Four-dimensional (4D) motion analysis of the left ventricle revealed six new motion phenotypes to identify Cardiovascular diseases Outcomes and genetic risks, according to researchers from Imperial College London, UK.
Conventional cardiac evaluation relies heavily on global volumetric measurements, which often fail to detect subtle or early abnormalities in cardiac function. Now, this large-scale analysis has demonstrated the potential of motion analysis to identify distinct cardiac outcomes.
Artificial intelligence analysis of 4D point cloud models
Participant samples were obtained from the UK Biobank, which recruited 500,000 participants aged 40 to 69 years, between 2006 and 2010.
The researchers used image processing artificial intelligence (computer vision) to analyze cardiac motion features in more than 20,000 eligible left ventricular samples. The image samples were converted into 4D point cloud models, capturing the full shape of the left ventricle and how the shape changed throughout the cardiac cycle.
Six groups (phenotypic groups) of movement traits were identified by disease prevalence, future cardiac events, and genetic risk scores. The researchers compared these clusters to population average (control) models to determine the diversity of movement patterns.
Six apparent groups of cardiac movement
Phenotypic groups 1 and 2 (PG1 and PG2) were identified as the lowest risk groups, with the lowest risk for obesity, diabetes, and hypertension, suggesting that healthy individuals with low risk for adverse outcomes can also be identified through movement analysis.
While PG3 did not show any strong distinct risk factors, PG4 was closely associated with metabolic heart diseases, such as diabetic cardiomyopathy, demonstrating that motion analysis can also identify metabolic risk.
PG5 and PG6 showed the most prevalence and incidence of heart disease, as well as the highest genetic risk scores. While PG6 showed a significant prevalence of incident cardiac arrest, PG5 was more associated with hypertension and dilated cardiomyopathy. PG6 was also the only group to show higher incidences of myocardial infarction and heart disease, suggesting that movement traits may also predict fatal outcomes.
Study limitations and clinical implications
Despite its promise, the study has limitations. The model focuses only on the left ventricle, excluding other heart chambers, challenging the broader applicability of virtual ensembles.
The UK Biobank samples were predominantly composed of participants of European origin, and older people and people living in socio-economically deprived areas were underrepresented. Additional research will be needed to understand movement analysis capabilities across diverse populations and social groups.
However, if validated in clinical settings, 4D cardiac motion analysis has the potential to redefine heart disease screening and management, promoting advances in personalized cardiac care.
reference
Shirati PR et al. Four-dimensional left ventricular motion clustering reveals cardiovascular phenotypes on a population scale. Science Rep 2026; doi: 10.1038/s41598-026-56151-y
Featured Image: Flow 37 on Abode Stock
