Automatic 3D+t four-chamber CMR quantification of the UK biobank: integrating imaging and non-imaging data priors at scale

Xia, Yan; Xiang, Chen; Ravikumar, Nishant; Kelly, Christopher; Attar, Rahman; Aung, Nay; Neubauer, Stefan; Petersen, Steffen E.; Frangi, Alejandro F.

doi:10.1016/j.media.2022.102498

Cited by 17 publications

(13 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The proposed method is outlined in Figure 1. It starts with the extraction of 3D meshes representing the LV from CMR images using an automatic segmentation method [12]. We then train several models with different metaparameters (network architecture, random seeds controlling weight initialization and dataset partitioning, and relative weight of the variational loss) to learn low-dimensional representations of the 3D meshes which capture anatomical variations using an encoder-decoder model.…”

Section: Methodsmentioning

confidence: 99%

“…Images were processed through a segmentation pipeline based on the deep learning approach in [12]. This network was trained on a dataset of manually segmented images at end-diastole and end-systole (4700 subjects), for which the manual 2D contours were registered to a 3D cardiac atlas encompassing the 4 chambers [40], producing a 3D mesh for each subject and time point.…”

Section: Additional Informationmentioning

confidence: 99%

“…The volume of the different chambers was calculated as explained in [12]. In particular, the volume for LV at end-diastole and end-systole was computed (LVEDV and LVESV).…”

Section: Additional Informationmentioning

confidence: 99%

“…First, we extracted a surface mesh representation of the anatomical structures. In particular, we studied 3D meshes representing the LV at end-diastole from CMR images of the UK Biobank database using an automatic deep-learningbased segmentation method [12]. We then learn a low-dimensional representation of the 3D meshes which captures anatomical variations using an encoder-decoder model.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Unsupervised ensemble-based phenotyping helps enhance the discoverability of genes related to heart morphology

Bonazzola¹,

Ferrante²,

Ravikumar³

et al. 2023

Preprint

View full text Add to dashboard Cite

Recent genome-wide association studies (GWAS) have been successful in identifying associations between genetic variants and simple cardiac parameters derived from cardiac magnetic resonance (CMR) images. However, the emergence of big databases including genetic data linked to CMR, facilitates investigation of more nuanced patterns of shape variability. Here, we propose a new framework for gene discovery entitled Unsupervised Phenotype Ensembles (UPE). UPE builds a redundant yet highly expressive representation by pooling a set of phenotypes learned in an unsupervised manner, using deep learning models trained with different hyperparameters. These phenotypes are then analyzed via (GWAS), retaining only highly confident and stable associations across the ensemble. We apply our approach to the UK Biobank database to extract left-ventricular (LV) geometric features from image-derived three-dimensional meshes. We demonstrate that our approach greatly improves the discoverability of genes influencing LV shape, identifying 11 loci with study-wide significance and 8 with suggestive significance. We argue that our approach would enable more extensive discovery of gene associations with image-derived phenotypes for other organs or image modalities.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Additional Informationmentioning

confidence: 99%

“…The volume of the different chambers was calculated as explained in [12]. In particular, the volume for LV at end-diastole and end-systole was computed (LVEDV and LVESV).…”

Section: Additional Informationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Unsupervised ensemble-based phenotyping helps enhance the discoverability of genes related to heart morphology

Bonazzola¹,

Ferrante²,

Ravikumar³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…generalised coherent point drift) developed previously by our group [47] to establish soft correspondences between atlas and subject-specific contours, and subsequent thine-platespline based warping of the atlas mesh to each subjects' contours. Further details on the registration process used to create the cohort of subject-specific meshes used in this study are available in [48]. We randomly split the available cohort of whole heart meshes from 2360 subjects into 422/59/1879 for training, validation and testing, respectively.…”

Section: A Datasetsmentioning

confidence: 99%

A Generative Shape Compositional Framework: Towards Representative Populations of Virtual Heart Chimaeras

