Method and device for reconstructing personalized heart model with infarction area and medium

Abstract

The invention relates to the technical field of medical images, and provides a method and equipment for reconstructing a personalized heart model with an infarction region, and a medium. The method comprises the following steps: preprocessing a heart magnetic resonance delay gadolinium enhanced CMR-LGE image of a patient to obtain a first high-resolution tag image containing a ventricular cavity and infarction tissue, and constructing a heart finite element mesh model based on high-precision heart cavity data of the patient; performing direction adjustment on the first high-resolution label image to obtain a second high-resolution label image; and mapping the infarction region in the second high-resolution label image into the heart finite element mesh model to obtain a personalized heart model with the infarction region. Based on the scheme of the invention, the accuracy of the reconstruction of the personalized heart model with the infarction region can be improved.

Description

Technical Field

[0001] This invention relates to the field of medical imaging technology, and in particular to a method, device and medium for reconstructing a personalized cardiac model with an infarcted region. Background Technology

[0002] Personalized cardiac models with infarct regions play a crucial role in electrophysiological simulation, risk stratification, and treatment optimization for cardiac diseases. LGE-CMR imaging technology provides the data foundation for these models. However, in practice, cardiac models constructed based on clinical imaging data often fail to perfectly replicate the geometry and spatial location of the infarct region compared to real anatomical structures, posing a risk of accuracy loss. This deviation in model accuracy directly impacts the reliability of subsequent electrophysiological simulation results, potentially weakening their effectiveness in guiding clinical decision-making and research. Therefore, improving the accuracy of personalized cardiac model reconstruction with infarct regions has become an urgent technical challenge. Summary of the Invention

[0003] This application provides a method, device, and medium for reconstructing a personalized heart model with an infarct region, in order to solve the technical problem of how to improve the accuracy of reconstructing a personalized heart model with an infarct region.

[0004] In a first aspect, embodiments of this application provide a method for reconstructing a personalized cardiac model with an infarct region, comprising: Preprocessing the patient's CMR-LGE images yields a first high-resolution labeled image containing ventricular chambers and infarcted tissue, and a cardiac finite element mesh model is constructed based on the patient's high-precision cardiac chamber data; The orientation of the first high-resolution label image is adjusted to obtain the second high-resolution label image; The infarct region in the second high-resolution labeled image is mapped onto the finite element mesh model of the heart to obtain a personalized heart model with the infarct region.

[0005] In conjunction with the first aspect, in some possible implementations, the patient's CMR-LGE images are preprocessed to obtain a first high-resolution labeled image containing ventricular chambers and infarct tissue, including: The patient's CMR-LGE images were segmented to obtain ventricular myocardial images and infarct tissue images. The infarct tissue images included completely infarcted tissue regions and hemi-infarcted tissue regions. Interpolation and reconstruction were performed on ventricular myocardial images to obtain high-resolution labeled images of ventricular myocardium; Interpolation and reconstruction are performed on infarct tissue images to obtain high-resolution labeled images of infarct tissue; The high-resolution labeled image of the ventricular myocardium is merged with the high-resolution labeled image of the infarct tissue to generate a first high-resolution labeled image containing the ventricular cavity and the infarct tissue.

[0006] Combining the first aspect and the above implementation methods, in some possible implementation methods, a cardiac finite element mesh model is constructed based on the patient's high-precision cardiac cavity data, including: The contour of the heart model was determined based on high-precision cardiac cavity data; The heart model outline is meshed into surfaces, the mesh quality is optimized, and the volume is meshed to generate a heart finite element mesh model. The heart finite element mesh model is saved as a node coordinate file that records the node coordinates and a mesh topology file that records the vertex indices of the tetrahedral elements.

[0007] Combining the first aspect and the above implementation methods, in some possible implementation methods, the orientation of the first high-resolution label image is adjusted to obtain a second high-resolution label image, including: The direction of the apex of the heart pointing towards the base of the heart is identified in the first high-resolution label image. The orientation of the first high-resolution label image is then adjusted so that the direction of the apex pointing towards the base of the heart is changed to the positive z-axis direction. The left ventricle in the first high-resolution label image is identified as pointing towards the right ventricle. The orientation of the first high-resolution label image is then adjusted so that the left ventricle pointing towards the right ventricle points towards the negative x-axis. The direction in which the rear wall points to the front wall is identified in the first high-resolution label image. The orientation of the first high-resolution label image is adjusted so that the direction in which the rear wall points to the front wall is changed to the negative y-axis direction. Based on the first high-resolution label image after orientation adjustment, a second high-resolution label image is generated that is consistent with the coordinate system of the heart finite element mesh model.

[0008] Combining the first aspect and the above implementation methods, in some possible implementation methods, identifying the direction from the apex of the heart to the base of the heart in the first high-resolution label image, and adjusting the orientation of the first high-resolution label image so that the direction from the apex of the heart to the base of the heart is changed to point in the positive z-axis direction, including: Each two-dimensional slice along the z-axis in the first high-resolution label image is defined as a layer, with the layer with the largest z-coordinate value being the top layer and the layer with the smallest z-coordinate value being the bottom layer. The number of myocardial pixels contained in the top and bottom layers are counted separately. The layer with more myocardial pixels is identified as the bottom layer of the heart, and the layer with fewer myocardial pixels is identified as the apical layer of the heart. Compare the z-coordinate values ​​of the basal and apical layers of the heart to determine the direction from the apex to the basal layer; If the direction from the apex of the heart to the base of the heart indicates that the apical layer is above the base of the heart, then the first high-resolution label image is flipped along the z-axis so that the direction from the apex of the heart to the base of the heart is changed to the positive direction of the z-axis.

[0009] Combining the first aspect and the above implementation methods, in some possible implementation methods, identifying the left ventricle pointing towards the right ventricle in the first high-resolution label image, and adjusting the orientation of the first high-resolution label image so that the left ventricle pointing towards the right ventricle points towards the negative x-axis direction, includes: Based on connectivity rules, the non-myocardial regions in the first high-resolution labeled image are divided into the left ventricular cavity, the right ventricular cavity, and the image background; among them, the circularity of the left ventricular cavity at the cardiac base is closer to 1 than that of the right ventricular cavity at the cardiac base. Determine the vector R between the centroid of the left ventricle and the centroid of the right ventricle; Compare the angle between vector R and the negative x-axis to determine the direction in which the left ventricle points towards the right ventricle; If the direction from the left ventricle to the right ventricle is not pointing in the negative x-axis direction, then the first high-resolution label image is rotated and / or flipped so that the direction from the left ventricle to the right ventricle is changed to point in the negative x-axis direction.

[0010] Combining the first aspect and the above implementation methods, in some possible implementation methods, identifying the direction in which the rear wall points to the front wall in the first high-resolution label image, and adjusting the orientation of the first high-resolution label image so that the direction of the rear wall pointing to the front wall is changed to point to the negative y-axis direction, including: In the first high-resolution labeled image, the coordinates of the centroid of the right ventricle in the basal layer, the basal 1 / 4 height layer, and the basal 1 / 2 height layer are obtained respectively. Analyze the trend of the y-coordinate change of the centroid of the right ventricle as it changes from the bottom of the heart to the apex, and determine the direction from the posterior wall to the anterior wall based on the trend of the y-coordinate change. If the direction from the rear wall to the front wall is not pointing in the negative y-axis direction, then the first high-resolution label image is flipped vertically to change the direction from the rear wall pointing to the front wall to pointing in the negative y-axis direction.

[0011] Combining the first aspect and the above implementation methods, in some possible implementations, the infarct region in the second high-resolution labeled image is mapped onto the cardiac finite element mesh model to obtain a personalized cardiac model with the infarct region, including: For each tetrahedral mesh in the finite element mesh model of the heart, determine the centroid of the tetrahedral mesh; Based on the image resolution of the second high-resolution label image, the coordinates of the grid centroid are transformed and rounded down to obtain the pixel coordinates of the grid centroid in the second high-resolution label image. When the pixel coordinates belong to the myocardial tissue region, the corresponding first label value is obtained in the second high-resolution label image according to the pixel coordinates, and the first label value is assigned to the tetrahedral mesh. If the pixel coordinates belong to a non-myocardial region, search for non-zero pixels in the first to third order neighborhoods of the pixel coordinates in turn, and assign the second label value of the first non-zero pixel found to the tetrahedral mesh. A personalized heart model with infarct regions is generated based on all tetrahedral meshes with assigned label values ​​in the finite element mesh model of the heart.

[0012] Secondly, embodiments of this application provide an electronic device, including a processor and a memory storing a computer program, wherein the processor executes the program to implement the steps of the personalized cardiac model reconstruction method with infarct region of the first aspect.

[0013] Thirdly, embodiments of this application provide a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the personalized cardiac model reconstruction method with infarct regions of the first aspect.

