Methods and systems for cardiac imaging

Abstract

This invention provides a method for refining a surface mesh for mapping cardiac chambers. The method includes obtaining a surface mesh mapping cardiac chamber anatomy, wherein the mapped surface mesh includes a central region representing the cardiac chamber and an outer region representing peripheral cardiac structures connected to the cardiac chamber, and wherein the mapped surface mesh includes a first view of anatomical landmarks within the cardiac chamber, and image data of the cardiac chamber anatomy of the object is obtained. The central region of the mapped surface mesh is deformed based on a first segmentation algorithm configured according to one or more predetermined shape constraints, and the outer region of the mapped surface mesh is deformed based on a second segmentation algorithm configured according to the image data, thereby producing a deformed outer region. The deformed central region and the deformed outer region are then combined to generate a refined mapped surface mesh.

Description

Technical Field

[0001] This invention relates to the field of cardiac chamber imaging, and more particularly to the field of 3D cardiac chamber imaging. Background Technology

[0002] Electrophysiological mapping systems can be used to guide ablation, for example, in the pulmonary vein (PV). These systems typically create a real-time estimate of the atrial surface via electromapping; however, errors relative to fluorescence imaging may still result in an unrefined mapped surface.

[0003] Mapped surfaces typically consist of two components: the atrium and the pulmonary veins (PV), whose number and branching patterns vary considerably among patients. Because mapped surfaces are only approximations, errors may still exist compared to fluorescence imaging. For example, the diameter of the PV may be underestimated or overestimated within the mapped surface.

[0004] One possible approach to refining the mapped surface using in-process image data is to utilize intracardiac echo (ICE) data. ICE data is typically used during ablation procedures, for example, to check catheter wall contact or detect thrombi and blockages. To optimize the mapped surface using ICE data, model-based segmentation (MBS) can be used. However, this is challenging because predefined models cannot adequately capture many different PV configurations. This is due to variations in the number of PVs and branching patterns between patients. Averaged MBS models fail to capture this adequately because MBS meshes have a fixed set of vertices and connecting triangles.

[0005] One approach is to identify the most common PV patterns and create different average models for the most common variations, then attempt to select the best model for the patient. However, this method still cannot cover the vast number of possible variations.

[0006] On the other hand, the surface mesh of the calibration does indeed encompass all anatomical details. However, arbitrary meshes like the surface mesh of the calibration cannot be simply provided to the MBS algorithm for tuning, because many parameters of the MBS algorithm are specifically defined for the vertices and triangles of a given model, meaning the algorithm requires knowledge of the model topology.

[0007] In O. Ecabert et al.'s paper, "Segmentation of the heart and great vessels in CT images using a model-based adaptation framework," Medical Image Analysis, Oxford University Press, Oxford, GB, Vol. 15, No. 6, pp. 863-876, June 7, 2011 (XP028312871), a technique for segmenting the heart and attached great vessels in computed tomography images in a multi-stage process is proposed. First, the generalized Hough transform is used to detect the heart. Subsequently, the heart chambers are adjusted. In the final stage, segments of the great vessel structures are continuously activated and adjusted.

[0008] Therefore, there is a need for a method for mapping the anatomical structure of the cardiac chambers, which includes a variety of possible anatomical variations. Summary of the Invention

[0009] This invention is defined by the claims.

[0010] According to an example of one aspect of the invention, a method is provided for refining a surface mesh for mapping cardiac chambers (e.g., atria), the method comprising:

[0011] A surface grid for mapping cardiac chamber anatomy is obtained, wherein the mapped surface grid includes a central region representing the cardiac chamber and an outer region representing peripheral cardiac structures connected to the cardiac chamber, and wherein the mapped surface grid includes a first view of anatomical landmarks within the cardiac chamber;

[0012] Obtain image data of the cardiac cavity anatomy of the object;

[0013] The central region of the measured surface mesh is deformed based on a first segmentation algorithm configured according to one or more predetermined shape constraints;

[0014] The outer region of the measured surface mesh is deformed based on a second segmentation algorithm configured according to the image data, thereby generating a deformed outer region; and

[0015] The deformed central region and the deformed outer region are combined to generate a refined, calibrated surface mesh.

