within the aortic region. Next, a region determination module determines a first region of interest containing aortic dissection in the second medical image based on the second medical image and the lesion location information. Finally, a segmentation module uses a segmentation model to perform multi-class segmentation on the first region of interest to obtain segmentation results corresponding to different pathological features of aortic dissection. Thus, preprocessing the input first medical image data through the preprocessing module helps improve the accuracy of subsequent localization. Firstly, the localization module obtains the lesion location information of aortic dissection, and then the region determination module obtains the first region of interest, ensuring that the aortic dissection is included within the first region of interest, thus improving the accuracy of obtaining the first region of interest. Based on this, the segmentation module segments the first region of interest, which helps improve the accuracy of the segmentation results, avoids unnecessary computational waste, and obtains segmentation results corresponding to different pathological features of aortic dissection, providing more detailed and reliable lesion information and assisting in the formulation of more effective treatment plans. The aortic dissection segmentation device automatically locates and segments the aortic dissection from the first medical image data through the preprocessing module, localization module, region determination module, and segmentation module, avoiding the large errors and time-consuming nature of manual operation, thereby improving the efficiency and accuracy of diagnosis.

[0038] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0039] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0040] Figure 1 This is a schematic diagram of the aortic dissection segmentation device provided in the embodiments of this application;

[0041] Figure 2 A schematic diagram illustrating the segmentation principle of the segmentation model provided in this application embodiment;

[0042] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0043] To make the technical solutions and beneficial effects of the present invention more apparent and understandable, the technical solutions in the embodiments of this application are clearly and completely described below by listing specific examples. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0044] Figure 1 This is a schematic diagram of the aortic dissection segmentation device provided in an embodiment of this application. Figure 1 As shown, the aortic dissection segmentation device includes:

[0045] The preprocessing module is used to preprocess the first medical image data containing aortic information to obtain the preprocessed second medical image data.

[0046] The localization module is used to detect the second medical image data using a region detection model, determine the aortic region in the second medical image data, and use the localization model to determine the location information of the aortic dissection lesion in the aortic region.

[0047] The region determination module is used to determine the first region of interest containing aortic dissection in the second medical imaging data based on the second medical imaging data and lesion location information.

[0048] The segmentation module uses a segmentation model to perform multi-class segmentation on the first region of interest to obtain segmentation results corresponding to different pathological features of aortic dissection.

[0049] The aortic dissection segmentation device provided in this embodiment preprocesses the input first medical image data through a preprocessing module, which helps improve the accuracy of subsequent localization. First, the localization module obtains the lesion location information of the aortic dissection, and then the region determination module obtains the first region of interest, ensuring that the aortic dissection is included in the first region of interest, thus improving the accuracy of obtaining the first region of interest. Based on this, the segmentation module segments the first region of interest, which helps improve the accuracy of the segmentation results, avoids unnecessary computational waste, and obtains segmentation results corresponding to different pathological features of the aortic dissection, providing more detailed and reliable lesion information and assisting in the formulation of more effective treatment plans. The aortic dissection segmentation device automatically locates and segments the aortic dissection from the first medical image data through the preprocessing module, localization module, region determination module, and segmentation module, avoiding the large errors and time-consuming nature of manual operation, thus improving the efficiency and accuracy of diagnosis.

[0050] It should be noted that the aortic dissection segmentation device provided in this application embodiment can be implemented by a computer, server or other device with computing power.

[0051] In this embodiment, the first medical image data is specifically, for example, CT image data. The CT image data contains multiple slices, each with a resolution of 512x512.

[0052] Optionally, the preprocessing includes at least one of the following methods: denoising, contrast enhancement, and normalization. Thus, denoising removes random noise from the first medical image data, improving its quality; contrast enhancement increases the grayscale difference between the aortic image and the surrounding environment, enhancing the aortic boundary and making it stand out from the surrounding tissue, thus facilitating the differentiation between the aortic image and the environment; normalization adjusts each pixel value in the data to a standard normal distribution with a mean of 0 and a standard deviation of 1, simplifying processing.

[0053] Furthermore, the denoising process includes bilateral filtering. This allows for the removal of noise from the first medical image data while preserving the edge information of the slices.

[0054] Furthermore, the contrast enhancement process includes adaptive histogram equalization. Adaptive histogram equalization has local processing characteristics, which can better adapt to regions with different contrast in the image, and while enhancing contrast, it can better preserve the details and texture of the image; through adaptive adjustment, it can avoid over-enhancing certain areas, which would lead to overexposure or loss of detail in the image.

