Left ventricular anatomy in obstructive hypertrophic cardiomyopathy: beyond basal septal hypertrophy

Supporting data for the study <strong>"Left ventricular anatomy in obstructive hypertrophic cardiomyopathy: beyond basal septal hypertrophy " </strong>published in the European Heart Journal - Cardiovascular Imaging <br> See: https://doi.org/10.1093/ehjci/jeac233<br> <br> <strong>Data summary:</strong> Collection of 2400 computational meshes of the anatomy of the left ventricle from individuals with hypertrophic cardiomyopathy (HCM) recruited as part of the Hypertrophic Cardiomyopathy Registry (HCMR) study (https://hcmregistry.org/). An external evaluation cohort of 110 HCM individuals with data collected at the Oxford Centre for Clinical Magnetic Resonance Research (OCMR) is also provided.<br> <br> <strong>Statistical shape model (SSM) shape analysis: </strong>Endocardial and epicardial contours were delineated at end-diastole (ED) for each slice of the LV short-axis magnetic resonance cine stack. A 3D model of each patient was automatically fitted by warping a template mesh with 64 elements and 130 nodes, as described in (1, 2). The resulting meshes are described by 3120 nodal variables or degrees of freedom. A statistical shape model was built with a principal component analysis (PCA), capturing relevant shape variations in the population.<br> <br> <strong>LV phenotype discriminant of obstruction: finding the LVOTO signature: </strong> Information of shape is decomposed by the SSM into several PCA modes, and the task is to find the linear combination that best discriminates between different groups (obstructive/non-obstructive at rest and/or stress). Following the methodology described by Varela et al. (27), a Fisher Linear Discriminant Analysis (LDA) was used to find the optimal anatomical discriminant mode (LDA mode). Two LDA models were built. The LDArest model compared LV shape phenotypes between groups R- (non-obstructive HCM cases at rest) and R+ (obstructive HCM cases at rest). The LDA∆stress aimed to identify the extra cases that show obstruction at stress but not at rest, and was built to discriminate groups R-S- (non-obstructive HCM cases at rest and stress) and R-S+ (obstructive HCM cases at stress but not at rest). LV anatomies for each subject were then then characterized by a single coefficient along each axis (i.e., Z-score). <br> A web service (Cardiac Atlas Project, www.cardiacatlas.org) is freely available to compute the shape score (Z-score) along each axis from MRI SAX segmentations.<br> <br> <strong>Shared data contents:</strong> <br> <strong>- Figure 1: Statistical shape analysis pipeline. </strong>From the endocardial and epicardial contours, automatic 3D meshes are generated for each patient. A statistical shape model is then built using a Principal Component Analysis (PCA), which results in a set of anatomical modes (linear directions of anatomical change from the average shape) that capture the most common changes in the cohort. A Fisher Linear Discriminant Analysis (LDA) is used to find the linear combination of anatomical modes that best discriminates between patient groups to explore their characteristic LV phenotypes. <br> <strong>- Figure 2:</strong> <strong>HCMR signature of LVOTO.</strong> Set of linear discriminant axes that characterize the role of LV shape in obstructive HCM (oHCM). Boxplots represent the Z-score distributions along each axis for each group. Orange cross and 3D shapes show -3SD from the mean shape (blue cross); purple cross and shape, +3SD. A) Linear discriminant axis between non-obstructive HCM cases at rest (blue – R-) and obstructive HCM (oHCM) cases at rest (green – R+) - LDArest. B) Linear discriminant axis between non-obstructive HCM cases at rest and stress (red – R-S-) and obstructive HCM cases at stress but not at rest (dark blue – R-S+) - LDA∆stress. Genotype negative (yellow – G-) and genotype positive (black – G+) scores are projected along the LDArest and LDA∆stress. C) Linear discriminant axis between genotype negative (yellow – G-) and genotype positive (black – G+) cases - LDAgen. Black arch: LVOT location (note Bull’s eye plot (bottom) is viewed from the apex, model views are from lateral (top-left) and base (top-right)). Red dot: septal wall location. A: Anterior; I: Inferior; L: Lateral; S: Septal.