[0016] The deformation of the central region of the measured surface mesh based on the first segmentation algorithm configured according to one or more predetermined shape constraints includes:

[0017] The average mesh model is deformed based on the central region of the surface mesh to match the central region of the surface mesh, thereby generating a first deformed average mesh model.

[0018] Based on the image data of the cardiac cavity anatomy, the first deformed average mesh model is deformed to register the first deformed average mesh model to the image data of the cardiac cavity anatomy, thereby generating an adjusted mesh model; and

[0019] The central region of the measured surface mesh is deformed based on the deformation applied to the first deformed average mesh model, thereby generating the deformed central region.

[0020] This method provides a way to accurately adjust the surface mesh for mapping the cardiac cavity anatomy.

[0021] For example, if the heart chamber is an atrium, the central region of the atrium is similar across multiple objects and can therefore be processed using a generalized model; however, the external regions of the pulmonary vessels, which include the atria, are highly variable across objects. Therefore, by processing the central and external regions separately, the accuracy of the mapped surface mesh can be improved.

[0022] In one embodiment, deforming the central region of the calibrated surface mesh based on the first segmentation algorithm includes aligning the stages of the first segmentation algorithm with the calibrated surface mesh based on a first anatomical marker.

[0023] This method can improve the accuracy of deformation.

[0024] In an embodiment, deforming the central region of the measured surface mesh based on the first segmentation algorithm includes one or more of the following:

[0025] The points of the calibrated surface mesh are pushed onto the output surface of the first segmentation algorithm; and

[0026] Drag the points of the measured surface mesh onto the output surface of the first segmentation algorithm.

[0027] In an embodiment, deforming the central region of the calibrated surface mesh based on the deformation applied to the first deformed average mesh model includes:

[0028] The elements of the first deformed average mesh model are linked to the elements of the central region of the measured surface mesh, thereby forming element pairs;

[0029] The deformation applied to the first deformed average mesh model is applied to the elements of the central region of the measured surface mesh.

[0030] In this way, the deformation applied to the first deformed mesh model can be directly applied to the measured surface mesh.

[0031] In an embodiment, the deformation of the outer region of the surface mesh being measured is also based on the deformation applied to the central region of the surface mesh being measured.

[0032] In this way, the deformation force in the central region can be taken into account in the outer region.

[0033] In this embodiment, the steps of deforming the first deformed average mesh model and deforming the central region of the measured surface mesh are performed simultaneously.

[0034] In this embodiment, the steps of deforming the central region of the calibrated surface mesh and deforming the outer region of the calibrated surface mesh are performed simultaneously.

[0035] In one embodiment, the method further includes repeating each deformation step iteratively.

[0036] In this way, the surface mesh of the calibration can be refined multiple times, thereby obtaining a more accurate surface match of the calibration.

[0037] In one embodiment, the method further includes displaying a refined, calibrated surface mesh to the user.

[0038] In this embodiment, the surface mesh of the mapping is obtained by electrophysiological mapping.

[0039] In one embodiment, deformation of the average mesh model is performed by model-based segmentation.

[0040] In this embodiment, the image data includes intracardiac echo data.

[0041] According to an example of one aspect of the present invention, a computer program including computer program code units is provided, wherein when the computer program is run on a computer, the computer program code units are adapted to implement the method as described above.

[0042] According to an example of one aspect of the invention, a system for refining a surface mesh for mapping cardiac chambers is provided, comprising a processor, wherein the processor is adapted to:

[0043] A surface grid for mapping cardiac chamber anatomy is obtained, wherein the mapped surface grid includes a central region representing the cardiac chamber and an outer region representing peripheral cardiac structures connected to the cardiac chamber, and wherein the mapped surface grid includes a first view of anatomical landmarks within the cardiac chamber;

[0044] Obtain image data of the cardiac cavity anatomy of the object;

[0045] The central region of the measured surface mesh is deformed based on a first segmentation algorithm configured according to one or more predetermined shape constraints;

[0046] The outer region of the measured surface mesh is deformed based on a second segmentation algorithm configured according to the image data, thereby generating a deformed outer region; and

[0047] The deformed central region and the deformed outer region are combined to generate a refined, calibrated surface mesh.