[0014] The personalized cardiac model reconstruction method, device, and medium with infarct regions provided in this application firstly preprocesses the patient's CMR-LGE images to obtain a first high-resolution labeled image containing ventricular cavities and infarct tissue. Simultaneously, a cardiac finite element mesh model is constructed based on the patient's high-precision cardiac cavity data. Next, the orientation of the first high-resolution labeled image is adjusted to obtain a second high-resolution labeled image. Finally, the infarct region in the second high-resolution labeled image is mapped onto the cardiac finite element mesh model, thereby obtaining a personalized cardiac model with infarct regions. Thus, obtaining the first high-resolution labeled image and the cardiac finite element mesh model through preprocessing improves the accuracy of geometric morphology from the data source. Then, by adjusting the orientation of the first high-resolution labeled image to obtain the second high-resolution labeled image, spatial alignment between the image and the model coordinate system is achieved, thereby avoiding reconstruction position deviations caused by inconsistent orientation and improving spatial accuracy. Finally, by mapping the infarct region in the correctly oriented second high-resolution labeled image onto the cardiac finite element mesh model, high-precision image information is accurately transferred to the three-dimensional model, comprehensively improving the overall reconstruction accuracy of the personalized cardiac model with infarct regions in terms of geometry and spatial location. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic flowchart of the personalized cardiac model reconstruction method with infarct region provided in the embodiments of this application; Figure 2 This is a schematic diagram of the preprocessing process provided in the embodiments of this application; Figure 3 This is a schematic diagram of the process for constructing a finite element mesh model of the heart provided in an embodiment of this application; Figure 4 This is a schematic diagram of the direction adjustment process provided in the embodiments of this application; Figure 5 This is a schematic diagram of the image mapping process provided in an embodiment of this application; Figure 6 This is a schematic diagram of the three-directional features of the dual-ventricle model provided in the embodiments of this application; Figure 7 This is a schematic diagram of the vector R between the centroids of the left and right ventricular cavities in the lower ventricle of the heart, provided in an embodiment of this application. Figure 8 This is a schematic diagram illustrating the change in the center of gravity of the right ventricle from the base to the apex of the heart, provided in an embodiment of this application. Figure 9 This is a schematic diagram of the infarct tissue mapping results provided in the embodiments of this application; Figure 10 This is a schematic diagram of the entire process of the personalized cardiac model reconstruction method with infarct region provided in the embodiments of this application; Figure 11 This is a schematic diagram illustrating the image segmentation, interpolation, and merging operations during the data preprocessing process provided in the embodiments of this application; Figure 12 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this 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 this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0018] Explanation of some terms used in the embodiments of this application: Late Gadolinium Enhancement Cardiac Magnetic Resonance (LGE-CMR): A magnetic resonance imaging technique that uses the difference in distribution of gadolinium contrast agent in myocardial tissue to identify myocardial activity, especially for detecting infarcted and fibrotic areas. In some cases, it can also be simply referred to as CMR-LGE.

[0019] Variational Implicit Function Interpolation: An interpolation method based on implicit functions that constructs a smooth surface passing through discrete constraint points by minimizing the energy function.

[0020] LogOdds Algorithm: An interpolation method based on log-odds transformation. It performs three-dimensional interpolation by mapping binarized infarct tissue slices to log-odds space, and then obtains a three-dimensional binary image through inverse transformation.

[0021] Label images: These are images used to identify different tissue regions in an image (also known as label maps). By assigning specific values ​​to different pixels or voxels in the image, different structural regions such as ventricular cavities, normal myocardial tissue, hemi-infarcted tissue, and completely infarcted tissue can be distinguished. They are one of the data carriers for image segmentation, interpolation, and subsequent mapping operations.

[0022] Personalized cardiac models with infarct regions play a crucial role in electrophysiological simulation, risk stratification, and treatment optimization for cardiac diseases. LGE-CMR imaging technology provides the data foundation for constructing these models. However, in practice, cardiac models built based on clinical imaging data often struggle to perfectly replicate the geometry and spatial location of the infarct region compared to real anatomical structures, posing a risk of accuracy loss. For instance, in related technologies, the reconstruction process commonly used for segmented images obtained from LGE-CMR imaging has limitations in ensuring the geometric accuracy of the infarct tissue, the accuracy of its three-dimensional spatial location, and the precise mapping between the infarct and the cardiac mesh model. This results in insufficient overall accuracy of the final personalized cardiac model, failing to accurately reflect the anatomical structure. This deviation in model accuracy directly affects the reliability of subsequent electrophysiological simulation results and may weaken its effectiveness in guiding clinical decision-making and research.

[0023] Therefore, improving the accuracy of personalized cardiac model reconstruction with infarcted areas has become an urgent technical problem to be solved.

[0024] To address the aforementioned issues, the main solution provided in this application includes: firstly, preprocessing the patient's CMR-LGE images to obtain a first high-resolution labeled image containing ventricular cavities and infarcted tissue; simultaneously, constructing a cardiac finite element mesh model based on the patient's high-precision cardiac cavity data; next, reorienting the first high-resolution labeled image to obtain a second high-resolution labeled image; finally, mapping the infarcted region in the second high-resolution labeled image onto the cardiac finite element mesh model to obtain a personalized cardiac model with an infarcted region. Thus, by obtaining the first high-resolution labeled image and the cardiac finite element mesh model through preprocessing, the accuracy of the geometric shape is improved from the data source; then, by reorienting the first high-resolution labeled image to obtain the second high-resolution labeled image, spatial alignment between the image and the model coordinate system is achieved, thereby avoiding reconstruction position deviations caused by inconsistent orientation and improving spatial position accuracy; finally, by mapping the infarcted region in the correctly oriented second high-resolution labeled image onto the cardiac finite element mesh model, high-precision image information is accurately transferred to the three-dimensional model, comprehensively improving the overall reconstruction accuracy of the personalized cardiac model with an infarcted region in terms of geometry and spatial position.

[0025] The following will provide a detailed description of the personalized cardiac model reconstruction method with infarcted regions provided in the embodiments of this application.

[0026] Please see Figure 1 , Figure 1 This is a schematic flowchart illustrating a personalized cardiac model reconstruction method with an infarct region, provided as an embodiment of this application. Figure 1 As shown, the method in this application embodiment may include the following steps S1-S3.

[0027] S1, preprocess the patient's CMR-LGE images to obtain the first high-resolution labeled image containing the ventricular cavity and infarct tissue, and construct a cardiac finite element mesh model based on the patient's high-precision cardiac cavity data.

[0028] Specifically, in order to obtain the three-dimensional ventricular structure and the geometry of the infarct area from the patient's two-dimensional medical image data, and to provide an image basis for subsequent spatial alignment and mapping operations, the patient's CMR-LGE images need to be preprocessed to obtain a first high-resolution labeled image containing the ventricular cavity and infarct tissue. CMR-LGE images refer to images acquired by magnetic resonance imaging (MRI) techniques that utilize the differences in gadolinium contrast agent distribution in myocardial tissue to identify myocardial activity, highlighting infarcted and fibrotic areas in the myocardium. Preprocessing refers to a series of processes performed on raw medical image data to extract key structural information, including at least image segmentation, 3D interpolation reconstruction, and data format conversion. Ventricular cavity refers to the spatial structure inside the ventricles of the heart, including the left and right ventricles. Infarcted tissue refers to the area of ​​myocardium that has died due to ischemia and hypoxia and has been replaced by fibrous tissue, including at least complete infarcted tissue and hemi-infarcted tissue. The first high-resolution labeled image containing ventricular cavity and infarcted tissue refers to a type of 3D image data in which different voxels are assigned specific label values ​​to distinguish ventricular cavity, normal myocardial tissue, hemi-infarcted tissue, and complete infarcted tissue, and this image has a resolution higher than that of the original input data.

[0029] Regarding this step, in some possible implementations, relevant image segmentation and reconstruction algorithms can be used to process the patient's CMR-LGE images. By performing the preprocessing operation, the first high-resolution labeled image containing the ventricular cavity and infarct tissue can be generated. In some possible implementations, one approach is to first perform multi-region segmentation on the patient's CMR-LGE images to obtain preliminary images of the ventricular myocardium and infarct tissue; then, perform three-dimensional interpolation reconstruction on key structures in the preliminary images to obtain high-resolution images of the cavity and infarct structures; finally, fuse the cavity and infarct structure images to generate the first high-resolution labeled image containing the ventricular cavity and infarct tissue.

[0030] Furthermore, a finite element mesh model of the heart needs to be constructed based on the patient's high-precision cardiac cavity data. Here, the high-precision cardiac cavity data refers to the set of coordinate points or surface data used to describe the geometry of the cardiac cavities, obtained by digitally representing the ventricular cavity model obtained during preprocessing. The finite element mesh model of the heart is a mathematical model that discretizes the continuous cardiac geometry into a set of finite simple elements (such as tetrahedrons), each element defined by nodes and connections, and can be used for numerical simulation.

[0031] Regarding this step, in some possible implementations, a relevant mesh generation algorithm can be used to process the patient's high-precision cardiac cavity data. This involves performing contour extraction, surface mesh generation, mesh optimization, and volume mesh generation to construct the cardiac finite element mesh model. In another possible implementation, based on the high-precision cardiac cavity data, a cardiac model contour can be obtained using geometric surface reconstruction technology. Subsequently, a relevant mesh generation technique is applied to discretize the contour, and the generated mesh is optimized for topology and geometric quality, ultimately outputting the cardiac finite element mesh model.

[0032] For example, the acquisition of high-precision cardiac cavity data of a patient can be performed as follows: select some constraint points located at the boundaries of the myocardium in the segmented myocardial image, use variational implicit function interpolation to interpolate these discrete points, generate fitting surfaces of the left ventricular endocardium and right ventricular endocardium respectively, combine the four surfaces to generate a cavity model of the two ventricles, and save it as high-resolution data.

