Carotid artery segmentation method, device, and equipment based on magnetic resonance imaging

Abstract

This invention provides a method, apparatus, and device for carotid artery segmentation based on magnetic resonance imaging (MRI). The method includes: acquiring three-dimensional magnetic resonance image data of the carotid artery of a target patient; dividing the three-dimensional magnetic resonance image data into a first dataset and a second dataset along the vascular axis; inputting the three-dimensional magnetic resonance image data into a pre-trained three-dimensional segmentation model to segment the carotid artery bifurcation region from the three-dimensional magnetic resonance image data; inputting the first dataset and the second dataset into a pre-trained two-dimensional segmentation model to segment the vascular region from each two-dimensional image data; and correcting the vascular region obtained from each two-dimensional image data based on the carotid artery bifurcation region obtained from the three-dimensional magnetic resonance image data. This achieves fully automatic segmentation of the carotid artery, which is time-efficient, highly effective, and highly repeatable. Furthermore, by combining three-dimensional and two-dimensional segmentation, interference from irrelevant regions on the segmentation results is removed, improving the accuracy of carotid artery segmentation.

Description

Technical Field

[0001] This invention relates to the field of image segmentation technology, specifically to a method, apparatus, and device for carotid artery segmentation based on magnetic resonance images. Background Technology

[0002] Carotid atherosclerosis is a common clinical disease caused by the accumulation of cholesterol, fat, calcium, and other substances on the walls of the carotid arteries. This buildup, commonly known as plaque, can block the carotid artery, causing localized changes in blood vessel size and reducing blood supply. If a plaque ruptures suddenly, it can form a blood clot, which can lead to stroke. Therefore, examining the carotid arteries is crucial for disease prevention and diagnosis. Magnetic resonance imaging (MRI) is a highly integrated instrument combining physical and chemical technologies. It can provide non-invasive magnetic resonance images reflecting the anatomical structure and pathological information of target tissues and has become one of the main methods for examining the carotid arteries in clinical practice.

[0003] After obtaining MRI images of the carotid artery, the region of interest (ROI) needs to be segmented for clinical analysis. Currently, this segmentation primarily relies on manual methods. Specifically, the 3D images acquired by MRI are sliced ​​along the vessel axis, and the carotid artery wall and lumen are manually identified within the 2D slices, then delineated based on this manual operation. Since each patient requires numerous 2D slices for segmentation, the entire process is very time-consuming, and human factors still play a dominant role, resulting in low reproducibility of the segmentation results. Therefore, a more efficient carotid artery segmentation method is urgently needed. Summary of the Invention

[0004] This invention provides a method, apparatus, and device for carotid artery segmentation based on magnetic resonance images, which solves the problems of long time consumption, low efficiency, and low repeatability of existing manual segmentation.

[0005] In a first aspect, embodiments of the present invention provide a carotid artery segmentation method based on magnetic resonance imaging, comprising:

[0006] Acquire 3D magnetic resonance imaging data of the carotid artery of the target patient;

[0007] The magnetic resonance three-dimensional image data were divided into a first dataset and a second dataset along the vascular axis. The first dataset included multiple two-dimensional image data of the left carotid artery of the target patient, and the second dataset included multiple two-dimensional image data of the right carotid artery of the target patient.

[0008] The 3D magnetic resonance imaging data is input into a pre-trained 3D segmentation model to segment the carotid bifurcation region from the 3D magnetic resonance imaging data.

[0009] The first and second datasets are input into the pre-trained two-dimensional segmentation model to segment the blood vessel region from each two-dimensional image data.

[0010] Based on the carotid bifurcation region obtained from the 3D magnetic resonance imaging data, the vascular region obtained from each 2D image data is corrected to obtain the region of interest in the carotid artery of the target patient.

[0011] In one embodiment, the vascular region obtained from each two-dimensional image data is corrected based on the carotid bifurcation region obtained from the magnetic resonance three-dimensional image data, including:

[0012] The blood vessel regions falling within the carotid bifurcation region are retained in each 2D image data.