[0048] In order to deform the central region of the measured surface mesh based on the first segmentation algorithm configured according to one or more predetermined shape constraints, the processor is further adapted to:

[0049] The average mesh model is deformed based on the central region of the measured surface mesh to match the central region of the measured surface mesh, thereby generating a first deformed average mesh model.

[0050] Based on the cardiac anatomical image data, the first deformed average mesh model is deformed to register the first deformed average mesh model to the image data of the cardiac anatomical structure, thereby generating an adjusted mesh model; and

[0051] The central region of the calibrated surface mesh is deformed based on the deformation applied to the first deformed average mesh model, thereby generating the deformed central region.

[0052] These and other aspects of the invention will become apparent and will be explained with reference to the embodiments described below. Attached Figure Description

[0053] To better understand the invention and to more clearly illustrate how it can be practiced, reference will now be made to the accompanying drawings by way of example only, wherein,

[0054] Figure 1 The method of the present invention is shown;

[0055] Figure 2 An example of a mesh model superimposed on a surface mesh of a calibration is shown;

[0056] Figure 3 An example of a first deformable mesh model superimposed on a calibrated surface mesh is shown; and

[0057] Figure 4 An example of surface mesh undergoing deformation is shown in the mesh model and the calibration. Detailed Implementation

[0058] The invention will be described with reference to the accompanying drawings.

[0059] It should be understood that while the detailed description and specific examples indicate exemplary embodiments of the devices, systems, and methods, they are intended for illustrative purposes only and are not intended to limit the scope of the invention. These and other features, aspects, and advantages of the devices, systems, and methods of the present invention will be better understood from the following description, the appended claims, and the accompanying drawings. It should be understood that the drawings are schematic only and are not drawn to scale. It should also be understood that the same reference numerals are used throughout the drawings to denote the same or similar parts.

[0060] This invention provides a method for refining a surface mesh for mapping cardiac chambers. The method includes obtaining a surface mesh of mapped cardiac chamber anatomy, wherein the mapped surface mesh includes a central region representing the cardiac chamber and an outer region representing peripheral cardiac structures connected to the cardiac chamber, and wherein the mapped surface mesh includes a first view of anatomical landmarks within the cardiac chamber, and obtaining image data of the cardiac chamber anatomy of the object.

[0061] The central region of the calibrated surface mesh is deformed based on a first segmentation algorithm configured according to one or more predetermined shape constraints, and the outer region of the calibrated surface mesh is deformed based on a second segmentation algorithm configured according to image data, thereby producing a deformed outer region. The deformed central region and the deformed outer region are then combined to generate a refined calibrated surface mesh.

[0062] In other words, the problem of obtaining a refined, mapped surface mesh of the heart chambers can be divided into two regions: the central region of the heart chambers, which typically has low anatomical variation, in which a conventional MBS mesh model can operate (the mesh model does not necessarily include accessory cardiac structures, such as blood vessels entering or leaving the heart chambers); and the outer region, which typically includes any accessory cardiac structures and has high anatomical variation.

[0063] This produces a refined, calibrated surface mesh based on trained and easily understood segmentation of ICE images in the large central region of the cardiac chamber, which is also applicable to images in variable external regions of the cardiac anatomy, including, for example, the PV.

[0064] For illustrative purposes, see references. Figures 1 to 4 The following description describes the implementation of the method of the present invention when the cardiac chamber is an atrium (more specifically, the left atrium). However, it should be noted that the method described herein can be applied to any cardiac chamber, such as: the left atrium; the right atrium; the left ventricle; and the right ventricle.

[0065] Figure 1 A method 100 for refining the surface grid of atrial mapping is shown.

[0066] The method begins in step 110 by obtaining a surface grid of atrial anatomy, wherein the surface grid includes a central region representing the atrial body and an outer region representing the pulmonary vessels connected to the atrial body, and wherein the surface grid includes a first view of anatomical landmarks within the atrial body.

