Our Research

Our research program is dedicated to pioneering innovative applications of artificial intelligence to significantly improve the early detection of Congenital Heart Defects (CHDs) in prenatal screenings. We focus on developing robust, AI-driven solutions that leverage advanced deep learning techniques, particularly unsupervised and generative models, to accurately characterize "normal" fetal cardiac anatomy from ultrasound data. This foundational work enables the identification of subtle structural or functional deviations, allowing for the precise flagging of potential anomalies for expert clinical review. Our technical roadmap is centered on establishing scalable and reliable anomaly detection pipelines that augment current diagnostic capabilities, aiming to enhance precision, reduce diagnostic variability, and ultimately facilitate earlier interventions to improve outcomes for affected infants. This page serves as a central hub for our ongoing experiments and findings, providing detailed write-ups on our methodological approaches, performance benchmarks, and insights derived from each research phase.