Children’s Hospital of Philadelphia uses open‑source NVIDIA AI to model pediatric hearts in seconds

The AI‑driven pipeline relies on MONAI Label and NVIDIA’s Auto3DSeg to train segmentation networks, delivering human‑level quality instantly. CHOP is extending the approach with the open‑source physics engine Newton, built on NVIDIA’s Warp framework, to simulate how cardiac devices will behave inside a child’s unique anatomy. Early results include a successful repair of a complex ventricular septal defect after previous failed attempts. The effort is part of a broader open‑source ecosystem—including SlicerHeart, OpenUSD, and collaborations with Stanford and Boston Children’s—aimed at providing real‑time, high‑fidelity simulations for congenital heart disease treatment.

By leveraging industrial‑scale GPU acceleration and community‑driven software, CHOP demonstrates how open‑source AI can overcome the economic barriers of rare, heterogeneous medical conditions, turning research prototypes into standard‑of‑care tools.