[0033] S2, the orientation of the first high-resolution label image is adjusted to obtain the second high-resolution label image.

[0034] Specifically, in order to align the spatial coordinate system of the first high-resolution label image with the coordinate system of the cardiac finite element mesh model, thereby achieving the correspondence of the infarct region in three-dimensional space, the orientation of the first high-resolution label image needs to be adjusted to obtain a second high-resolution label image. Orientation adjustment refers to performing a geometric transformation on the first high-resolution label image to align its anatomical orientation—for example, apex pointing towards base, left ventricle towards right ventricle, and posterior wall towards anterior wall—with a preset coordinate system orientation. This may include image rotation and / or flipping. The second high-resolution label image refers to a high-resolution label image containing the ventricular cavity and infarct tissue, whose spatial coordinate system matches the coordinate system of the cardiac finite element mesh model after orientation adjustment.

[0035] Regarding this step, in some possible implementations, relevant image recognition and geometric transformation algorithms can be used to process the first high-resolution label image. By performing the orientation adjustment operation, the anatomical orientation of the image can be identified and adjusted to generate a second high-resolution label image consistent with the orientation of the target coordinate system. In some possible implementations, in one approach, the current orientation of specific anatomical structures (such as the apex of the heart and ventricular cavities) in the first high-resolution label image can be identified by analyzing their geometric distribution characteristics. Based on a preset standard coordinate system, the image can be flipped or rotated through relevant mathematical operations to complete the orientation adjustment, ultimately obtaining a second high-resolution label image that matches the standard coordinate system.

[0036] S3, map the infarct region in the second high-resolution labeled image onto the finite element mesh model of the heart to obtain a personalized heart model with the infarct region.

[0037] Specifically, to transfer the infarct tissue information recorded in the second high-resolution labeled image to the cardiac finite element mesh model, and to assign a corresponding tissue type label to each mesh element to ultimately generate a personalized cardiac model suitable for numerical simulation, it is necessary to map the infarct region in the second high-resolution labeled image to the cardiac finite element mesh model, resulting in a personalized cardiac model with infarct regions. Here, the infarct region in the second high-resolution labeled image refers to the set of voxels labeled as representing semi-infarcted and / or completely infarcted tissue in the second high-resolution labeled image; the personalized cardiac model with infarct regions refers to a personalized three-dimensional cardiac model in which each element or node in the cardiac finite element mesh model has been assigned a label representing its tissue type, thus containing information on the spatial distribution of infarct tissue.

[0038] Regarding this step, in some possible implementations, mathematical operations can be performed on each element in the cardiac finite element mesh model to determine its corresponding position in the second high-resolution labeled image. Based on the label information at that position, the attribute information of the infarct region is assigned to the element, thereby generating the personalized cardiac model with the infarct region. In some possible implementations, a spatial correspondence between the cardiac finite element mesh model and the second high-resolution labeled image can be established through a correlation mapping algorithm. For each element in the model, its corresponding region in the image can be located based on its spatial coordinates. If the region is not a myocardial region, a search for myocardial or infarct regions is further conducted within its neighborhood, and the searched tissue type label is assigned to the element. Finally, the label information of all elements is integrated to construct the personalized cardiac model with the infarct region.

[0039] In this embodiment, the patient's CMR-LGE images are first preprocessed to obtain a first high-resolution labeled image containing ventricular cavities and infarcted tissue. Simultaneously, a cardiac finite element mesh model is constructed based on the patient's high-precision cardiac cavity data. Next, the orientation of the first high-resolution labeled image is adjusted to obtain a second high-resolution labeled image. Finally, the infarcted region in the second high-resolution labeled image is mapped onto the cardiac finite element mesh model, thereby obtaining a personalized cardiac model with the infarcted region. Thus, by obtaining the first high-resolution labeled image and the cardiac finite element mesh model through preprocessing, the accuracy of the geometric shape is improved from the data source. Then, by adjusting the orientation of the first high-resolution labeled image to obtain the second high-resolution labeled image, spatial alignment between the image and the model coordinate system is achieved, thereby avoiding reconstruction position deviations caused by inconsistent orientation and improving the accuracy of spatial position. Finally, by mapping the infarcted region in the correctly oriented second high-resolution labeled image onto the cardiac finite element mesh model, high-precision image information is accurately transferred to the three-dimensional model, comprehensively improving the overall reconstruction accuracy of the personalized cardiac model with the infarcted region in terms of geometry and spatial position.

[0040] Please see Figure 2 This document provides a schematic diagram of a preprocessing procedure for embodiments of this application, such as... Figure 2 As shown, the method of this application embodiment may include the following steps S101-S104, which can be used as a further refinement of the above step "preprocessing the patient's CMR-LGE image to obtain a first high-resolution labeled image containing the ventricular cavity and infarct tissue".

[0041] S101, segment the patient’s CMR-LGE images to obtain ventricular myocardial images and infarct tissue images. The infarct tissue images include complete infarct tissue regions and hemi-infarct tissue regions. S102, interpolation reconstruction is performed based on ventricular myocardial images to obtain high-resolution labeled images of ventricular myocardial images; S103, interpolation reconstruction is performed based on the infarct tissue image to obtain a high-resolution label image of the infarct tissue; S104, merge the high-resolution labeled image of the ventricular myocardium with the high-resolution labeled image of the infarct tissue to generate a first high-resolution labeled image containing the ventricular cavity and the infarct tissue.

[0042] Specifically, the patient's CMR-LGE images first need to be segmented to obtain ventricular myocardial images and infarct tissue images. Infarct tissue images include completely infarcted areas and hemi-infarcted areas. Segmentation refers to the process of distinguishing and identifying different tissue structures or regions in the CMR-LGE images based on their image characteristics. Ventricular myocardial images refer to two-dimensional or three-dimensional image data containing only information about ventricular myocardial tissue. Infarct tissue images refer to two-dimensional or three-dimensional image data that identifies the location of the myocardial infarction area. Completely infarcted areas refer to areas where cardiomyocytes are completely necrotic and replaced by fibrous tissue; hemi-infarcted areas refer to areas where cardiomyocytes are partially damaged and their electrophysiological properties have been remodeled.

[0043] Regarding this step, in some possible implementations, a deep learning-based image segmentation algorithm can be used to process the patient's CMR-LGE image, and the segmented ventricular myocardial image can be used as the ventricular myocardial image. At the same time, a Gaussian mixture model or threshold segmentation method can be used to identify the infarcted tissue region in the ventricular myocardial image, and the identified complete infarcted tissue region and half infarcted tissue region can be combined to generate an infarcted tissue image.

[0044] Furthermore, interpolation reconstruction based on the ventricular myocardial images is required to obtain high-resolution labeled images of the ventricular myocardium. These high-resolution labeled images refer to 3D image data where, after resolution enhancement through interpolation algorithms, each voxel is assigned a label value identifying the ventricular myocardial type.

[0045] Regarding this step, in some possible implementations, discrete constraint points located at the myocardial boundary can be selected in the ventricular myocardial image, and the discrete constraint points can be interpolated and reconstructed using the variational implicit function interpolation method to generate fitting surfaces of the left and right ventricular endocardium and endocardium. The fitting surfaces are then combined into a biventricular cavity structure, and the biventricular cavity structure is converted into a high-resolution labeled image of the ventricular myocardium.

[0046] Furthermore, interpolation reconstruction based on the infarct tissue image is required to obtain a high-resolution labeled image of the infarct tissue. Here, the high-resolution labeled image of the infarct tissue refers to the 3D image data where, after the resolution is increased through an interpolation algorithm, each voxel is assigned a label value identifying the infarct tissue type.

[0047] Regarding this step, in some possible implementations, a probabilistic logarithmic algorithm can be used to perform three-dimensional interpolation reconstruction on the fully infarcted tissue region and the half-infarcted tissue region in the infarcted tissue image, convert the interpolation result into a binary image, and use the binary image as a high-resolution label image of the infarcted tissue.

[0048] Furthermore, the high-resolution labeled image of the ventricular myocardium is merged with the high-resolution labeled image of the infarct tissue to generate a first high-resolution labeled image containing the ventricular cavity and the infarct tissue.

[0049] Regarding this step, in some possible implementations, the high-resolution labeled image of ventricular myocardium and the high-resolution labeled image of infarct tissue can be spatially registered. Based on the registration result, the label values ​​in the high-resolution labeled image of infarct tissue can be mapped to the high-resolution labeled image of ventricular myocardium to generate a first high-resolution labeled image containing ventricular cavity, normal myocardial tissue, half-infarct tissue region and complete infarct tissue region.

[0050] In this embodiment, the patient's CMR-LGE images are segmented to obtain ventricular myocardial images and infarct images containing both complete and partial infarct regions. Then, the ventricular myocardial images and infarct images are interpolated and reconstructed separately to obtain corresponding high-resolution labeled images of the ventricular myocardial system and the infarct tissue. Finally, these two images are merged to generate a first high-resolution labeled image containing the ventricular cavity and infarct tissue. This ensures the complete preservation and resolution enhancement of the three-dimensional geometry of the ventricular myocardial structure and infarct tissue, providing a high-precision and spatially consistent data foundation for subsequent orientation adjustment of the first high-resolution labeled image and accurate mapping of the infarct region in the orientation-adjusted second high-resolution labeled image to the cardiac finite element mesh model.