[0013] In one embodiment, before training the 3D segmentation model and the 2D segmentation model, the method further includes:

[0014] Linear interpolation is performed on partially labeled training samples to generate fully labeled training samples.

[0015] In one embodiment, before correcting the vascular region obtained from each two-dimensional image data based on the carotid bifurcation region obtained from the magnetic resonance three-dimensional image data, the method further includes:

[0016] Morphological corrections are performed on the segmented blood vessel regions in each 2D image data. These corrections include erosion, connected component detection, preservation of the largest connected component, and dilation.

[0017] In one embodiment, the method further includes:

[0018] Calculate the distance between the center of the lumen and the center of the pipe wall;

[0019] If the distance is greater than a preset threshold, then perform morphological correction operations iteratively until the distance is less than the preset threshold, or until the preset number of iterations is reached.

[0020] In one embodiment, acquiring three-dimensional magnetic resonance imaging data of the carotid artery of a target patient includes:

[0021] Real-time three-dimensional magnetic resonance imaging data of the target patient's carotid artery is acquired using magnetic resonance imaging equipment;

[0022] or,

[0023] Retrieve pre-stored 3D magnetic resonance imaging data of the target patient's carotid artery from a storage device.

[0024] In one embodiment, the method further includes: visualizing the segmented region of interest of the carotid artery based on the magnetic resonance three-dimensional image data.

[0025] Secondly, embodiments of the present invention provide a carotid artery segmentation device based on magnetic resonance imaging, comprising:

[0026] The acquisition module is used to acquire magnetic resonance three-dimensional image data of the carotid artery of the target patient;

[0027] The partitioning module is used to divide the magnetic resonance three-dimensional image data into a first dataset and a second dataset along the vascular axis. The first dataset includes multiple two-dimensional image data of the left carotid artery of the target patient, and the second dataset includes multiple two-dimensional image data of the right carotid artery of the target patient.

[0028] The 3D segmentation module is used to input magnetic resonance 3D image data into a pre-trained 3D segmentation model to segment the carotid bifurcation region from the magnetic resonance 3D image data.

[0029] The two-dimensional segmentation module is used to input the first dataset and the second dataset into the pre-trained two-dimensional segmentation model, and segment the blood vessel region from each two-dimensional image data.

[0030] The correction module is used to correct the vascular region obtained in each two-dimensional image data based on the carotid bifurcation region obtained in the magnetic resonance three-dimensional image data, so as to obtain the region of interest of the carotid artery of the target patient.

[0031] Thirdly, embodiments of the present invention provide an electronic device, comprising:

[0032] At least one processor and memory;

[0033] The memory stores the instructions that the computer executes;

[0034] At least one processor executes computer execution instructions stored in memory, causing the at least one processor to perform the carotid artery segmentation method based on magnetic resonance images as described in any of the first aspects.

[0035] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the carotid artery segmentation method based on magnetic resonance images as described in any of the first aspects.

[0036] The carotid artery segmentation method, apparatus, and device based on magnetic resonance imaging (MRI) provided in this invention acquire 3D MRI image data of the carotid artery of a target patient; divide the 3D MRI image data into a first dataset and a second dataset along the vascular axis. The first dataset includes multiple 2D images of the left carotid artery of the target patient, and the second dataset includes multiple 2D images of the right carotid artery of the target patient. The 3D MRI image data is input into a pre-trained 3D segmentation model to segment the carotid bifurcation region from the 3D MRI image data; the first dataset and the second dataset are respectively input into a pre-trained 2D segmentation model to segment the vascular region from each 2D image data; the vascular region obtained from each 2D image data is corrected based on the carotid bifurcation region obtained from the 3D MRI image data to obtain the region of interest (ROI) of the carotid artery of the target patient, achieving fully automatic segmentation of the carotid artery with short processing time, high efficiency, and high repeatability. Furthermore, by coordinating 3D and 2D segmentation, interference from irrelevant regions on the segmentation results is removed, improving the accuracy of carotid artery segmentation. Attached Figure Description