[0055] Optionally, the region detection model includes a machine learning model. Thus, after pre-training, the region detection model can be used to identify specific objects in medical imaging data. For example, the region detection model can be used to identify the aortic region in medical imaging data. In this embodiment, the aortic region may include the entire aorta and adjacent surrounding tissues in the second medical imaging data.

[0056] In this embodiment, the region detection model may include a ResNet (Residual Network) model. Understandably, the ResNet model contains fully connected layers and alternating layers of convolutional and pooling layers. The convolutional and pooling layers are used to extract deep features from the second medical image data, while the fully connected layers are used to reflect the extracted features as the aortic region detection result. The ResNet model has good deep feature extraction capabilities, thus extracting richer feature information, which helps improve the sensitivity and specificity of detection and reduce misdiagnosis and missed diagnosis; it is also less prone to gradient vanishing, exhibiting good stability and generalization ability, and avoiding overfitting.

[0057] Optionally, the localization module is further configured to: after determining the aortic region in the second medical image data, use a region growing algorithm to identify the aortic wall within the aortic region; specifically, the localization module is configured to: at least based on the identified aortic wall, use a localization model to determine the lesion location information of aortic dissection within the aortic region. Understandably, in this embodiment, the aortic region includes the aorta and some adjacent environmental tissues. After determining the aortic region, further identifying the aortic wall can more accurately define the aortic boundary. The localization model determines the lesion location information based on the identified aortic wall, which helps improve the localization accuracy of the localization model. The region growing algorithm can adapt to aortic walls of different shapes and thicknesses, which helps to more comprehensively capture the morphological features of the aortic wall and enhances the robustness of the localization module.

[0058] Specifically, the region growing algorithm is used to: use one or more pixels in the aortic wall of the aortic region as seed points for region growing; check the difference between the gray values ​​of the neighboring pixels of the seed point and the gray value of the seed point, and identify neighboring pixels whose differences are within a specific range as part of the aortic wall; continue growing using these neighboring pixels as seed points until all pixels have been checked, thus completing the identification of the aortic wall. Understandably, in medical imaging data, the gray values ​​of pixels constituting the same tissue should be similar, that is, the difference between the gray values ​​of pixels should be within a specific range. If the gray values ​​of two pixels differ significantly, exceeding this specific range, then these two pixels can be determined not to belong to the same tissue. Therefore, the region growing algorithm selects pixels on the aortic wall of the aortic region as seed points, uses gray value information to identify and distinguish the aortic wall from surrounding tissues, and iteratively expands the identification range to comprehensively and accurately depict the aortic wall.

[0059] Optionally, the localization model includes a machine learning model. Thus, after pre-training, the localization model can be used to identify specific objects in medical imaging data and output the location information of those objects within the medical imaging data. For example, the localization model can identify aortic dissection in the aortic region and output the location information of the aortic dissection lesion.

[0060] In this embodiment, the localization model may include a DenseNet (Densely Connected Convolutional Networks) model. Understandably, the DenseNet model includes convolutional layers and fully connected layers. The convolutional layers are used to extract features, and the fully connected layers are used to output lesion location information. The multiple convolutional layers in the DenseNet model are densely connected, and the input to each convolutional layer is the union of the output feature maps of all preceding convolutional layers. Therefore, the features output from different layers can be effectively fused together, increasing the diversity and richness of features and improving the network's ability to integrate features at different scales and levels, thus helping to more accurately locate the lesion location information of aortic dissection.

[0061] Optionally, the lesion location information includes information used to determine the location of the aortic dissection in the second medical imaging data. This information is then used as key input to the region determination module, enabling it to quickly and accurately identify the first region of interest containing the aortic dissection within the input second medical imaging data.

[0062] Furthermore, the lesion location information includes: the center coordinates of the aortic dissection and the size information of the aortic dissection; the region determination module is specifically used to: determine the center coordinates of the first region of interest based on the center coordinates of the aortic dissection, and determine the boundary of the first region of interest based on the size information of the aortic dissection, so as to determine the first region of interest containing the aortic dissection in the second medical image. Thus, the approximate location of the aortic dissection can be quickly located based on the center coordinates, and the size information ensures that the definition of the region of interest is neither too broad nor overlooks relevant pathological features, more accurately determining the first region of interest in the second medical image. This provides a more targeted working area for the segmentation model, reduces unnecessary calculations and analysis, and helps improve the efficiency and accuracy of segmentation.

[0063] In this embodiment, the size information may include length, width, and height.

[0064] Optionally, the segmentation model includes a machine learning model. Thus, after pre-training, the segmentation model can be used to identify and segment specific anatomical structures or pathological features within a region of interest. For example, the localization model can be used to perform multi-class segmentation of the region of interest to obtain segmentation results corresponding to different pathological features of aortic dissection.