[0051] Please see Figure 3 This application provides a schematic diagram of the process for constructing a finite element mesh model of the heart, as illustrated in the embodiments below. Figure 3 As shown, the method of this application embodiment may include the following steps S105-S106, which can be used as a further refinement of the above step "constructing a cardiac finite element mesh model based on the patient's high-precision cardiac cavity data".

[0052] S105, Determine the outline of the heart model based on high-precision heart cavity data; S106, perform surface meshing, mesh quality optimization, and volume meshing on the heart model outline to generate a heart finite element mesh model; wherein, the heart finite element mesh model is saved as a node coordinate file that records node coordinates and a mesh topology file that records the vertex indices of tetrahedral elements.

[0053] Specifically, the first step is to determine the contour of the heart model based on high-precision cardiac cavity data. The heart model contour refers to the three-dimensional boundary surface that characterizes the geometry of the heart.

[0054] Regarding this step, in some possible implementations, a surface reconstruction algorithm can be used to process high-precision cardiac cavity data. By extracting key geometric feature points of the endocardium and endocardium and fitting them into a continuous surface, the resulting surface can be used as the contour of the heart model.

[0055] Furthermore, the heart model outline is subjected to surface meshing, mesh quality optimization, and volume meshing to generate a finite element mesh model of the heart. The heart finite element mesh model is saved as a node coordinate file recording node coordinates and a mesh topology file recording the vertex indices of tetrahedral elements. Surface meshing refers to the process of discretizing the heart model outline into a set of triangular or quadrilateral facets; mesh quality optimization refers to adjusting the shape and distribution of facets to eliminate distorted elements and improve geometric accuracy; volume meshing refers to the process of filling the surface mesh with tetrahedral or hexahedral volume elements to form a three-dimensional mesh. The node coordinate file is a data file storing the spatial position information of all nodes in the mesh, and can be represented as a file with the .pts extension. The mesh topology file is an index file describing the connection relationships of mesh elements, and can be represented as a file with the .elem extension.

[0056] Regarding this step, in some possible implementations, surface meshing can be performed on the contour of the heart model to generate an initial surface mesh. The quality of the surface mesh can be adjusted by an optimization algorithm, and volume meshing can be performed based on the optimized surface mesh. The generated node coordinates and element topology information can be saved as node coordinate files and mesh topology files, respectively, and finally the finite element mesh model of the heart can be obtained.

[0057] In this embodiment, the contour of the heart model is determined based on high-precision cardiac cavity data, which can capture the geometric features of the patient's heart. By sequentially performing surface mesh generation, mesh quality optimization, and volume mesh generation on the heart model contour, the rationality of the mesh elements in terms of shape and distribution is ensured. Finally, the generated node coordinates and element topology information are saved as node coordinate files and mesh topology files, respectively, to construct a heart finite element mesh model containing complete geometric and topological information. This provides a three-dimensional structural basis for subsequently mapping the infarct area in the second high-resolution labeled image to the heart finite element mesh model.

[0058] Please see Figure 4 This document provides a schematic flowchart for a direction adjustment process in an embodiment of this application. Figure 4 As shown, the method of this application embodiment may include the following steps S201-S204. Steps S201-S204 can be used as a further refinement of the above step "adjusting the orientation of the first high-resolution label image to obtain a second high-resolution label image".

[0059] S201, Identify the direction from the apex of the heart to the base of the heart in the first high-resolution label image, and adjust the orientation of the first high-resolution label image so that the direction from the apex of the heart to the base of the heart is changed to the positive z-axis direction; S202, Identify the direction in which the left ventricle points towards the right ventricle in the first high-resolution label image, and adjust the orientation of the first high-resolution label image so that the direction in which the left ventricle points towards the right ventricle is changed to the negative x-axis direction; S203, identify the direction in which the rear wall points to the front wall in the first high-resolution label image, and adjust the orientation of the first high-resolution label image so that the direction in which the rear wall points to the front wall is changed to the negative y-axis direction; S204. Based on the first high-resolution label image after orientation adjustment, generate a second high-resolution label image that is consistent with the coordinate system of the heart finite element mesh model.

[0060] Specifically, on the one hand, it is necessary to identify the direction from the apex of the heart to the base of the heart in the first high-resolution label image, and then adjust the orientation of the first high-resolution label image so that the direction from the apex of the heart to the base of the heart is changed to point in the positive z-axis direction. Here, the direction from the apex of the heart to the base of the heart refers to the anatomically extending direction from the apex of the heart to the base of the heart in the first high-resolution label image, which is usually consistent with the long axis of the heart; the positive z-axis direction refers to the direction extending along the positive z-axis in the preset three-dimensional coordinate system.

[0061] Regarding this step, in some possible implementations, the direction from the apex to the base of the heart can be determined by analyzing the myocardial area distribution characteristics of different layers in the first high-resolution labeled image. Specifically, this includes counting the number of myocardial pixels in each two-dimensional slice along the z-axis, identifying the layer with the largest myocardial area as the base of the heart, and the layer with the smallest myocardial area as the apex of the heart, and thus determining the direction from the apex to the base of the heart. If the apex of the heart is located above the base of the heart, then a flip operation along the z-axis is performed on the first high-resolution labeled image to align the direction from the apex to the base of the heart with the positive z-axis direction.

[0062] On the other hand, it is necessary to identify the direction in which the left ventricle points towards the right ventricle in the first high-resolution labeled image, and then adjust the orientation of the first high-resolution labeled image so that the direction of the left ventricle pointing towards the right ventricle is changed to the negative x-axis direction. Here, the direction of the left ventricle pointing towards the right ventricle refers to the anatomically extending direction from the left ventricular cavity to the right ventricular cavity in the first high-resolution labeled image; the negative x-axis direction refers to the direction extending along the negative x-axis in the preset three-dimensional coordinate system.

[0063] Regarding this step, in some possible implementations, the direction from left ventricle to right ventricle can be determined by segmenting the non-myocardial region in the first high-resolution labeled image and calculating its geometric features. Specifically, this includes dividing the non-myocardial region into the left ventricular cavity, the right ventricular cavity, and the image background according to connectivity rules, where the image background is the region with the largest distribution range; calculating the roundness of the left and right ventricular cavities at the basal layer of the heart, identifying the cavity with a roundness closer to 1 as the left ventricular cavity; calculating the vector R between the centroids of the two cavities, and if the angle difference between vector R and the negative x-axis is greater than 45 degrees, then performing a rotation or flip operation on the first high-resolution labeled image to align the direction from left ventricle to right ventricle with the negative x-axis.

[0064] On the other hand, it is necessary to identify the direction from the posterior wall to the anterior wall in the first high-resolution label image, and then adjust the orientation of the first high-resolution label image so that the direction from the posterior wall to the anterior wall is changed to point in the negative y-axis direction. Here, the direction from the posterior wall to the anterior wall refers to the anatomically extending direction from the posterior wall of the heart to the anterior wall in the first high-resolution label image; the negative y-axis direction refers to the direction extending along the negative y-axis in the preset three-dimensional coordinate system.

[0065] Regarding this step, in some possible implementations, the direction of the posterior wall pointing towards the anterior wall can be determined by analyzing the spatial variation characteristics of the centroid of the right ventricular cavity. Specifically, this involves obtaining the centroid coordinates of the right ventricular cavity at the basal layer, the basal 1 / 4 height layer, and the basal 1 / 2 height layer, and analyzing the trend of the y-coordinate change of the centroid as it changes from the basal layer towards the apex. If the y-coordinate decreases, it is determined that the posterior wall is facing the negative y-axis, and the first high-resolution label image needs to be flipped vertically. If the y-coordinate increases, it is determined that the anterior wall is facing the negative y-axis, and no adjustment is needed. Further verification is performed on the average y-coordinate of the minimum x-coordinate point of the right ventricular cavity at the basal layer. If it is less than the midpoint of the y-axis of the right ventricular cavity, it indicates an abnormal posterior wall orientation, requiring manual intervention.

[0066] Finally, based on the first high-resolution label image after orientation adjustment, a second high-resolution label image is generated that is consistent with the coordinate system of the heart finite element mesh model.

[0067] Regarding this step, in some possible implementations, the first high-resolution label image, after being adjusted by the direction from the apex to the base of the heart, from the left ventricle to the right ventricle, and from the posterior wall to the anterior wall, can be used as the second high-resolution label image. This ensures that its spatial coordinate system is completely aligned with the coordinate system of the cardiac finite element mesh model, providing an accurate geometric reference for subsequent infarct area mapping.

[0068] In this embodiment, the anatomical orientation of the first high-resolution label image is ensured to be consistent with the preset three-dimensional coordinate system by identifying and adjusting the apex-to-base direction in the first high-resolution label image, aligning this direction with the positive z-axis; the left ventricle-to-right ventricle direction is identified and adjusted, aligning this direction with the negative x-axis; and the posterior wall-to-anterior wall direction is identified and adjusted, aligning this direction with the negative y-axis. A second high-resolution label image is then generated based on the first high-resolution label image with adjusted orientation, achieving spatial registration between the coordinate system of the second high-resolution label image and the coordinate system of the cardiac finite element mesh model. This provides a precise geometric basis for accurately mapping the infarct region in the second high-resolution label image to the cardiac finite element mesh model.