[0037] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0038] Figure 1 A flowchart illustrating a carotid artery segmentation method based on magnetic resonance images according to an embodiment of the present invention;

[0039] Figure 2 This is a schematic diagram of a visual display interface provided in an embodiment of the present invention;

[0040] Figure 3 A schematic diagram of a carotid artery segmentation device based on magnetic resonance imaging provided in an embodiment of the present invention;

[0041] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0042] The accompanying drawings have illustrated specific embodiments of the invention, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the invention in any way, but rather to illustrate the concept of the invention to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0043] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings. Similar elements in different embodiments are referred to by associated similar element reference numerals. In the following embodiments, many details are described to facilitate a better understanding of this application. However, those skilled in the art will readily recognize that some features may be omitted in different situations, or may be replaced by other elements, materials, or methods. In some cases, certain operations related to this application are not shown or described in the specification. This is to avoid obscuring the core parts of this application with excessive description. For those skilled in the art, detailed description of these related operations is not necessary; they can fully understand the related operations based on the description in the specification and general technical knowledge in the art.

[0044] Furthermore, the features, operations, or characteristics described in the specification can be combined in any suitable manner to form various embodiments. At the same time, the steps or actions in the method description can be rearranged or adjusted in a manner obvious to those skilled in the art. Therefore, the various orders in the specification and drawings are only for the clear description of a particular embodiment and do not imply a necessary order, unless otherwise stated that a particular order must be followed.

[0045] The serial numbers assigned to components in this document, such as "first" and "second," are used only to distinguish the described objects and have no sequential or technical meaning. The terms "connection" and "linkage" used in this application, unless otherwise specified, include both direct and indirect connections (linkages).

[0046] The carotid bifurcation region is a common site of atherosclerosis. Compared with other traditional medical segmentation tasks, carotid artery segmentation presents the following unique challenges: First, the shape of the lumen and vessel contour is generally a concentric circle, a smooth circle, or a closed circle, meaning that all points on the contour are close to the centroid of the vessel. However, the lumen shape of vessels with atherosclerosis is diverse, including elliptical, eccentric small circles, crescent shapes, and even those invisible on MRI images (i.e., occlusion). The lumen shape mainly depends on the distribution of plaques. Second, considering the smoothness and tubular nature of the carotid artery, the lumen and vessel wall should have a sufficiently smooth shape without sharp areas. Therefore, in carotid artery wall segmentation tasks, the shapes of the lumen and vessel wall can vary greatly, mainly due to atherosclerosis, i.e., the presence of plaques. Third, the focus of interpretation is concentrated on the area 4 cm above and below the carotid bifurcation. How to accurately segment the region of interest and reduce interference from other vessels is also a key focus of carotid artery segmentation.

[0047] Traditional segmentation algorithms, such as thresholding, are only suitable for images where the target and background occupy different grayscale ranges, and are not applicable to high-resolution carotid artery MRI images. When segmenting based on edge detection, wavelet transform, etc., further processing or combinations with other related algorithms are required to complete the segmentation task. While existing deep learning neural network segmentation algorithms can consider multi-scale feature information from different sub-regions and perform automatic and rapid pixel-level segmentation, achieving significant results in image segmentation, current deep learning models have not been further optimized for carotid artery segmentation. They do not consider the special properties of the vessel wall and the clinically relevant areas, resulting in poor segmentation performance and a tendency to produce incorrect segmentations in similar vascular tissues.

[0048] To address at least one of the problems existing in the prior art, this application proposes a carotid artery segmentation method based on a combination of three-dimensional and two-dimensional segmentation. This method can remove interference from irrelevant region segmentation results and accurately segment the region of interest of clinical interest. Detailed descriptions are provided below through specific embodiments.