[0067] The anatomical landmark can be any anatomical feature within the atrium.

[0068] As described above, the surface grid of the mapping can be obtained through electrophysiological mapping. For example, electrophysiological mapping can be performed using a catheter introduced through a vein or artery of the subject to measure electrical activity from within the heart.

[0069] In step 120, image data of the atrial anatomy of the subject is obtained. The image data may include intracardiac echo data or any other suitable image data type, such as transesophageal echo (TEE), transthoracic echo (TTE), or 3D X-ray angiography.

[0070] In step 130, the central region of the calibrated surface mesh is deformed based on a first segmentation algorithm configured according to one or more predetermined shape constraints.

[0071] The first segmentation algorithm can be any suitable segmentation algorithm that has been trained on a known shape of the central region of the heart cavity. For example, the first segmentation algorithm can be a deep learning-based segmentation algorithm that includes one or more shape constraints based on the region of interest (i.e., the central region of the heart cavity).

[0072] The following provides a detailed example of the implementation of the first segmentation algorithm based on Model-Based Segmentation (MBS).

[0073] Deforming the central region of the calibrated surface mesh may include one or more of the following: pushing the points of the calibrated surface mesh onto the output surface of the first segmentation algorithm; or pulling the points of the calibrated surface mesh onto the output surface of the first segmentation algorithm.

[0074] In the example, the first segmentation algorithm may include a voxel classifier that labels all image voxels belonging to the left atrium (LA) body. The output surface of the first segmentation algorithm will be the boundary between labeled and unlabeled voxels in the image. Alternatively, the output of the first segmentation algorithm may be a probability map, where each voxel is assigned a probability of belonging to the LA body. In this case, the surface may be an isosurface in the probability map, for example, including voxels with specific values ​​or probability ranges, such as between 0.45 and 0.55.

[0075] In step 160, the stage of the first segmentation algorithm can be aligned to the surface mesh of the survey based on the first anatomical landmark. The stage of the first segmentation algorithm can be any stage of the segmentation algorithm, such as the initial segmentation mesh, intermediate segmentation results, or final segmentation output.

[0076] For example, the average grid model of the cardiac chambers included in the first segmentation algorithm can also include views of anatomical landmarks. In this case, the average grid model and the mapped surface grid can be aligned, wherein the alignment is based solely on the two grids, for example, on the anatomical landmarks present in the two grids.

[0077] The deformation of the central region of the surface mesh, as measured by the MBS algorithm, will now be described.

[0078] In step 170, the average mesh model is deformed based on the central region of the calibrated surface mesh to generate a first deformed average mesh model. The deformation of the average mesh model can be performed based on model segmentation.

[0079] In other words, the average mesh model is deformed to match the central region of the measured surface model. For example, based on the selected model degrees of freedom, the distance from the average mesh model to the measured surface can be minimized in the central region shared by the two meshes. This deformation can be achieved using the point-to-plane iterative nearest point (ICP) method, for example, by registering points of the average mesh model to neighboring points on the surface of the measured surface mesh.

[0080] Alternatively, deformation can be achieved by projecting rays from triangles in the average mesh model and searching for target points on the measured surface mesh using the intersections with the target points of the surface mesh to deform the mesh model. The mesh model can then be deformed while taking into account other model properties such as shape constraints and internal energy.

[0081] Furthermore, the correspondence between the surface of the calibrated surface mesh and the mesh model can be established in the overlapping central region, such that for each element in the calibrated surface mesh, a corresponding element in the average mesh model is found (e.g., the distance between corresponding elements is limited to below a given threshold).

[0082] In step 180, the first deformed average grid model is deformed based on image data of the atrial anatomy to generate an adjusted grid model. Deforming the first deformed average grid model includes registering the first deformed average grid model to the image data of the atrial anatomy, for example, through generalized Hough transform, deep learning-based marker detection, or any other suitable method for registering the first deformed grid model to the image data.

[0083] As a clarification, up to step 180, no image data relating to atrial anatomy was used in the deformation of the central region of the mapped surface grid.