[0069] In one embodiment, the above step of "identifying the direction from the apex of the heart to the base of the heart in the first high-resolution label image, and adjusting the orientation of the first high-resolution label image so that the direction from the apex of the heart to the base of the heart is changed to the positive z-axis direction" can be further refined and may include the following steps: Each two-dimensional slice along the z-axis in the first high-resolution label image is defined as a layer, with the layer with the largest z-coordinate value being the top layer and the layer with the smallest z-coordinate value being the bottom layer. The number of myocardial pixels contained in the top and bottom layers are counted separately. The layer with more myocardial pixels is identified as the bottom layer of the heart, and the layer with fewer myocardial pixels is identified as the apical layer of the heart. Compare the z-coordinate values ​​of the basal and apical layers of the heart to determine the direction from the apex to the basal layer; If the direction from the apex of the heart to the base of the heart indicates that the apical layer is above the base of the heart, then the first high-resolution label image is flipped along the z-axis so that the direction from the apex of the heart to the base of the heart is changed to the positive direction of the z-axis.

[0070] Specifically, it is first necessary to determine that each two-dimensional slice along the z-axis in the first high-resolution label image is a layer, with the layer with the largest z-coordinate value being the top layer and the layer with the smallest z-coordinate value being the bottom layer.

[0071] Regarding this step, in some possible implementations, the image can be traversed layer by layer along the z-axis based on the three-dimensional coordinate axis definition of the first high-resolution label image, with each two-dimensional slice treated as an independent layer; wherein, the layer with the largest z-coordinate value is marked as the top layer, and the layer with the smallest z-coordinate value is marked as the bottom layer.

[0072] Furthermore, the number of myocardial pixels contained in the top and bottom layers are counted separately. Layers with a large number of myocardial pixels are identified as the basal layer of the heart, and layers with a small number of myocardial pixels are identified as the apical layer of the heart.

[0073] Regarding this step, in some possible implementations, it is possible to traverse every pixel at the top and bottom layers and count the number of pixels marked as myocardial tissue; based on the comparison of the number of myocardial pixels, the layer with a larger number of myocardial pixels is identified as the basal layer of the heart, and the layer with a smaller number of myocardial pixels is identified as the apical layer of the heart.

[0074] Furthermore, the z-coordinate values ​​of the basal and apical layers are compared to determine the direction from the apex to the basal layer.

[0075] Regarding this step, in some possible implementations, the z-coordinate values ​​of the basal and apical layers of the heart can be obtained, and the direction from the apex to the basal layer can be determined by comparing the magnitudes of the two values. If the z-coordinate value of the apical layer is greater than that of the basal layer, then the direction from the apex to the basal layer is from bottom to top; if the z-coordinate value of the apical layer is less than that of the basal layer, then the direction from the apex to the basal layer is from top to bottom.

[0076] If the direction from the apex of the heart to the base of the heart indicates that the apical layer is above the base of the heart, then the first high-resolution label image is flipped along the z-axis so that the direction from the apex of the heart to the base of the heart is changed to the positive direction of the z-axis.

[0077] Regarding this step, in some possible implementations, when it is determined that the apex layer is above the basal layer, a flipping operation along the z-axis can be performed on the first high-resolution label image; in the flipped first high-resolution label image, the direction from the apex to the basal layer is consistent with the positive z-axis direction.

[0078] If the direction from the apex of the heart to the base of the heart indicates that the apical layer is below the base of the heart, then the original orientation of the first high-resolution label image is maintained, and no flipping operation is required.

[0079] In this embodiment, by determining each layer along the z-axis in the first high-resolution label image and counting the number of myocardial pixels contained in the top and bottom layers, the basal and apical layers of the heart can be accurately identified based on the anatomical feature that the myocardial area of ​​the basal layer is much larger than that of the apical layer. Furthermore, by comparing the z-coordinate values ​​of the basal and apical layers, the actual direction of the current image's center apex pointing towards the basal layer can be determined. If this direction indicates that the apical layer is located above the basal layer, the first high-resolution label image is flipped along the z-axis to adjust the direction of the apex pointing towards the basal layer to the positive z-axis direction. This ensures that the axis of the first high-resolution label image is aligned with the preset three-dimensional coordinate system, providing an accurate spatial reference for subsequently mapping the infarct area in the second high-resolution label image to the cardiac finite element mesh model.

[0080] In one embodiment, the step of "identifying the direction of the left ventricle pointing towards the right ventricle in the first high-resolution label image, and adjusting the orientation of the first high-resolution label image so that the direction of the left ventricle pointing towards the right ventricle is changed to the negative x-axis direction" can be further refined and may include the following steps: Based on connectivity rules, the non-myocardial regions in the first high-resolution labeled image are divided into the left ventricular cavity, the right ventricular cavity, and the image background; among them, the circularity of the left ventricular cavity at the cardiac base is closer to 1 than that of the right ventricular cavity at the cardiac base. Determine the vector R between the centroid of the left ventricle and the centroid of the right ventricle; Compare the angle between vector R and the negative x-axis to determine the direction in which the left ventricle points towards the right ventricle; If the direction from the left ventricle to the right ventricle is not pointing in the negative x-axis direction, then the first high-resolution label image is rotated and / or flipped so that the direction from the left ventricle to the right ventricle is changed to point in the negative x-axis direction.

[0081] Specifically, the non-myocardial regions in the first high-resolution labeled image first need to be divided into the left ventricular cavity, right ventricular cavity, and image background according to connectivity rules. The circularity of the left ventricular cavity at the cardiac basal layer is closer to 1 than that of the right ventricular cavity at the cardiac basal layer. Connectivity rules refer to criteria for classifying image regions based on pixel neighborhood relationships. By judging the spatial connectivity between pixels, non-myocardial regions are divided into independent connected regions, and these connected regions are labeled as left ventricular cavity, right ventricular cavity, or image background based on their geometric features.

[0082] Regarding this step, in some possible implementations, a region growing algorithm or a watershed algorithm can be used to perform connectivity analysis on the non-myocardial regions in the first high-resolution labeled image to generate multiple connected regions. By calculating the area and shape parameters of each connected region, the connected region with the largest area is marked as the image background. For the remaining connected regions, their roundness at the cardiac basal layer is further calculated, and the connected regions with a roundness close to 1 are identified as the left ventricular cavity, while the remaining connected regions are marked as the right ventricular cavity.

[0083] Furthermore, the vector R between the centroid of the left ventricle and the centroid of the right ventricle is determined.

[0084] Regarding this step, in some possible implementations, all pixels in the left and right ventricular cavities can be traversed, the average pixel coordinates of the two cavities can be calculated, and the average value can be used as the centroid coordinates of the corresponding cavity. By subtracting the centroid coordinates of the left and right ventricular cavities, the vector R can be obtained.

[0085] Furthermore, the angle between vector R and the negative x-axis is compared to determine the direction in which the left ventricle points towards the right ventricle.

[0086] Regarding this step, in some possible implementations, the angle between vector R and the negative x-axis can be calculated, where the negative x-axis is a unit vector along the negative x-axis in a preset coordinate system; if the angle is less than or equal to a preset threshold (such as 45 degrees), it is determined that the direction from the left ventricle to the right ventricle is consistent with the negative x-axis; if the angle is greater than the preset threshold, it is determined that the direction from the left ventricle to the right ventricle is inconsistent with the negative x-axis.

[0087] If the left ventricle points towards the right ventricle in a direction that does not point towards the negative x-axis, then the first high-resolution label image needs to be rotated and / or flipped to change the left ventricle pointing towards the right ventricle to point towards the negative x-axis.

[0088] Regarding this step, in some possible implementations, the first high-resolution label image can be rotated or flipped based on the angle between vector R and the negative x-axis. Specifically, if the angle is greater than 90 degrees, the image is rotated 180 degrees; if the angle is between 45 and 90 degrees, the image is rotated 90 degrees or flipped along a specific axis so that the adjusted vector R is aligned with the negative x-axis.

[0089] If the direction from the left ventricle to the right ventricle is in the negative x-axis direction, then the current orientation of the first high-resolution label image is maintained, and no rotation or flipping operation is required.

[0090] In this embodiment, non-myocardial regions are classified using connectivity rules, and the left and right ventricular cavities are accurately identified by combining roundness differences. The direction of the left ventricle pointing towards the right ventricle is quantitatively determined by calculating the centroid vector R of the cavity and comparing the angle with the negative x-axis. If the directions are inconsistent, adjustments are made by rotation or flipping to ensure that the left ventricle pointing towards the right ventricle in the first high-resolution label image is aligned with the negative x-axis, providing an accurate coordinate system basis for subsequent infarct region mapping.

[0091] In one embodiment, the above step of "identifying the direction in which the rear wall points to the front wall in the first high-resolution label image, and adjusting the orientation of the first high-resolution label image so that the direction in which the rear wall points to the front wall is changed to the negative y-axis direction" can be further refined and may include the following steps: In the first high-resolution labeled image, the coordinates of the centroid of the right ventricle in the basal layer, the basal 1 / 4 height layer, and the basal 1 / 2 height layer are obtained respectively. Analyze the trend of the y-coordinate change of the centroid of the right ventricle as it changes from the bottom of the heart to the apex, and determine the direction from the posterior wall to the anterior wall based on the trend of the y-coordinate change. If the direction from the rear wall to the front wall is not pointing in the negative y-axis direction, then the first high-resolution label image is flipped vertically to change the direction from the rear wall pointing to the front wall to pointing in the negative y-axis direction.

