Project Details

Improving Risk Stratification in Repaired Tetralogy of Fallot Using Radiomics and Deep Learning

Joshua Mayourian, MD, PhD

Summary

BACKGROUND: While children with repaired tetralogy of Fallot (rTOF) have excellent early survival, the growing adult rTOF population remains at increased risk for morbidity and premature mortality. GAP: While conventional cardiac magnetic resonance parameters lead to promising risk stratification in the vulnerable rTOF population, there remains room for improvement. The application of radiomics and deep learning algorithms to native CMR has shown promise in the field of cardiovascular risk stratification, but to our knowledge it has not yet been applied to rTOF. HYPOTHESIS: We hypothesize that applying radiomics and deep learning to native CMR images will improve all-cause mortality risk stratification in patients with rTOF compared to conventional models. METHODS: This is a single-center retrospective cohort study, using patients enrolled in the INDICATOR cohort at Boston Children’s Hospital. We aim to develop and validate a rTOF all-cause mortality risk stratification model that incorporates native CMR “radiomic signatures” and deep learning-based “cine risk score” predictors, which will then be benchmarked to conventional modeling techniques. RESULTS: Pending. IMPACT: Our findings may guide clinical decision-making and the design of future interventional trials in this population, while also providing insight into novel imaging biomarkers to risk stratify rTOF and potentially other forms of congenital heart disease.