[0049] Figure 1 This is a flowchart illustrating a carotid artery segmentation method based on magnetic resonance imaging, as provided in an embodiment of the present invention. Figure 1 As shown, the carotid artery segmentation method based on magnetic resonance images provided in this embodiment may include:

[0050] S101. Obtain magnetic resonance three-dimensional image data of the target patient's carotid artery.

[0051] In this embodiment, the acquisition of the target patient's carotid artery 3D magnetic resonance image data can be achieved by acquiring the target patient's carotid artery 3D magnetic resonance image data in real time through a magnetic resonance imaging device; or it can be obtained from a storage device by acquiring the target patient's carotid artery 3D magnetic resonance image data that has been stored in advance.

[0052] S102. Divide the magnetic resonance three-dimensional image data into a first dataset and a second dataset along the vascular axis. The first dataset includes multiple two-dimensional image data of the left carotid artery of the target patient, and the second dataset includes multiple two-dimensional image data of the right carotid artery of the target patient.

[0053] To improve segmentation accuracy, this embodiment segments the left and right carotid arteries separately. Carotid artery data obtained from magnetic resonance imaging (MRI) scans is typically exported from the machine in Digital Imaging and Communications in Medicine (DICOM) format. In this embodiment, the 3D MRI image data can be halved according to the left-right direction defined in the DICOM metadata, dividing it into a first dataset and a second dataset. The first dataset corresponds to the left carotid artery of the target patient; performing two-dimensional slicing along the axial direction generates multiple two-dimensional image data of the target patient's left carotid artery. The second dataset corresponds to the right carotid artery of the target patient; performing two-dimensional slicing along the axial direction generates multiple two-dimensional image data of the target patient's right carotid artery.

[0054] S103. Input the magnetic resonance three-dimensional image data into the pre-trained three-dimensional segmentation model to segment the carotid bifurcation region from the magnetic resonance three-dimensional image data.

[0055] In this embodiment, the model used for 3D segmentation can be the nnU-net model. Through key representation analysis of the training dataset, various hyperparameters are adjusted without any manual intervention. The optimal model is then ensembled using 5x cross-validation, ultimately achieving automated training of the optimal 3D segmentation model. Specifically, a 3D cascaded nnU-net can be used for 3D blood vessel wall segmentation.

[0056] It should be noted that the key area of ​​focus in clinical practice is the region approximately 4 cm above and below the carotid bifurcation, i.e., the carotid bifurcation region. Therefore, in order to remove interference from other irrelevant areas for carotid artery segmentation, this embodiment can use 3D training samples labeled with the vessel wall mask of the carotid bifurcation region to train the 3D segmentation model.

[0057] S104. Input the first dataset and the second dataset into the pre-trained two-dimensional segmentation model respectively, and segment the blood vessel region from each two-dimensional image data.

[0058] In this embodiment, the model used for 2D segmentation can be the nnU-net model. Through key representation analysis of the training dataset, various hyperparameters are adjusted without any manual intervention. The optimal model is then ensembled using 5x cross-validation, ultimately achieving automated training of the optimal 2D segmentation model. Specifically, a 2D cascaded nnU-net can be used for 2D blood vessel wall segmentation.

[0059] In this embodiment, the left and right carotid arteries are processed separately, so each 2D image only produces a segmentation result for one blood vessel. In this embodiment, the 2D segmentation model can be trained using 2D training samples labeled with blood vessel wall masks. These 2D training samples can be generated by slicing 3D training samples labeled with blood vessel wall masks along the axial direction.

[0060] S105. Based on the carotid bifurcation region obtained from the magnetic resonance three-dimensional image data, the vascular region obtained from each two-dimensional image data is corrected to obtain the region of interest of the carotid artery of the target patient.