[0092] Specifically, the first step is to obtain the centroid coordinates of the right ventricular cavity in the first high-resolution labeled image at the base of the heart, the lateral quarter-height layer, and the lateral half-height layer. Here, the base of the heart refers to the two-dimensional slice layer corresponding to the bottom of the heart in the first high-resolution labeled image; the lateral quarter-height layer refers to the two-dimensional slice layer extending from the base of the heart towards the apex to a position one-quarter of the total height; the lateral half-height layer refers to the two-dimensional slice layer extending from the base of the heart towards the apex to a position half of the total height; and the centroid coordinates of the right ventricular cavity refer to the average coordinates of all pixels within the right ventricular cavity region.

[0093] Regarding this step, in some possible implementations, an image processing algorithm can be used to scan the first high-resolution label image layer by layer to identify the pixel regions belonging to the right ventricular cavity in the basal layer, the basal 1 / 4 height layer, and the basal 1 / 2 height layer. By calculating the average coordinates of the right ventricular cavity pixels in each layer, the centroid coordinates of the right ventricular cavity in each layer can be obtained.

[0094] Furthermore, the trend of the y-coordinate change of the right ventricular centroid as it changes from the base of the heart to the apex was analyzed. Based on this trend, the direction from the posterior wall to the anterior wall was determined. Here, the trend of the y-coordinate change refers to the numerical change of the right ventricular centroid along the y-axis; the direction from the posterior wall to the anterior wall refers to the anatomically extending direction from the posterior wall to the anterior wall in the first high-resolution labeled image.

[0095] Regarding this step, in some possible implementations, the y-coordinate values ​​of the centroid of the right ventricle in the basal layer, the basal 1 / 4 height layer, and the basal 1 / 2 height layer can be compared. If the y-coordinate value decreases with increasing height, then the direction from the posterior wall to the anterior wall is determined to be the negative y-axis direction; if the y-coordinate value increases with increasing height, then the direction from the posterior wall to the anterior wall is determined to be the positive y-axis direction.

[0096] If the direction from the rear wall to the front wall is not pointing in the negative y-axis direction, then the first high-resolution label image needs to be flipped vertically to change the direction from the rear wall pointing to the front wall to pointing in the negative y-axis direction.

[0097] Regarding this step, in some possible implementations, when it is determined that the direction from the rear wall to the front wall is the positive direction of the y-axis, a flip operation along the y-axis can be performed on the first high-resolution label image so that the adjusted direction from the rear wall to the front wall is consistent with the negative direction of the y-axis.

[0098] If the direction from the rear wall to the front wall is in the negative y-axis direction, then the current orientation of the first high-resolution label image is maintained, and no flipping operation is required.

[0099] In this embodiment, by acquiring the coordinates of the right ventricular centroid in the basal layer, the basal 1 / 4 height layer, and the basal 1 / 2 height layer, the spatial variation characteristics of the right ventricular centroid can be captured. By analyzing the trend of the y-coordinate change of the right ventricular centroid, the actual direction of the posterior wall pointing towards the anterior wall in the current image can be determined. If this direction does not point towards the negative y-axis, an adjustment is made by flipping the image up and down to ensure that the posterior wall pointing towards the anterior wall in the first high-resolution label image is aligned with the negative y-axis. This process utilizes the anatomical feature that the right ventricular outflow tract is located on the anterior wall and disappears relatively quickly when changing towards the apex. By quantitatively analyzing the spatial displacement of the right ventricular centroid, the identification and adjustment of the anterior and posterior wall directions of the first high-resolution label image are achieved, thereby ensuring that the adjusted second high-resolution label image is consistent with the coordinate system of the cardiac finite element mesh model in the anterior and posterior wall directions, providing a geometric reference for subsequent infarct area mapping.

[0100] Please see Figure 5 This document provides a schematic diagram of an image mapping process in an embodiment of this application, as shown below. Figure 5 As shown, the method of this application embodiment may include the following steps S301-S305. Steps S301-S305 can be used as a further refinement of the above step "mapping the infarct region in the second high-resolution labeled image to the cardiac finite element mesh model to obtain a personalized cardiac model with the infarct region".

[0101] S301, for each tetrahedral mesh in the finite element mesh model of the heart, determine the mesh centroid of the tetrahedral mesh; S302, perform coordinate transformation on the grid centroid based on the image resolution of the second high-resolution label image and round down to obtain the pixel coordinates of the grid centroid in the second high-resolution label image; S303, when the pixel coordinates belong to the myocardial tissue region, obtain the corresponding first label value in the second high-resolution label image according to the pixel coordinates, and assign the first label value to the tetrahedral grid. S304, when the pixel coordinates belong to the non-myocardial region, sequentially search for non-zero pixels in the first to third order neighborhoods of the pixel coordinates, and assign the second label value of the first non-zero pixel found to the tetrahedral mesh. S305 generates a personalized heart model with infarct regions based on all tetrahedral meshes with assigned label values ​​in the finite element mesh model of the heart.

[0102] Specifically, the first step is to determine the centroid of each tetrahedral mesh in the cardiac finite element mesh model. A tetrahedral mesh refers to a three-dimensional geometric unit composed of four vertices, which is the basic building block of the cardiac finite element mesh model. The centroid of a tetrahedral mesh refers to the spatial location represented by the average coordinates of its four vertices.

[0103] Regarding this step, in some possible implementations, all tetrahedral meshes in the heart finite element mesh model can be traversed, and the arithmetic mean of the coordinates of the four vertices of each tetrahedral mesh can be calculated. The average value obtained can then be used as the centroid of the tetrahedral mesh.

[0104] Furthermore, the grid centroid needs to be coordinate transformed and rounded down based on the image resolution of the second high-resolution label image to obtain the pixel coordinates of the grid centroid in the second high-resolution label image. Specifically, coordinate transformation and rounding down refers to converting the three-dimensional physical coordinates of the grid centroid into discrete pixel coordinates in the image coordinate system based on the image resolution, and then rounding the transformed coordinate values ​​down.

[0105] Regarding this step, in some possible implementations, the image resolution of the second high-resolution label image in the x, y, and z directions can be obtained. Each physical coordinate value of the grid centroid is divided by the image resolution in the corresponding direction, and the division result is rounded down to obtain the pixel coordinates of the grid centroid in the second high-resolution label image.

[0106] When the pixel coordinates belong to the myocardial tissue region, the corresponding first label value is obtained from the second high-resolution labeled image based on the pixel coordinates, and the first label value is assigned to the tetrahedral grid. Here, the first label value refers to the numerical value in the second high-resolution labeled image used to identify the type of myocardial tissue, such as the label value corresponding to normal myocardial tissue, hemi-infarcted myocardial tissue, or complete infarcted tissue.

[0107] Regarding this step, in some possible implementations, it can be determined whether the pixel coordinates in the second high-resolution labeled image belong to the myocardial tissue region. If so, the corresponding label value at the pixel coordinate position is read as the first label value, and the first label value is assigned to the corresponding tetrahedral mesh.

[0108] When the pixel coordinates belong to a non-myocardial region, non-zero pixels within the first-order to third-order neighborhoods of the pixel coordinates are searched sequentially, and the second label value of the first non-zero pixel found is assigned to the tetrahedral mesh. The first-order to third-order neighborhoods refer to the range of neighborhoods expanded sequentially in three-dimensional space centered on the pixel coordinates; for example, the first-order neighborhood is 26 neighborhoods, the second-order neighborhood is 124 neighborhoods, and the third-order neighborhood is 342 neighborhoods. Non-zero pixels refer to pixels with label values ​​that are not zero, typically representing myocardial tissue or infarcted tissue. The second label value refers to the label value of the first non-zero pixel found within the neighborhood.

[0109] Regarding this step, in some possible implementations, when the pixel coordinates belong to a non-myocardial region, the search can start from the first-order neighborhood of the pixel coordinates and sequentially search to the third-order neighborhood. The label value of the first non-zero pixel found is used as the second label value, and the second label value is assigned to the corresponding tetrahedral mesh.

[0110] Finally, a personalized heart model with infarct regions is generated based on all tetrahedral meshes with assigned label values ​​in the finite element mesh model of the heart.

[0111] Regarding this step, in some possible implementations, all tetrahedral meshes with assigned label values ​​in the cardiac finite element mesh model can be integrated, and the tissue type can be determined based on the label value of each mesh, thereby generating a personalized cardiac model containing information on the spatial distribution of the infarct area.

[0112] In this embodiment, the centroid of each tetrahedral mesh in the cardiac finite element mesh model is determined, thereby representing the spatial position of the three-dimensional discrete geometric unit. Then, the centroid is coordinate-transformed and rounded down based on the image resolution of the second high-resolution label image, establishing a mapping relationship between the centroid and the image pixel coordinates. When the pixel coordinates belong to the myocardial tissue region, the corresponding first label value is obtained and assigned to the tetrahedral mesh, conveying the myocardial tissue type information. When the pixel coordinates belong to a non-myocardial region, the label assignment problem when the centroid falls in a non-myocardial region is solved by sequentially searching non-zero pixels in the first to third-order neighborhoods and assigning the second label value of the first non-zero pixel to the tetrahedral mesh. Finally, based on all the tetrahedral meshes with assigned label values ​​in the cardiac finite element mesh model, a personalized cardiac model with an infarct region is generated, realizing the mapping of the infarct region from the second high-resolution label image to the cardiac finite element mesh model.

