Pediatric Right Ventricular Assessment

Congenital and acquired heart disease affects about 1% of children worldwide, and accurate assessment of right ventricular (RV) function is challenging due to the ventricle’s complex geometry and variability in pediatric patients. Using nearly 25,000 echocardiograms from almost 4,000 children across four international centers, the authors developed a video-based deep learning model that performs automated RV segmentation, beat-by-beat fractional area change estimation, disease classification, and exploratory left ventricular ejection fraction prediction with high accuracy. The framework enables expert-level, real-time and consistent ventricular assessment, reducing manual workload and supporting earlier diagnosis and intervention, particularly in resource-limited settings.