[0084] In step 190, the central region of the measured surface mesh is deformed based on the deformation applied to the first deformed average mesh model, thereby generating a deformed central region.

[0085] Deforming the central region of the measured surface mesh based on the deformation applied to the first deformed average mesh model may include linking elements of the first deformed mesh model to elements of the central region of the measured surface mesh to form element pairs, and applying the deformation applied to the first deformed mesh model to the element pairs.

[0086] In other words, the deformations of corresponding regions in the mesh model and the measured surface mesh can be linked together. In this way, the MBS boundary detector can create forces that actively deform the central regions of both the average mesh model and the measured surface mesh. This allows for the simultaneous execution of the steps of deforming the first deformed mesh model and deforming the central regions of the measured surface mesh.

[0087] In step 140, the outer region of the measured surface mesh is deformed based on a second segmentation algorithm configured according to the image data, thereby generating a deformed outer region. In other words, the outer region can be actively deformed based solely on image data acquired from an individual object.

[0088] Deformation of the outer region of the calibrated surface mesh can also be based on deformation applied to the central region of the calibrated surface mesh. In other words, the outer region of the calibrated surface mesh, which may include a structure such as a PV, can be passively moved by a deformation force applied to the central region. Furthermore, the steps of deforming the central region of the calibrated surface mesh and deforming the outer region of the calibrated surface mesh are performed simultaneously. In particular, the passive deformation force can be applied to the outer region at the same time as the active deformation force is applied to the central region.

[0089] Using ICE image data, the outer region of the calibrated surface mesh is deformed. This can occur either after or directly simultaneously with the deformation of the central body.

[0090] For the current surface mesh topology, the boundary detector for the PV in the outer region can be determined based on the specific circumstances. The boundary detector can be determined in several ways.

[0091] For example, boundary conditions can be transferred from a portion of the central region adjacent to the portion of the external region under discussion (e.g., the region around the PV ostium). In another example, boundary detection can be performed using a pre-computed boundary detector for each specific pulmonary vein (e.g., assuming all branches of the upper left PV (LUPV) and the central LUPV are assigned a pre-computed boundary detector for the LUPV, where branches with unknown labels might be assigned an average backtracking boundary detector).

[0092] Alternatively, regions of interest are identified, where deep learning-based algorithms are applied to detect structures of interest. For example, to feed this information into the calibrated surface mesh, the segmentation mask can be converted into a grayscale image, and the grayscale features can be used to adjust the mesh model. Furthermore, the boundary detector of the calibrated surface mesh can be refined multiple times after the first iteration of deformation.

[0093] When deforming the outer region within the MBS framework, no statistically average shape is available for the region due to the high variability in the anatomical structure and the number of mesh triangles. Therefore, during deformation of the outer region, the surveyed surface mesh can be used as an average shape model for the outer region to penalize unreasonable deformations of the original surveyed shape. Further tuning parameters, such as allowed affine transformations, can be defined in a manner similar to that of the boundary detector, for example, based on the nearest region of the average mesh model, or, if necessary, based on a predefined shape of the anatomical region.

[0094] In step 150, the deformed central region and the deformed outer region are then combined to generate a refined, calibrated surface mesh.

[0095] The above deformation steps can be repeated iteratively to further refine the measured surface mesh.

[0096] After the deformation process is complete, the output of the first segmentation algorithm, such as the MBS mesh model, may be discarded. The result is a refined, calibrated surface mesh, based on well-trained and easily understood segmentation of the ICE image in the large central region, but also applicable to images in variable external regions. Furthermore, the refined, calibrated surface mesh matches the ICE image and can be displayed as an overlay of image data in a way that is clear to the operator.

[0097] In other words, the refined surface mesh can be displayed to the user as an overlay of image data or as a separate surface mesh.

[0098] Figure 2 An example 200 of the average grid model 210 is shown as a solid line, overlaid on the surface grid 220 of the left atrium as a dashed line.

[0099] As described above, the average grid model 210 controls the deformation of the central region 230 (atrial body), while the outer region 240 (including...) Figure 1 The deformation of the left PV 250 and right PV 260 in the example shown is based on the surface mesh 220 in the calibration.