[0065] In this embodiment, the segmentation model may include a SegResNet (Segmentation Residual Network) model. Understandably, the SegResNet model combines an encoder-decoder structure with residual network techniques. The encoder part is used to extract features, the decoder part is used to map the features extracted by the encoder back to the original image size for accurate pixel-level segmentation, and the residual network part assists the encoder in deeper feature extraction and also assists the decoder in feature fusion to enhance the accuracy of the segmentation results.

[0066] Optionally, the segmentation module is also used for at least one of the following: outputting a 2D slice view based on the segmentation results; and constructing a 3D reconstructed view based on the 2D slice views corresponding to multiple first medical image data. Thus, the 2D slice view allows for a direct view of the segmentation results of different sections of the aortic dissection, facilitating the observation of subtle pathological features on the sections, aiding in accurate diagnosis, and making it easier to switch between different slices; the 3D reconstructed view displays the three-dimensional morphology and spatial location of the aortic dissection, facilitating the observation of the overall structure for comprehensive analysis, and can also aid in surgical simulation and planning, improving the user experience.

[0067] Understandably, the segmentation result can be a segmentation mask containing each category; the segmentation module can be specifically used to overlay the segmentation mask with the corresponding tile image data to output a 2D tile view.

[0068] Furthermore, the segmentation module is specifically used to: assign corresponding visual features to each pixel in the segmentation mask according to its category, and overlay the segmentation mask with the corresponding slice image data to output a 2D slice view. Please refer to [reference needed]. Figure 2 In this embodiment, the visual feature is specifically, for example, color.

[0069] Understandably, the pathological features of aortic dissection include: the true lumen, the false lumen, and the tear site. The true lumen refers to the original vascular cavity in aortic dissection, which is usually compressed during the dissection. The false lumen is the extra space formed by the tear in the intima, through which blood flows, leading to separation of the aortic wall. The tear site is the location of the intima tear, the entry point for blood from the true lumen into the false lumen. Aortic dissection segmentation can be understood as the classification and segmentation of the true lumen, false lumen, and tear site. Therefore, by setting visual features corresponding to the categories when outputting the segmentation results, doctors can more intuitively see the segmentation results of aortic dissection.

[0070] Optionally, the aortic dissection segmentation device further includes: a first model training module, used to: use fourth medical image data obtained by preprocessing the third medical image data through a preprocessing module as input to the localization model, and obtain the predicted lesion location information output by the localization model; based on the predicted lesion location information and the actual lesion location information of the third medical image data, a first loss function is used to train the localization model; wherein, the first loss function includes mean square error (MSE) and Dice loss. Thus, by training the localization model through the first model training module, the localization model can more accurately identify the lesion location information, improving the accuracy of segmentation and diagnosis; the mean square error (MSE) measures the square of the difference between the model's lesion location information and the actual lesion location information, helping the model learn to minimize the prediction error; the Dice loss can be used to evaluate and improve the similarity between the predicted lesion location information and the actual lesion location information. Using both loss functions helps improve the localization accuracy of the localization model.

[0071] In this embodiment, the training process of the first model training module may include:

[0072] The fourth medical image data, obtained by preprocessing the third medical image data through the preprocessing module, is used as the dataset, which is divided into a training set and a validation set.

[0073] The training set is used as input to the localization model, and the first predicted lesion location information output by the localization model is obtained.

[0074] The first loss function value is obtained based on the first predicted lesion location information and the actual lesion location information of the training set. Training is stopped when the first loss function value is less than or equal to the first preset loss function value, and the trained localization model is obtained.

[0075] The validation set is used as input to the trained localization model, and the second predicted lesion location information output by the trained localization model is obtained.

[0076] The first loss function value is obtained based on the second predicted lesion location information and the actual lesion location information of the validation set. When the first loss function value is less than or equal to the first preset loss function value, the training of the localization model is completed.

[0077] Therefore, the localization model is trained using a training set, and the localization model is evaluated using a validation set to determine whether it overfits during training, thereby improving the localization accuracy of the localization model.

[0078] In some optional specific examples, the preprocessing module is also used to normalize the actual lesion location information to normalize the lesion coordinates indicated by the actual lesion location information. This facilitates the unified processing of coordinates from different image data, improving training efficiency.