<br> <strong>- Figure 3:</strong> <strong>LV shape and LVOTO. </strong>Impact of LV shape on vortex formation during early systole, which changes the angle of attack of blood flow with respect to the MV leaflets, increasing the systolic anterior motion of the MV and consequently, LVOTO. Orange shape represents -3SD from the average shape along the LDArest axis (extreme non-obstructive HCM phenotype at rest). Purple shape, +3SD (obstructive HCM phenotype at rest). Phenotypes resulting from the LDA∆stress are overlayed. An estimation of the left atrium, mitral valve, aorta, and aortic valve is shown with dotted lines. <br> <strong>- HCMR cohort:</strong> <strong>1. AtlasMeshData.zip</strong>, where, for each case: 1.1 CaseName.vtk: The mask of the left ventricular (LV) myocardium and right ventricular blood pool from the short axis magnetic resonance image at end-diastole. 1.2 Thickness.mat: binary mask information. 1.3 Folder with mesh-related files (MeshingLoD3): 1.3.1 CurvatureMetrics.mat: curvature metrics derived from the personalised 3D mesh. 1.3.2 GeometricalMetrics.mat: geometrical metrics derived from the personalised 3D mesh. 1.3.3 ThicknessMetrics.mat: thickness metrics derived from the personalised 3D mesh. 1.3.4 GeoReport.txt: Basic clinical geometrical metrics. 1.3.5 Mesh files (.exelem and .exnode): Cubic Hermite mesh that was personalised to the anatomy of the left ventricle. 1.3.6 Image_Initialization.jpg: Image of the overlay between the template LV mesh used for personalization and the mask of the LV myocardium. 1.3.7 Image_CaseName_mesh.jpg: Image of the overlay between the fitted mesh and the mask of the LV myocardium. 1.3.8 ImageDistances2ClosestPointCaseNamexxxxxxxxxx.png: Image of the distances between the LV mesh and the mask of the LV myocardium (xxxxxxxxxx is a digit encoding for the time when the image was generated). 1.3.9 PersonalizationReportCaseName.txt: File with metrics of mesh fitting accuracy and quality. 1.3.10 CaseNameThicknessMap.png: image of the Bull's eye map of the LV wall thickness.<br> <br> <strong>2. VTKmeshes</strong>: Collection of all LV meshes in VTK format, after correction of centre of mass and circumferential orientation. <br> <strong>3. ListCases_with_classification.xlsx</strong>: Spreadsheet with a list of cases and their classification as obstructive/non-obstructive HCM. <br> <br> <strong>- OCMR cohort:</strong> <strong>1. AtlasMeshData.zip</strong>, where, for each case: 1.1 CaseName.gipl: The mask of the left ventricular (LV) myocardium and right ventricular blood pool from the short axis magnetic resonance image at end-diastole. <br> 1.2 Thickness.mat: binary mask information. <br> 1.3 Folder with mesh-related files (MeshingHCMR3): 1.3.1 CurvatureMetrics.mat: curvature metrics derived from the personalised 3D mesh. 1.3.2 GeometricalMetrics.mat: geometrical metrics derived from the personalised 3D mesh. 1.3.3 ThicknessMetrics.mat: thickness metrics derived from the personalised 3D mesh. 1.3.4 GeoReport.txt: Basic clinical geometrical metrics. 1.3.5 Mesh files (.exelem and .exnode): Cubic Hermite mesh that was personalised to the anatomy of the left ventricle. 1.3.6 Image_Initialization.jpg: Image of the overlay between the template LV mesh used for personalization and the mask of the LV myocardium. 1.3.7 Image_CaseName_mesh.jpg: Image of the overlay between the fitted mesh and the mask of the LV myocardium. 1.3.8 ImageDistances2ClosestPointCaseNamexxxxxxxxxx.png: Image of the distances between the LV mesh and the mask of the LV myocardium (xxxxxxxxxx is a digit encoding for the time when the image was generated). 1.3.9 PersonalizationReportCaseName.txt: File with metrics of mesh fitting accuracy and quality. 1.3.10 CaseNameThicknessMap.png: image of the Bull's eye map of the LV wall thickness.<br> <br> <strong>2. VTKmeshes</strong>: Collection of all LV meshes in VTK format, after correction of centre of mass and circumferential orientation. <br>