Deep-learning analysis of 3D microarchitectural remodeling in hypertrophic cardiomyopathy

Area of Science:

• Cardiovascular Research
• Medical Imaging Analysis
• Genetics

Background:

• Hypertrophic cardiomyopathy (HCM) is a genetic heart disease causing unexplained cardiac wall thickening and sudden death.
• The three-dimensional cardiac tissue organization in left ventricular hypertrophy is not well understood.

Purpose of the Study:

• To characterize cardiac microarchitecture in hypertrophic cardiomyopathy (HCM) using a novel deep-learning approach.
• To identify genotype-specific differences in cardiac tissue organization in HCM patients.

Main Methods:

• Development of CaMVIA-3D, a deep-learning pipeline for volumetric imaging and analysis of cardiac microarchitecture.
• Analysis of cardiac tissues from HCM patients and a pig HCM model.
• Integration of transcriptomic and morphologic data.

Main Results:

• HCM hearts showed genotype-specific differences in cardiomyocyte and extracellular volume.
• Pathogenic variants were associated with greater cellular hypertrophy and disarray, while variant-negative cases showed fibrosis.
• Early-onset fibrosis preceded cardiomyocyte hypertrophy in a pig HCM model.
• Identified genes linked to cellular and extracellular remodeling.

Conclusions:

• The study defines genotype-specific microstructural differences in hypertrophic cardiomyopathy (HCM).
• Findings provide insights for improved diagnostics and targeted therapies for HCM.
• CaMVIA-3D pipeline offers a novel method for cardiac microarchitecture characterization.