[0100] Figure 3 An example 300 of a first deformed average mesh model 310 superimposed on a surface mesh 320 of a calibration is shown. After the average mesh model 310 has been deformed to match the surface mesh of the calibration, corresponding / proximity elements 330 can be identified and linked together so that they move together when further deformation is applied.

[0101] Figure 4 An example 400 of the mesh model 410 is shown, and a surface mesh 420 of the calibration is deformed based on ICE image data captured from the field of view 430 of the ICE conduit.

[0102] Within the field of view 430, the segmentation algorithm can search for image boundaries 440 on the average grid model. For the calibrated surface grid 420, an auxiliary boundary detector can be created after grid registration to detect further image boundaries 450 in the height-variable region of the PV that the first segmentation algorithm cannot accurately describe. These forces can be used in successive steps, either first deforming only the MBS grid model and passively moving the linked calibrated surface grid elements, or combining both forces in parallel to update the calibrated surface grid.

[0103] Those skilled in the art, through studying the accompanying drawings, the disclosure, and the claims, will be able to understand and implement variations of the disclosed embodiments when practicing the claimed invention. In the claims, the word "comprising" does not exclude other elements or steps, and the words "a" or "an" do not exclude a plurality.

[0104] A single processor or other unit can perform the functions of several items described in the claims.

[0105] Although specific measures are described in different dependent claims, this does not imply that combinations of these measures cannot be used advantageously.

[0106] Computer programs can be stored / distributed on suitable media such as optical storage media or solid-state media that are provided together with or as part of other hardware, but they can also be distributed in other forms such as via the Internet or other wired or wireless telecommunications systems.

[0107] If the term “suitable” is used in the claims or description, it should be noted that the term “suitable” is intended to be equivalent to the term “configured as”.

[0108] Any reference numerals in the claims should not be construed as limiting the scope.

Claims

1. A method (100) for refining a mapped surface mesh of a cardiac chamber, the method comprising: obtaining (110) a mapped surface mesh of the cardiac chamber anatomy, wherein the mapped surface mesh includes a central region representing a cardiac chamber and an outer region representing a peripheral cardiac structure connected to the cardiac chamber, and wherein the mapped surface mesh includes a first view of an anatomical landmark within the cardiac chamber; obtaining (120) image data of the cardiac chamber anatomy of a subject; deforming (130) the central region of the mapped surface mesh based on a first segmentation algorithm configured according to one or more predetermined shape constraints; deforming (140) the outer region of the mapped surface mesh based on a second segmentation algorithm configured according to the image data, thereby generating a deformed outer region; and combining (150) the deformed central region and the deformed outer region, thereby generating a refined mapped surface mesh, wherein deforming the central region of the mapped surface mesh based on the first segmentation algorithm configured according to one or more predetermined shape constraints comprises: deforming (170) an average mesh model based on the central region of the mapped surface mesh to match the central region of the mapped surface mesh, thereby generating a first deformed average mesh model; deforming (180) the first deformed average mesh model based on the image data of the cardiac chamber anatomy to register the first deformed average mesh model to the image data of the cardiac chamber anatomy, thereby generating an adjusted mesh model; and deforming (190) the central region of the mapped surface mesh based on the deformation applied to the first deformed average mesh model, thereby generating a deformed central region.

2. The method (100) of claim 1, wherein Deforming the central region of the mapped surface mesh based on a first segmentation algorithm comprises aligning (160) a stage of the first segmentation algorithm with the mapped surface mesh based on a first anatomical landmark.

3. The method (100) according to any one of claims 1 to 2, wherein Deforming the central region of the mapped surface mesh based on a first segmentation algorithm comprises one or more of: pushing points of the mapped surface mesh to an output surface of the first segmentation algorithm; and pulling points of the mapped surface mesh to an output surface of the first segmentation algorithm.

4. The method (100) according to any one of claims 1 to 3, wherein Deforming the central region of the mapped surface mesh based on the deformation applied to the first deformed average mesh model comprises: linking elements of the first deformed average mesh model to elements of the central region of the mapped surface mesh, thereby forming element pairs; applying the deformation applied to the first deformed average mesh model to the elements of the central region of the mapped surface mesh.

