4Dheart
Research type
Research Study
Full title
Imaging the heart in 4D using advanced MRI
IRAS ID
281924
Contact name
Declan O'Regan
Contact email
Sponsor organisation
Imperial College London
Duration of Study in the UK
10 years, 0 months, 1 days
Research summary
Cardiomyopathy is a leading cause of sudden death in the UK. Cardiomyopathies represent a spectrum of intrinsic heart muscle diseases arising from a complex interaction between genetic and environmental factors associated with differing adaptations of ventricular contractility and geometry frequently leading to functional decompensation or sudden death. Current tools for characterising dysfunction and predicting adverse events rely on crude explanatory variables (e.g. left ventricular ejection fraction) that are insensitive to the complex pathophysiology of heart disease, and fail to make use of the rich information that is inherent in routine cardiac imaging (e.g. cardiac magnetic resonance imaging).
Previous work by our group has demonstrated it is possible to generate accurate and reproducible 3D segmentations in pulmonary hypertension (PH) patients using a fully convolutional network, and utilising cardiac atlases derived from heathy volunteers to create smooth 3D renderings of frame-by-frame cardiac motion. These renderings then serve as input to a supervised machine learning prediction network (“4Dsurvival”) that can predict time-to-events (e.g. death) more accurately than conventional risk parameters.
There is a need to more optimally predict risk in cardiomyopathy patients, a clinically heterogenous group. At present we are lacking cardiomyopathy-specific reference models, or “atlases”, to develop accurate 3D motion-based survival models for these conditions, and utilising atlases derived from healthy volunteers would be imprecise in terms of modelling cardiac anatomy in cardiomyopathy. Additionally, advances in MRI allow for the measurement of non-laminar flow in all three spatial directions across the cardiac cycle (4D-Flow), which can further enhance the atlases generated.
This approach could be transformative in delivering patient-specific tools for predicting time-to-events, and hence initiating therapy earlier, identifying causative mechanisms, and accelerating the discovery of molecular targets amenable to therapy.
REC name
London - Bromley Research Ethics Committee
REC reference
21/PR/1710
Date of REC Opinion
3 Feb 2022
REC opinion
Further Information Favourable Opinion