[0079] As an optional specific implementation, the aortic dissection segmentation device further includes: a second model training module, used to: use the second region of interest of the fifth medical image data determined by the region determination module as the input of the segmentation model, and obtain the predicted segmentation result output by the segmentation model; based on the predicted segmentation result and the actual segmentation result of the fifth medical image data, a second loss function is used to train the segmentation model; wherein, the second loss function includes multi-class cross-entropy loss and Dice loss. Thus, by training the segmentation model through the second model training module, the segmentation model can more accurately segment aortic dissection, improving diagnostic accuracy; multi-class cross-entropy loss is suitable for multi-class segmentation tasks, helping the model to more accurately classify the category of each pixel, thereby improving segmentation accuracy; Dice loss can be used to evaluate and improve the similarity between the predicted segmentation result and the actual segmentation result. Using both loss functions helps to improve the segmentation accuracy of the segmentation model.

[0080] As another optional implementation, the aortic dissection segmentation device further includes: a second model training module, used to: acquire a second region of interest (ROI) of the fifth medical image data determined by the region determination module; perform image deformation processing on the second ROI; use the image-deformed second ROI as input to the segmentation model and obtain the predicted segmentation result of the segmentation model; and, based on the predicted segmentation result and the actual segmentation result of the fifth medical image data, employ a second loss function to train the segmentation model; wherein the second loss function includes multi-class cross-entropy loss and Dice loss. Therefore, performing image deformation processing on the second ROI before training helps increase the diversity of training samples and improve the robustness of the model.

[0081] Optionally, image deformation processing includes, but is not limited to: rotation, translation, scaling, and elastic deformation.

[0082] In this embodiment, the training process of the second model training module may include:

[0083] The second region of interest of the fifth medical image data determined by the region determination module is obtained as the dataset, which is divided into a training set and a validation set.

[0084] The training set is used as input to the segmentation model, and the first predicted segmentation result output by the segmentation model is obtained.

[0085] The second loss function value is obtained based on the first predicted segmentation result and the actual segmentation result of the training set. Training is stopped when the second loss function value is less than or equal to the second preset loss function value, and the trained segmentation model is obtained.

[0086] The validation set is used as input to the trained segmentation model, and the second predicted segmentation result output by the trained segmentation model is obtained.

[0087] The second loss function value is obtained based on the second predicted segmentation result and the actual segmentation result of the validation set. When the second loss function value is less than or equal to the second preset loss function value, the training of the segmentation model is completed.

[0088] Therefore, the segmentation model is trained using a training set and evaluated using a validation set to determine whether the model overfits during training, thereby improving the segmentation accuracy of the model.

[0089] This application also provides a computer-readable storage medium storing instructions that, when executed by a processor of an electronic device, enable the electronic device to control the aortic dissection segmentation apparatus of any of the above embodiments.

[0090] Embodiments of this application may be systems, methods, and / or computer program products. A computer program product may include a computer-readable storage medium on which computer-readable program instructions are loaded to cause a processor to implement various aspects of this application. The computer program product may be written in any combination of one or more programming languages ​​to perform operations of embodiments of this application. Programming languages ​​include object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code may execute entirely on a user's computing device, partially on a user's device, as a standalone software package, partially on a user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computers, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuits, such as programmable logic circuits, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), are personalized by utilizing state information of computer-readable program instructions. These electronic circuits can execute computer-readable program instructions to implement various aspects of this application.

[0091] Computer-readable storage media can take the form of any combination of one or more readable media. A readable medium can be a readable signal medium or a readable storage medium. A computer-readable storage medium is a tangible device capable of holding and storing instructions for use by an instruction execution device. A readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combinations thereof. The computer-readable storage medium as used herein is not to be construed as a transient signal itself, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0092] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0093] Various aspects of this application are described herein with reference to block diagrams of an apparatus according to embodiments of this application. It should be understood that each block of the block diagram, and combinations of blocks in the block diagram, can be implemented by computer-readable program instructions.

[0094] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processor of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner; thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the block diagram.

[0095] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more blocks of the block diagram.

[0096] This application also provides an electronic device, such as... Figure 3 As shown, it includes: a processor; a memory for storing computer-executable instructions; and a processor for executing the computer-executable instructions to control the aortic dissection segmentation apparatus of any of the above embodiments.

[0097] A processor can be a central processing unit (CPU) or other form of processing unit with data processing and / or instruction execution capabilities, and can control other components in an electronic device to perform desired functions.

[0098] The memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and a processor may execute the program instructions to implement the steps in the text recognition methods of the various embodiments of this application above, and / or other desired functions.

[0099] In one example, the electronic device may also include input devices and output devices, which are interconnected via a bus system and / or other forms of connection mechanism (not shown in the figure).

[0100] In addition, input devices may include, for example, a keyboard, a mouse, a microphone, etc. Output devices can output various information to the outside, and may include, for example, a monitor, speakers, a printer, and communication networks and their connected remote output devices, etc.