5. The method (100) according to any one of claims 1 to 4, wherein Deforming the outer region of the mapped surface mesh is further based on the deformation applied to the central region of the mapped surface mesh.

6. The method (100) according to any one of claims 1 to 5, wherein The steps of deforming the first deformed average mesh model and deforming the central region of the mapped surface mesh are performed simultaneously.

7. The method (100) according to any one of claims 1 to 6, wherein The steps of deforming the central region of the mapped surface mesh and deforming the outer region of the mapped surface mesh are performed simultaneously.

8. The method (100) according to any one of claims 1 to 7, wherein The method further comprises repeating each of the deforming steps in an iterative manner.

9. The method (100) according to any one of claims 1 to 8, wherein The method further comprises displaying the refined mapped surface mesh to a user.

10. The method (100) of any of claims 1 to 9, the mapped surface mesh being obtained by electro-physiological mapping.

11. The method (100) according to any one of claims 1 to 10, wherein Deforming the average mesh model is performed by model-based segmentation.

12. The method (100) according to any one of claims 1 to 11, wherein The image data comprises intracardiac echo data.

13. A computer program comprising computer program code means adapted to implement the method of any of claims 1 to 12 when said computer program is run on a computer.

14. A system for refining a surface mesh of a cardiac chamber's mapping, comprising a processor, wherein, The processor is adapted to: obtain a mapped surface mesh of a cardiac chamber anatomy, wherein the mapped surface mesh comprises a central region representing a cardiac chamber and an outer region representing a peripheral cardiac structure connected to the cardiac chamber, and wherein the mapped surface mesh comprises a first view of an anatomical landmark within the cardiac chamber; obtain image data of the cardiac chamber anatomy of a subject; deform the central region of the mapped surface mesh based on a first segmentation algorithm configured according to one or more predetermined shape constraints; deform the outer region of the mapped surface mesh based on a second segmentation algorithm configured according to the image data, thereby generating a deformed outer region; and combine the deformed central region and the deformed outer region, thereby generating a refined mapped surface mesh, wherein, to deform the central region of the mapped surface mesh based on the first segmentation algorithm configured according to one or more predetermined shape constraints, the processor is further adapted to: deform an average mesh model based on the central region of the mapped surface mesh to match the central region of the mapped surface mesh, thereby generating a first deformed average mesh model; deform the first deformed average mesh model based on the image data of the cardiac chamber anatomy to register the first deformed average mesh model to the image data of the cardiac chamber anatomy, thereby generating an adjusted mesh model; and deform the central region of the mapped surface mesh based on the deformation applied to the first deformed average mesh model, thereby generating a deformed central region. The processor is adapted to: obtain a mapped surface mesh of a cardiac chamber anatomy, wherein the mapped surface mesh comprises a central region representing a cardiac chamber and an outer region representing a peripheral cardiac structure connected to the cardiac chamber, and wherein the mapped surface mesh comprises a first view of an anatomical landmark within the cardiac chamber; obtain image data of the cardiac chamber anatomy of a subject; deform the central region of the mapped surface mesh based on a first segmentation algorithm configured according to one or more predetermined shape constraints; deform the outer region of the mapped surface mesh based on a second segmentation algorithm configured according to the image data, thereby generating a deformed outer region; and combine the deformed central region and the deformed outer region, thereby generating a refined mapped surface mesh, wherein, to deform the central region of the mapped surface mesh based on the first segmentation algorithm configured according to one or more predetermined shape constraints, the processor is further adapted to: deform an average mesh model based on the central region of the mapped surface mesh to match the central region of the mapped surface mesh, thereby generating a first deformed average mesh model; deform the first deformed average mesh model based on the image data of the cardiac chamber anatomy to register the first deformed average mesh model to the image data of the cardiac chamber anatomy, thereby generating an adjusted mesh model; and deform the central region of the mapped surface mesh based on the deformation applied to the first deformed average mesh model, thereby generating a deformed central region.

Images