[0113] In some embodiments, see Figure 6 , Figure 6This diagram illustrates the three directional features of the biventricular model provided in this embodiment. Wherein, T represents the direction from the apex to the base of the heart, R represents the direction from the left ventricle to the right ventricle, and A represents the direction from the posterior wall to the anterior wall. In this diagram, the T direction extends along the positive z-axis, the R direction extends along the negative x-axis, and the A direction extends along the negative y-axis. In this embodiment, by identifying the apex-to-base direction, the left ventricle-to-right ventricle direction, and the posterior wall-to-anterior wall direction, and adjusting the label image through flipping, rotation, etc., so that the apex-to-base direction points to the positive z-axis, the left ventricle-to-right ventricle direction points to the negative x-axis, and the posterior wall-to-anterior wall direction points to the negative y-axis, the coordinate system of the label image is ensured to be consistent with the coordinate system of the cardiac finite element mesh model.

[0114] In some embodiments, see Figure 7 , Figure 7 This is a schematic diagram of the vector R between the centroids of the left and right ventricles in the lower ventricle, as provided in an embodiment of this application. R represents the vector obtained by subtracting the centroid coordinates of the left and right ventricles. In this embodiment, the angle difference between vector R and the negative x-axis is calculated. If this angle difference is less than a preset threshold, the orientation of the left and right ventricles is correct; otherwise, the first high-resolution label image is rotated and / or flipped to align vector R with the negative x-axis.

[0115] In some embodiments, see Figure 8 , Figure 8 This is a schematic diagram illustrating the change in the center of gravity of the right ventricle from the base to the apex, provided in an embodiment of this application. The base of the heart, the right ventricular quarter-height layer, and the right ventricular half-height layer represent three specific slice layers from the base of the heart towards the apex. Specifically, in this embodiment, by obtaining the coordinates of the center of gravity of the right ventricular cavity in the base of the heart, the right ventricular quarter-height layer, and the right ventricular half-height layer, the trend of the y-coordinate change of the center of gravity of the right ventricular cavity as it changes from the base of the heart towards the apex is analyzed. If the y-coordinate decreases with increasing height, it is determined that the direction from the posterior wall to the anterior wall is the negative y-axis direction, and the first high-resolution label image needs to be flipped vertically. If the y-coordinate increases with increasing height, it is determined that the anterior wall is facing the negative y-axis direction, and no adjustment is needed. After adjustment, the average y-coordinate of the minimum x-coordinate point of the right ventricular cavity in the base of the heart is calculated. If this average value is less than the midpoint of the y-axis of the right ventricular cavity, it indicates that the posterior wall is facing the negative y-axis direction, and the image features do not conform to the normal ventricular structure, requiring manual observation and adjustment.

[0116] In some embodiments, see Figure 9 , Figure 9This is a schematic diagram of the infarct tissue mapping results provided in this application embodiment. The green area represents semi-infarct tissue, and the blue area represents completely infarct tissue. Specifically, in this embodiment, by mapping the infarct region in the second high-resolution label image to a cardiac finite element mesh model, the centroid of each tetrahedral mesh is calculated, and the centroid coordinates are converted into pixel coordinates in the second high-resolution label image. The tissue type is determined based on the Label value corresponding to the pixel coordinates. If the pixel coordinates are in a non-myocardial region, non-zero pixels in the neighborhood are searched sequentially and assigned values.

[0117] In one embodiment, for a better understanding of the method for reconstructing a personalized cardiac model with an infarct region, please refer to [link to relevant documentation]. Figure 10 , Figure 10 This is a schematic diagram illustrating the entire process of a personalized cardiac model reconstruction method with an infarct region provided in an embodiment of this application.

[0118] Specifically, the process is divided into three parts: the first part is data preprocessing, the second part is direction adjustment, and the third part is mapping.

[0119] In the first part, data preprocessing, using CMR-LGE images as input, image segmentation and interpolation operations are first performed: on one hand, segmentation and interpolation are performed based on the CMR-LGE images to obtain high-resolution labeled images of the ventricular myocardium; on the other hand, interpolation is performed based on the infarcted tissue regions segmented from the CMR-LGE images to obtain high-resolution labeled images of the infarcted tissue. Subsequently, the high-resolution labeled images of the ventricular myocardium and the high-resolution labeled images of the infarcted tissue are merged to generate a first high-resolution labeled image; simultaneously, a finite element mesh model of the heart is constructed based on the patient's high-precision cardiac cavity data, thus completing the data preprocessing stage.

[0120] In the second part, orientation adjustment, the first high-resolution label image is used as input, and three orientation recognition and adjustment operations are performed sequentially: First, the operation of recognizing and adjusting the direction from the apex of the heart to the base of the heart is performed, so that the direction from the apex of the first high-resolution label image to the base of the heart is aligned with the positive z-axis of the preset coordinate system; then, the operation of recognizing and adjusting the direction from the left ventricle to the right ventricle is performed, so that the direction from the left ventricle to the right ventricle in the first high-resolution label image is aligned with the negative x-axis of the preset coordinate system; finally, the operation of recognizing and adjusting the direction from the posterior wall to the anterior wall is performed, so that the direction from the posterior wall to the anterior wall in the first high-resolution label image is aligned with the negative y-axis of the preset coordinate system. After the orientation adjustment is completed, the second high-resolution label image is obtained.

[0121] In the third part: mapping, the second high-resolution label image and the heart finite element mesh model are processed simultaneously: For the heart finite element mesh model, the operation of determining the centroid of the tetrahedral mesh is first performed to obtain the centroid of the mesh corresponding to each tetrahedral mesh; For the second high-resolution label image, the operation of determining the pixel coordinates of the centroid in the second high-resolution label image is performed, specifically by dividing the coordinate value of the centroid by the resolution of the second high-resolution label image and rounding down to obtain the corresponding pixel coordinates.

[0122] Subsequently, the pixel coordinates are evaluated: if the pixel coordinates belong to the myocardial tissue region (i.e., the label value corresponding to the pixel coordinates is not 0), then the corresponding first label value is obtained from the second high-resolution label image based on the pixel coordinates, and the first label value is assigned to the tetrahedral mesh; if the pixel coordinates belong to a non-myocardial region (i.e., the label value corresponding to the pixel coordinates is 0), then the non-zero pixels in the first to third order neighborhoods of the pixel coordinates are searched sequentially, and the second label value of the first non-zero pixel found is assigned to the tetrahedral mesh. After assigning label values ​​to all tetrahedral meshes, a personalized heart model with infarct regions can be generated.

[0123] In this embodiment, the reconstruction of a personalized heart model with infarct regions from CMR-LGE images is fully realized through the above three-part process: First, image segmentation and interpolation operations in the data preprocessing stage are used to obtain a first high-resolution labeled image containing ventricular cavities and infarct tissue, and a corresponding cardiac finite element mesh model, providing basic data for subsequent processes; then, through three orientation recognition and adjustment operations in the orientation adjustment stage, the anatomical orientation of the first high-resolution labeled image is aligned with the preset coordinate system to obtain a second high-resolution labeled image with higher spatial matching; finally, through mesh centroid calculation, pixel coordinate transformation, and label value assignment operations in the mapping stage, the infarct region information in the second high-resolution labeled image is accurately mapped to the tetrahedral mesh of the cardiac finite element mesh model, ultimately generating a personalized heart model with infarct regions that meets the requirements of anatomical structure and spatial position. This process not only ensures high data accuracy but also realizes spatial alignment and information transmission in each step, providing a reliable model foundation for subsequent electrophysiological simulations.

[0124] In one embodiment, for a better understanding of the image processing and model construction of the central ventricular cavity and infarcted tissue during the data preprocessing stage of this application, please refer to [link to relevant documentation]. Figure 11 , Figure 11 This is a schematic diagram illustrating the image segmentation, interpolation, and merging operations during the data preprocessing process provided in the embodiments of this application.

[0125] Specifically, Figure 11It consists of three parts, each corresponding to a different operational step in the data preprocessing stage: In the corresponding step of part (a), the CMR-LGE image is used as input. First, a segmentation operation is performed to extract the ventricular myocardial image from the CMR-LGE image. Then, an interpolation operation is performed on the ventricular myocardial image to generate a Label image. This Label image can serve as a specific representation of the high-resolution label image of the ventricular myocardial image in the data preprocessing stage.

[0126] In the corresponding step of part (b), based on the CMR-LGE image, the infarct tissue segmentation operation is first performed to obtain an image containing the infarct tissue region; then, the infarct tissue interpolation operation is performed on the image to obtain the interpolated infarct tissue image; then, the interpolated infarct tissue image is merged with the Label image obtained in (a) to generate a merged Label image, which is a specific representation of the first high-resolution label image in the data preprocessing stage.

[0127] In the corresponding step of part (c), a personalized ventricular model of the patient is constructed based on the image data obtained from (a) and (b). This model can serve as the basic geometric structure of the cardiac finite element mesh model in the data preprocessing stage. Subsequently, the final cardiac finite element mesh model can be generated through operations such as surface mesh generation and volume mesh generation.