[0061] In one optional implementation, the vascular region obtained from each two-dimensional image is corrected based on the carotid bifurcation region obtained from the 3D magnetic resonance imaging data. Specifically, this may include retaining the vascular regions in each two-dimensional image that fall within the carotid bifurcation region. That is, the intersection of the carotid bifurcation region obtained from the 3D magnetic resonance imaging data and the vascular regions obtained from each two-dimensional image is taken as the final segmentation result. In other words, the 2D segmentation result is retained within the range of the 3D segmentation result, because only the vascular regions falling within the carotid bifurcation region are the areas of clinical interest. Correcting the vascular regions obtained from each two-dimensional image based on the carotid bifurcation region obtained from the 3D magnetic resonance imaging data can further improve the accuracy of segmentation.

[0062] The carotid artery segmentation method based on magnetic resonance imaging (MRI) provided in this embodiment acquires 3D MRI image data of the target patient's carotid artery. The 3D MRI image data is divided into a first dataset and a second dataset along the vessel axis. The first dataset includes multiple 2D images of the target patient's left carotid artery, and the second dataset includes multiple 2D images of the target patient's right carotid artery. The 3D MRI image data is input into a pre-trained 3D segmentation model to segment the carotid bifurcation region from the 3D MRI image data. The first and second datasets are then input into a pre-trained 2D segmentation model to segment the vessel region from each 2D image. Based on the carotid bifurcation region obtained from the 3D MRI image data, the vessel regions obtained from each 2D image data are corrected to obtain the region of interest (ROI) of the target patient's carotid artery. This achieves fully automatic carotid artery segmentation with low time consumption, high efficiency, and high repeatability. Furthermore, by coordinating 3D and 2D segmentation, interference from irrelevant regions on the segmentation results is removed, improving the accuracy of carotid artery segmentation.

[0063] It should be noted that training the 3D and 2D segmentation models requires a large number of labeled training samples. However, manual annotation is not only time-consuming but also extremely costly. Therefore, the actual training samples usually only depict the vessel wall contour coordinates in a limited axial 2D slice. Directly using these partially labeled training samples to train the model will reduce the segmentation accuracy; while manually annotating all training samples is prohibitively time-consuming and economically costly. To balance cost and segmentation accuracy, based on the above embodiments, the carotid artery segmentation method based on magnetic resonance images provided in this embodiment first uses linear interpolation to generate fully labeled training samples from the partially labeled training samples before training the 3D and 2D segmentation models. Considering that the manually drawn vessel wall contour coordinates are only drawn in a limited axial 2D slice, it is necessary to use linear interpolation on the unlabeled slices to generate a complete but coarse 3D vessel wall mask. Carotid artery data obtained from an MRI scan is exported from the machine in DICOM format for training a 3D segmentation network. The image is roughly divided into left and right carotid artery segments according to the left-right direction defined in the DICOM metadata, for training a 2D segmentation network. After obtaining the trained 2D and 3D segmentation models, in practical applications, only DICOM format images of the patient's carotid arteries need to be acquired and exported, and then input into the trained 2D and 3D segmentation models respectively for 2D and 3D segmentation.

[0064] To further improve segmentation accuracy, based on any of the above embodiments, the carotid artery segmentation method based on magnetic resonance images provided in this embodiment performs morphological correction on the segmented blood vessel regions in each two-dimensional image data before correcting the blood vessel regions obtained from each two-dimensional image data according to the carotid bifurcation region obtained from the magnetic resonance three-dimensional image data. This is to remove discontinuous blood vessels and thus avoid the adverse effects of similar blood vessel tissue on segmentation. Specifically, morphological correction includes erosion, connected component detection, preservation of the largest connected component, and dilation operation, retaining the result with the largest area. For the two-dimensional segmentation model, the left and right carotid arteries are processed separately, so each two-dimensional image only has the segmentation result of one blood vessel. During morphological correction, the largest connected component is retained to correct discontinuous and erroneous blood vessel segmentation results. This morphological optimization operation helps correct erroneous bifurcation results at the carotid bifurcation.