Dou¹,

Virtanen²,

Ravikumar³

et al. 2022

Preprint

View full text Add to dashboard Cite

Generating virtual populations of anatomy that capture sufficient variability while remaining plausible is essential for conducting in-silico trials of medical devices. However, not all anatomical shapes of interest are always available for each individual in a population. Imaging examinations and modalities employed and available can vary across individuals. Different imaging modalities may have different fields of view, be sensitive to signals from different tissues/organs, or both. Hence, missing/partially-overlapping anatomical information is often available across individuals in a population. We introduce a generative shape model for complex anatomical structures, learnable from datasets of unpaired datasets, i.e. where each substructure in the complex comes from datasets that have missing or partially overlapping substructures from disjoint subjects of the same population. The proposed generative model can synthesise complete whole-complex shape assemblies coined virtual chimaeras, as opposed to natural human chimaeras. We applied this framework to build virtual chimaeras from databases of whole-heart shape assemblies that each contribute samples for heart substructures. Specifically, we propose a graph neural network-based generative shape compositional framework which comprises two components -a part-aware generative shape model which captures the variability in shape observed for each structure of interest in the training population; and a spatial composition network which assembles/composes the structures synthesised by the former into multi-part shape assemblies (viz. virtual chimaeras). We also propose a novel self-supervised learning scheme that enables the spatial composition network to be trained with partially overlapping data and weak labels. We trained and validated our approach using shapes of cardiac structures derived from cardiac magnetic resonance images available in the UK Biobank. When trained with complete and partially overlapping data, our approach significantly outperforms a PCAbased shape model (trained with complete data) in terms of generalisability and specificity. This demonstrates the superiority of the proposed approach as the synthesised cardiac virtual populations are more plausible and capture a greater degree of variability in shape than those generated by the PCA-based shape model.

show abstract

Risk factors for raised left ventricular filling pressure by cardiovascular magnetic resonance: Prognostic insights

Thomson,

Grafton‐Clarke,

Matthews

et al. 2024

ESC Heart Failure

View full text Add to dashboard Cite

BackgroundCardiovascular magnetic resonance (CMR) imaging shows promise in estimating pulmonary capillary wedge pressure (PCWP) non‐invasively. At the population level, the prognostic role of CMR‐modelled PCWP remains unknown. Furthermore, the relationship between CMR‐modelled PCWP and established risk factors for cardiovascular disease has not been well characterized.ObjectiveThe main aim of this study was to investigate the prognostic value of CMR‐modelled PCWP at the population level.MethodsEmploying data from the imaging substudy of the UK Biobank, a very large prospective population‐based cohort study, CMR‐modelled PCWP was calculated using a model incorporating left atrial volume, left ventricular mass and sex. Logistic regression explored the relationships between typical cardiovascular risk factors and raised CMR‐modelled PCWP (≥15 mmHg). Cox regression was used to examine the impact of typical risk factors and CMR‐modelled PCWP on heart failure (HF) and major adverse cardiovascular events (MACE).ResultsData from 39 163 participants were included in the study. Median age of all participants was 64 years (inter‐quartile range: 58 to 70), and 47% were males. Clinical characteristics independently associated with raised CMR‐modelled PCWP included hypertension [odds ratio (OR) 1.57, 95% confidence interval (CI) 1.44–1.70, P < 0.001], body mass index (BMI) [OR 1.57, 95% CI 1.52–1.62, per standard deviation (SD) increment, P < 0.001], male sex (OR 1.37, 95% CI 1.26–1.47, P < 0.001), age (OR 1.33, 95% CI 1.27–1.41, per decade increment, P < 0.001) and regular alcohol consumption (OR 1.10, 95% CI 1.02–1.19, P = 0.012). After adjusting for potential confounders, CMR‐modelled PCWP was independently associated with incident HF [hazard ratio (HR) 2.91, 95% CI 2.07–4.07, P < 0.001] and MACE (HR 1.48, 95% CI 1.16–1.89, P = 0.002).ConclusionsRaised CMR‐modelled PCWP is an independent risk factor for incident HF and MACE. CMR‐modelled PCWP should be incorporated into routine CMR reports to guide HF diagnosis and further management.

show abstract

Automatic 3D+t four-chamber CMR quantification of the UK biobank: integrating imaging and non-imaging data priors at scale

Cited by 17 publications

References 11 publications

Unsupervised ensemble-based phenotyping helps enhance the discoverability of genes related to heart morphology

Unsupervised ensemble-based phenotyping helps enhance the discoverability of genes related to heart morphology

A Generative Shape Compositional Framework: Towards Representative Populations of Virtual Heart Chimaeras

Risk factors for raised left ventricular filling pressure by cardiovascular magnetic resonance: Prognostic insights

Contact Info

Product

Resources

About