[0128] In this embodiment, through Figure 11 The demonstrated workflow fully presents the core operations in the data preprocessing stage, from CMR-LGE images to the first high-resolution labeled image and the cardiac finite element mesh model: First, the label image corresponding to the ventricular myocardium is obtained through segmentation and interpolation. Then, the infarcted tissue is segmented, interpolated, and merged with the label image to obtain a merged label image (i.e., the first high-resolution labeled image). Finally, a personalized ventricular model (i.e., the basis of the cardiac finite element mesh model) is constructed based on these image data. This workflow realizes the transformation from raw medical images to high-precision labeled images and three-dimensional geometric models, providing accurate basic data support for subsequent orientation adjustments and infarct region mapping.

[0129] Figure 12 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 12As shown, the electronic device may include: a processor 1301, a communication interface 1302, a memory 1303, and a communication bus 1304, wherein the processor 1301, the communication interface 1302, and the memory 1303 communicate with each other via the communication bus 1304. The processor 1301 can call a computer program in the memory 1303 to execute steps of a personalized cardiac model reconstruction method with infarct regions, such as including: Preprocessing the patient's CMR-LGE images yields a first high-resolution labeled image containing ventricular chambers and infarcted tissue, and a cardiac finite element mesh model is constructed based on the patient's high-precision cardiac chamber data; The orientation of the first high-resolution label image is adjusted to obtain the second high-resolution label image; The infarct region in the second high-resolution labeled image is mapped onto the finite element mesh model of the heart to obtain a personalized heart model with the infarct region.

[0130] Furthermore, when the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0131] On the other hand, embodiments of this application also provide a non-transitory computer-readable storage medium storing a computer program. The computer program is used to cause a processor to execute the steps of the methods provided in the above embodiments, including, for example: Preprocessing the patient's CMR-LGE images yields a first high-resolution labeled image containing ventricular chambers and infarcted tissue, and a cardiac finite element mesh model is constructed based on the patient's high-precision cardiac chamber data; The orientation of the first high-resolution label image is adjusted to obtain the second high-resolution label image; The infarct region in the second high-resolution labeled image is mapped onto the finite element mesh model of the heart to obtain a personalized heart model with the infarct region.

[0132] Non-transitory computer-readable storage media can be any available medium or data storage device that can be accessed by a processor, including but not limited to magnetic storage (e.g., floppy disks, hard disks, magnetic tapes, magneto-optical disks (MOs), etc.), optical storage (e.g., CDs, DVDs, BDs, HVDs, etc.), and semiconductor storage (e.g., ROMs, EPROMs, EEPROMs, non-volatile memory (NAND flash), solid-state drives (SSDs)).

[0133] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0134] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0135] All actions involving the acquisition of signal information or data in this application were carried out in compliance with the relevant data protection laws and policies of the country where the application is located, and with the authorization granted by the owner of the relevant device. Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for reconstructing a personalized cardiac model with an infarct region, characterized in that, Includes the following steps: Preprocessing the patient's cardiac magnetic resonance delayed gadolinium-enhanced CMR-LGE images yields a first high-resolution labeled image containing ventricular cavities and infarcted tissue, and a cardiac finite element mesh model is constructed based on the patient's high-precision cardiac cavity data; The orientation of the first high-resolution label image is adjusted to obtain a second high-resolution label image; The infarct region in the second high-resolution labeled image is mapped onto the finite element mesh model of the heart to obtain a personalized heart model with the infarct region.

2. The method according to claim 1, characterized in that, The preprocessing of the patient's CMR-LGE images to obtain a first high-resolution labeled image containing ventricular chambers and infarct tissue includes: The patient's CMR-LGE images are segmented to obtain ventricular myocardial images and infarct tissue images, wherein the infarct tissue images include completely infarcted tissue regions and hemi-infarcted tissue regions; Interpolation and reconstruction are performed on the ventricular myocardial images to obtain high-resolution labeled images of the ventricular myocardium; Interpolation and reconstruction are performed on the infarcted tissue image to obtain a high-resolution labeled image of the infarcted tissue; The high-resolution labeled image of the ventricular myocardium is merged with the high-resolution labeled image of the infarcted tissue to generate a first high-resolution labeled image containing the ventricular cavity and the infarcted tissue.

3. The method according to claim 1, characterized in that, The construction of a cardiac finite element mesh model based on the patient's high-precision cardiac cavity data includes: The contour of the heart model is determined based on the high-precision heart cavity data. The heart model outline is subjected to surface mesh generation, mesh quality optimization, and volume mesh generation to generate a heart finite element mesh model; wherein, the heart finite element mesh model is saved as a node coordinate file that records node coordinates and a mesh topology file that records vertex indices of tetrahedral elements.

4. The method according to claim 1, characterized in that, The process of adjusting the orientation of the first high-resolution label image to obtain the second high-resolution label image includes: Identify the direction from the apex of the heart to the base of the heart in the first high-resolution label image, and adjust the orientation of the first high-resolution label image so that the direction from the apex of the heart to the base of the heart is changed to the positive z-axis direction; The left ventricle in the first high-resolution label image is identified as pointing towards the right ventricle. The orientation of the first high-resolution label image is then adjusted so that the left ventricle pointing towards the right ventricle is changed to point towards the negative x-axis direction. The direction in which the rear wall points to the front wall in the first high-resolution label image is identified, and the orientation of the first high-resolution label image is adjusted so that the direction in which the rear wall points to the front wall is changed to point to the negative y-axis direction. Based on the first high-resolution label image after orientation adjustment, a second high-resolution label image is generated that is consistent with the coordinate system of the heart finite element mesh model.

5. The method according to claim 4, characterized in that, The step of identifying the direction from the apex of the heart to the base of the heart in the first high-resolution label image, and then adjusting the orientation of the first high-resolution label image so that the direction from the apex of the heart to the base of the heart is changed to point in the positive z-axis direction, includes: Each two-dimensional slice along the z-axis in the first high-resolution label image is defined as a layer, with the layer with the largest z-coordinate value being the top layer and the layer with the smallest z-coordinate value being the bottom layer. The number of myocardial pixels contained in the top and bottom layers are counted respectively. The layer with a large number of myocardial pixels is identified as the bottom layer of the heart, and the layer with a small number of myocardial pixels is identified as the apical layer of the heart. Compare the z-coordinate values ​​of the basal layer and the apical layer to determine the direction from the apex to the basal layer; If the direction from the apex of the heart to the base of the heart indicates that the apex layer is located above the base of the heart, then the first high-resolution label image is flipped along the z-axis so that the direction from the apex of the heart to the base of the heart is changed to point in the positive z-axis direction.

6. The method according to claim 4, characterized in that, The step of identifying the left ventricle pointing towards the right ventricle in the first high-resolution label image and adjusting the orientation of the first high-resolution label image so that the left ventricle pointing towards the right ventricle points towards the negative x-axis includes: Based on connectivity rules, the non-myocardial regions in the first high-resolution labeled image are divided into the left ventricular cavity, the right ventricular cavity, and the image background; wherein, the circularity of the left ventricular cavity at the cardiac basal layer is closer to 1 than that of the right ventricular cavity at the cardiac basal layer. Determine the vector R between the centroid of the left ventricle and the centroid of the right ventricle; Compare the angle between the vector R and the negative x-axis to determine the direction in which the left ventricle points towards the right ventricle; If the direction in which the left ventricle points to the right ventricle is not pointing in the negative x-axis direction, then the first high-resolution label image is rotated and / or flipped so that the direction in which the left ventricle points to the right ventricle is changed to point in the negative x-axis direction.

7. The method according to claim 4, characterized in that, The step of identifying the rear wall pointing towards the front wall in the first high-resolution label image and adjusting the orientation of the first high-resolution label image so that the rear wall pointing towards the front wall points towards the negative y-axis includes: In the first high-resolution labeled image, the coordinates of the centroid of the right ventricle are obtained in the basal layer, the basal side 1 / 4 height layer, and the basal side 1 / 2 height layer, respectively. Analyze the trend of the y-coordinate change of the centroid of the right ventricle as it changes from the bottom of the heart to the apex, and determine the direction from the posterior wall to the anterior wall based on the trend of the y-coordinate change. If the direction from the rear wall to the front wall is not pointing in the negative y-axis direction, then the first high-resolution label image is flipped vertically to change the direction from the rear wall pointing to the front wall to pointing in the negative y-axis direction.

8. The method according to claim 1, characterized in that, The step of mapping the infarct region in the second high-resolution labeled image to the cardiac finite element mesh model to obtain a personalized cardiac model with the infarct region includes: For each tetrahedral mesh in the cardiac finite element mesh model, determine the centroid of the tetrahedral mesh; Based on the image resolution of the second high-resolution label image, the coordinates of the grid centroid are transformed and rounded down to obtain the pixel coordinates of the grid centroid in the second high-resolution label image. If the pixel coordinates belong to the myocardial tissue region, the corresponding first label value is obtained in the second high-resolution label image according to the pixel coordinates, and the first label value is assigned to the tetrahedral mesh. If the pixel coordinates belong to a non-myocardial region, search for non-zero pixels in the first to third order neighborhoods of the pixel coordinates in sequence, and assign the second label value of the first non-zero pixel found to the tetrahedral mesh. Based on all the tetrahedral meshes with assigned label values ​​in the aforementioned cardiac finite element mesh model, a personalized cardiac model with infarct regions is generated.

9. An electronic device comprising a processor and a memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the personalized cardiac model reconstruction method with infarct region as described in any one of claims 1 to 8.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the personalized cardiac model reconstruction method with infarct regions as described in any one of claims 1 to 8.

Images