[0065] Simultaneously, centroid analysis is performed on the segmentation results, meaning the distance between the lumen center and the vessel wall center should be less than a certain threshold to keep the lumen within the vessel wall. Based on the above embodiments, the carotid artery segmentation method based on magnetic resonance imaging provided in this embodiment may further include: calculating the distance between the lumen center and the vessel wall center; if the distance is greater than a preset threshold, iteratively performing morphological correction operations until the distance is less than the preset threshold, or until the number of iterations reaches a preset number.

[0066] Understandably, since the dataset used for training the 3D segmentation network is generated from discontinuous 2D slices, there is some error between the generated data labels and the true labels. Consequently, the segmentation effect of the 3D network is slightly worse than that of the 2D network. However, due to the presence of 3D spatial information, it has a better effect on locating the area before and after the carotid bifurcation, which is of clinical interest. The 2D segmentation results are more refined, but since the network input is slices from each layer, inaccurate segmentation is also performed in areas that are not of clinical interest. These results should be ignored to improve the final segmentation effect. Therefore, when summarizing the segmentation results, this application uses the 2D morphologically corrected segmentation as the final segmentation. When a sudden decrease in the vessel area is detected in the 3D segmentation result, it is considered that the vessel has left the critical region of interest, and the segmentation results of other distal slices are removed, while the segmentation results of the clinically relevant areas are retained. Finally, the 2D segmentation result corrected based on the 3D results is output as the final result.

[0067] Based on any of the above embodiments, to enable users to conveniently and intuitively view the segmentation results of the carotid artery, the carotid artery segmentation method based on magnetic resonance images provided in this embodiment may further include: visualizing the segmented region of interest of the carotid artery based on the magnetic resonance three-dimensional image data. Please refer to... Figure 2 , Figure 2 This is a schematic diagram of a visual display interface provided in an embodiment of the present invention. Figure 2 As shown in the image, the white area represents the blood vessel wall, and the area enclosed by the white area is the blood vessel lumen. From... Figure 2 The left-hand side shows the area of ​​clinical interest, approximately 4 cm above and below the carotid artery bifurcation. This can also be visualized as follows: Figure 2 The two images on the right show the left and right carotid arteries respectively.

[0068] In summary, the carotid artery segmentation method based on magnetic resonance imaging provided in this application, by combining a 3D segmentation network and a 2D segmentation network, can accurately locate the carotid bifurcation region of clinical interest and remove interference from irrelevant segmentation results. This is because the 2D segmentation model helps improve the accuracy of the segmentation results, while the 3D segmentation model helps provide localization of the clinically relevant region; the combination of the two achieves better segmentation results. Morphological correction operations such as maximum connected component analysis, erosion, and dilation are applied to the 2D segmentation results to help remove erroneous segmentation results, achieving better segmentation performance. This application solves the problem of long segmentation and annotation time for existing carotid artery wall and lumen segmentation methods. Compared with the currently used manual annotation methods, the automatic annotation method using linear interpolation saves more manpower and time. Compared with manual segmentation algorithms, this application has higher repeatability and more stable segmentation results. Compared with other segmentation methods based on a single neural network, this application uses a 2D and 3D collaborative segmentation neural network to perform 2D segmentation of the left and right carotid arteries separately, and fully considers the carotid artery characteristics of clinical interest, improving segmentation accuracy. Compared with traditional algorithms, it has faster processing speed and more stable performance.

[0069] Figure 3 This is a schematic diagram of a carotid artery segmentation device based on magnetic resonance imaging, provided in an embodiment of the present invention. Figure 3 As shown, the carotid artery segmentation device 30 based on magnetic resonance images provided in this embodiment may include: an acquisition module 301, a segmentation module 302, a three-dimensional segmentation module 303, a two-dimensional segmentation module 304, and a correction module 305.