[0101] Of course, for the sake of simplicity, Figure 3 Only a portion of the components of the electronic device relevant to this application are shown in this illustration, omitting components such as buses, input devices / output interfaces, etc. In addition, the electronic device may include any other suitable components depending on the specific application.

[0102] It should be noted that the aortic dissection segmentation device, computer-readable storage medium, and electronic device embodiments provided in this application belong to the same concept; the technical features in the technical solutions described in each embodiment can be arbitrarily combined without conflict.

[0103] It should be understood that the above embodiments are exemplary and are not intended to encompass all possible implementations included in the claims. Various modifications and changes can be made to the above embodiments without departing from the scope of this disclosure. Similarly, the various technical features of the above embodiments can be arbitrarily combined to form other embodiments of the present invention that may not be explicitly described. Therefore, the above embodiments only illustrate several implementations of the present invention and do not limit the scope of protection of this patent.

Claims

1. A device for cutting aortic dissection, characterized in that, include: The preprocessing module is used to preprocess the first medical image data containing aortic information to obtain the preprocessed second medical image data. The positioning module is used to detect the second medical image data using a region detection model, determine the aortic region in the second medical image data, and determine the lesion location information of aortic dissection in the aortic region using the positioning model. The region determination module is used to determine a first region of interest containing the aortic dissection in the second medical image data based on the second medical image data and the lesion location information; The segmentation module uses a segmentation model to perform multi-class segmentation on the first region of interest to obtain segmentation results corresponding to different pathological features of the aortic dissection.

2. The aortic dissection segmentation device according to claim 1, characterized in that, The positioning module is further configured to: after determining the aortic region in the second medical image data, use a region growing algorithm to identify the aortic wall in the aortic region; The positioning module is specifically used for: Based at least on the identified aortic wall, the localization model is used to determine the lesion location information of the aortic dissection in the aortic region.

3. The aortic dissection segmentation device according to claim 1, characterized in that, The lesion location information includes: the center coordinates of the aortic dissection and the size information of the aortic dissection; The region determination module is specifically used for: The center coordinates of the first region of interest are determined based on the center coordinate information of the aortic dissection, and the boundary of the first region of interest is determined based on the size information of the aortic dissection, so as to determine the first region of interest containing the aortic dissection in the second medical image.

4. The aortic dissection segmentation device according to claim 1, characterized in that, The segmentation module is also used for at least one of the following: Output a 2D slice view based on the segmentation results; A 3D reconstructed view is constructed based on the 2D slice views corresponding to multiple primary medical image data.

5. The aortic dissection segmentation device according to claim 1, characterized in that, The device further includes: a first model training module, used for: The fourth medical image data, obtained by preprocessing the third medical image data through the preprocessing module, is used as the input to the localization model, and the predicted lesion location information output by the localization model is obtained. Based on the predicted lesion location information and the actual lesion location information of the third medical image data, a first loss function is used to train the localization model; wherein, the first loss function includes mean squared error (MSE) and Dice loss.

6. The aortic dissection segmentation device according to claim 5, characterized in that, The preprocessing module is further used to: normalize the actual lesion location information to normalize the lesion coordinates indicated by the actual lesion location information.

7. The aortic dissection segmentation device according to claim 1, characterized in that, The device further includes: a second model training module, used for: The second region of interest of the fifth medical image data determined by the region determination module is used as the input of the segmentation model, and the predicted segmentation result output by the segmentation model is obtained. Based on the predicted segmentation result and the actual segmentation result of the fifth medical image data, a second loss function is used to train the segmentation model; wherein, the second loss function includes multi-class cross-entropy loss and Dice loss.

8. The aortic dissection segmentation device according to claim 1, characterized in that, The device further includes: a second model training module, used for: Obtain the second region of interest of the fifth medical image data determined by the region determination module; Image deformation processing is performed on the second region of interest; The second region of interest, after the image deformation processing is completed, is used as the input to the segmentation model, and the predicted segmentation result of the segmentation model is obtained. Based on the predicted segmentation result and the actual segmentation result of the fifth medical image data, a second loss function is used to train the segmentation model; wherein, the second loss function includes multi-class cross-entropy loss and Dice loss.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed by a processor of an electronic device, enable the electronic device to control the aortic dissection segmentation apparatus according to any one of claims 1 to 8.

10. An electronic device, characterized in that, The electronic device includes: processor; Memory used to store executable instructions for a computer; The processor is configured to execute the computer-executable instructions to control the aortic dissection segmentation device according to any one of claims 1 to 8.

Images