[0070] The acquisition module 301 is used to acquire magnetic resonance three-dimensional image data of the carotid artery of the target patient;

[0071] The partitioning module 302 is used to partition the magnetic resonance three-dimensional image data into a first dataset and a second dataset along the vascular axis. The first dataset includes multiple two-dimensional image data of the left carotid artery of the target patient, and the second dataset includes multiple two-dimensional image data of the right carotid artery of the target patient.

[0072] The three-dimensional segmentation module 303 is used to input magnetic resonance three-dimensional image data into a pre-trained three-dimensional segmentation model to segment the carotid bifurcation region from the magnetic resonance three-dimensional image data.

[0073] The two-dimensional segmentation module 304 is used to input the first dataset and the second dataset into the pre-trained two-dimensional segmentation model respectively, and segment the blood vessel region from each two-dimensional image data.

[0074] The correction module 305 is used to correct the vascular region obtained in each two-dimensional image data based on the carotid bifurcation region obtained in the magnetic resonance three-dimensional image data, so as to obtain the region of interest of the carotid artery of the target patient.

[0075] The apparatus of this embodiment can be used to perform Figure 1 The technical solutions of the method embodiments shown are similar in principle and in effect, and will not be described again here.

[0076] This invention also provides an electronic device, please refer to [link to relevant documentation]. Figure 4 As shown, the embodiments of the present invention are only used as examples. Figure 4 The examples are provided for illustration only and do not imply that the invention is limited to these examples. Figure 4 This is a schematic diagram of the structure of an electronic device provided according to an embodiment of the present invention. Figure 4 As shown, the electronic device 40 provided in this embodiment may include: a memory 401, a processor 402, and a bus 403. The bus 403 is used to connect the various components.

[0077] The memory 401 stores a computer program, which, when executed by the processor 402, can implement the technical solutions of any of the above method embodiments.

[0078] The memory 401 and processor 402 are electrically connected directly or indirectly to enable data transmission or interaction. For example, these components can be electrically connected to each other via one or more communication buses or signal lines, such as bus 403. The memory 401 stores a computer program that implements a carotid artery segmentation method based on magnetic resonance imaging, including at least one software functional module that can be stored in the memory 401 in the form of software or firmware. The processor 402 executes various functional applications and data processing by running the software program and modules stored in the memory 401.

[0079] The memory 401 may be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), etc. The memory 401 stores programs, and the processor 402 executes the programs after receiving execution instructions. Furthermore, the software programs and modules within the memory 401 may also include an operating system, which may include various software components and / or drivers for managing system tasks (such as memory management, storage device control, power management, etc.) and can communicate with various hardware or software components to provide an operating environment for other software components.

[0080] Processor 402 can be an integrated circuit chip with signal processing capabilities. The aforementioned processor 402 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor. It is understood that... Figure 4 The structure shown is for illustrative purposes only and may include more... Figure 4 The more or fewer components shown, or having the same Figure 4 The different configurations shown. Figure 4 The components shown can be implemented in hardware and / or software.

[0081] This invention also provides a computer-readable storage medium storing a computer program thereon, which is executed by a processor to implement the technical solutions of any of the above method embodiments.

[0082] The various embodiments in this disclosure are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0083] The scope of protection of this disclosure is not limited to the embodiments described above. Obviously, those skilled in the art can make various modifications and variations to this disclosure without departing from its scope and spirit. If such modifications and variations fall within the scope of the claims of this disclosure and their equivalents, then the intent of this disclosure also includes such modifications and variations.

Claims

1. A method for carotid artery segmentation based on magnetic resonance imaging, characterized in that, include: Acquire 3D magnetic resonance imaging data of the carotid artery of the target patient; The magnetic resonance three-dimensional image data is divided into a first dataset and a second dataset along the vascular axis. The first dataset includes multiple two-dimensional image data of the left carotid artery of the target patient, generated by two-dimensional slicing along the axial direction corresponding to the left carotid artery of the target patient. The second dataset includes multiple two-dimensional image data of the right carotid artery of the target patient, generated by two-dimensional slicing along the axial direction corresponding to the right carotid artery of the target patient. The magnetic resonance three-dimensional image data is input into a pre-trained three-dimensional segmentation model to segment the carotid artery bifurcation region from the magnetic resonance three-dimensional image data; The first dataset and the second dataset are respectively input into the pre-trained two-dimensional segmentation model to segment the blood vessel region from each two-dimensional image data; Based on the carotid bifurcation region obtained from the magnetic resonance three-dimensional image data, the vascular region obtained from each two-dimensional image data is corrected to obtain the region of interest of the carotid artery of the target patient.

2. The method according to claim 1, characterized in that, The step of correcting the vascular region obtained from each two-dimensional image data based on the carotid bifurcation region obtained from the magnetic resonance three-dimensional image data includes: The blood vessel regions falling within the carotid bifurcation region are retained in each two-dimensional image data.

3. The method according to claim 1, characterized in that, Before training the 3D segmentation model and the 2D segmentation model, the method further includes: Linear interpolation is performed on partially labeled training samples to generate fully labeled training samples.

4. The method according to claim 1, characterized in that, Before correcting the vascular region obtained from each two-dimensional image data based on the carotid bifurcation region obtained from the magnetic resonance three-dimensional image data, the method further includes: Morphological corrections are performed on the segmented blood vessel regions in each two-dimensional image data. These morphological corrections include erosion, connected component detection, preservation of the largest connected component, and dilation operations.

5. The method according to claim 4, characterized in that, The method further includes: Calculate the distance between the center of the lumen and the center of the pipe wall; If the distance is greater than a preset threshold, the morphological correction operation is performed iteratively until the distance is less than the preset threshold, or the number of iterations reaches a preset number.

6. The method according to claim 1, characterized in that, The acquisition of the target patient's carotid artery 3D magnetic resonance imaging data includes: Real-time three-dimensional magnetic resonance imaging data of the target patient's carotid artery is acquired using magnetic resonance imaging equipment; or, Retrieve pre-stored 3D magnetic resonance imaging data of the target patient's carotid artery from a storage device.

7. The method according to any one of claims 1-6, characterized in that, The method further includes: The segmented region of interest of the carotid artery is visualized based on the magnetic resonance three-dimensional image data.

8. A carotid artery segmentation device based on magnetic resonance imaging, characterized in that, include: The acquisition module is used to acquire magnetic resonance three-dimensional image data of the carotid artery of the target patient; The segmentation module is used to divide the magnetic resonance three-dimensional image data into a first dataset and a second dataset along the vascular axis. The first dataset includes multiple two-dimensional image data of the left carotid artery of the target patient, generated by performing two-dimensional slicing along the axial direction corresponding to the left carotid artery of the target patient. The second dataset includes multiple two-dimensional image data of the right carotid artery of the target patient, generated by performing two-dimensional slicing along the axial direction corresponding to the right carotid artery of the target patient. The three-dimensional segmentation module is used to input the magnetic resonance three-dimensional image data into a pre-trained three-dimensional segmentation model and segment the carotid bifurcation region from the magnetic resonance three-dimensional image data. The two-dimensional segmentation module is used to input the first dataset and the second dataset into the pre-trained two-dimensional segmentation model, respectively, to segment the blood vessel region from each two-dimensional image data; The correction module is used to correct the vascular region obtained in each two-dimensional image data according to the carotid bifurcation region obtained in the magnetic resonance three-dimensional image data, so as to obtain the region of interest of the carotid artery of the target patient.

9. An electronic device, characterized in that, include: At least one processor and memory; The memory stores computer-executed instructions; The at least one processor executes computer execution instructions stored in the memory, causing the at least one processor to perform the carotid artery segmentation method based on magnetic resonance images as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the carotid artery segmentation method based on magnetic resonance images as described in any one of claims 1-7.

Images