CHOP Models Children’s Hearts with Open-Source NVIDIA AI Tools

Children’s Hospital of Philadelphia (CHOP) is running a cardiac modeling service built on MONAI, an open-source medical imaging framework cofounded by NVIDIA, that turns the CT scans, MRI and 3D ultrasound images a child’s care team already has into anatomically precise heart models, NVIDIA reported on September 15, 2026. NVIDIA said the service produces each model in seconds, replacing a workflow that once took a skilled researcher about four hours, making the models fast enough for routine clinical use.

About 1% of live births involve a congenital heart defect, and no two are alike, according to NVIDIA’s account. A child born with a hole between the heart’s lower chambers, or a leaking valve in the single pumping chamber keeping them alive, needs care matched to that exact anatomy, while the devices surgeons reach for have historically almost never been designed with that specific child in mind.

“You’ve got a one-of-a-kind kid and an off-the-shelf device,” said Dr. Matthew Jolley, a cardiologist and researcher at Children’s Hospital of Philadelphia, in NVIDIA’s blog post. “Our job is to find what fits — and modeling lets us do that before anyone goes into the cath lab or operating room.”

The approach is spreading, NVIDIA reports. Cardiac modeling programs now operate at more than 20 children’s hospitals across the U.S. At Boston Children’s Hospital, modeling supports more than half of all cardiac surgeries, roughly 500 cases a year, and CHOP expects to reach about 200 modeled cases in 2026. Where the work began in cardiac care, CHOP now aims to apply the same tools across multiple disciplines through the IDEA Lab, part of the hospital’s Morgan Center for Research and Innovation. CHOP’s Research Institute has described the IDEA Lab, short for Innovate, Design, Engineer, and Accelerate, as a centerpiece of the Morgan Center offering on-campus biofabrication resources, led by Elizabeth Silvestro, PhD.

From SlicerHeart to Standard of Care

Jolley joined CHOP in 2015, when 3D echocardiography was just coming online. Tools existed for modeling adult valves, but almost nothing had been built for the complex, small anatomies he was treating, according to NVIDIA’s account. Jolley’s lab and the open-source community built SlicerHeart, an extension of the open-source 3D Slicer software used to visualize, segment and analyze 3D medical images, and the team developed workflows for modeling pediatric hearts and valves from multiple imaging modalities.

For years, producing a single model meant a skilled research assistant spending hours at a workstation. Using MONAI Label and NVIDIA’s Auto3DSeg implementation, Jolley’s team trained segmentation networks on pairs of prior images and models, and NVIDIA reports the output meets the same quality standard a trained human would produce, in seconds rather than hours. Jolley said machine learning has become routine for the lab: as soon as the team has made 10 or 20 image-model pairs, it trains a model and starts applying it.

MONAI’s own project site describes a PyTorch-based, community-led ecosystem for medical-imaging AI: MONAI Label provides active-learning annotation that plugs into 3D Slicer, MONAI Core supplies the training and research library, and MONAI Deploy packages trained models for clinical inference pipelines, with Auto3DSeg among its reference implementations. The framework is maintained by researchers and engineers at NVIDIA, NIH, King’s College London, Mayo Clinic, MSKCC, Stanford, DKFZ and more than 30 other institutions, and is distributed under an Apache 2.0 license.

The clinical impact appeared quickly, according to NVIDIA. For complex ventricular septal defects, holes between the heart’s lower chambers, CHOP now models routinely before surgery. In one early case, a child had already undergone two failed repair attempts, with surgeons unable to locate the defect using traditional methods; the 3D model clarified the anatomy, and the repair succeeded on the first try. For cases like these, Jolley said, cardiac modeling has moved from research to standard of care.

Simulating Devices Before Surgery

For Jolley’s team, the goal extends past visualization: they want to predict what will happen when a device is deployed inside a child’s heart, before any procedure begins. That work runs on Newton, an open-source physics engine built on the NVIDIA Warp Python framework that runs physics simulations on GPUs and was originally intended for simulation-based AI robot training.

CHOP is working with NVIDIA and the open-source community to build biomechanics-focused simulation frameworks with Warp that can be brought into Newton. Once integrated with 3D Slicer and SlicerHeart, the frameworks are intended to help doctors understand the tissue material properties that determine how a device will deploy in a given patient. NVIDIA reports that GPU acceleration can cut the time needed for cardiac device simulation from up to four hours, or a full overnight run for multiple configurations, to near real time, fast enough in practice for a clinician comparing how different devices fit a child’s anatomy to inform a same-day decision.

CHOP has begun implementing features built on Warp and Newton for the closure devices used to seal holes in children’s hearts, and it hopes to apply similar methods to simulations of transcatheter valves. The open-source architecture of Warp and Newton is being connected with SlicerHeart, with the long-term goal of bringing real-time simulations into clinical workflows. A coupler using SlicerHeart and NVIDIA Omniverse digital twins, powered by OpenUSD, is also in development: simulations built from patient images would flow into virtual reality environments with embedded vision-language models, letting clinicians query and interact with a child’s cardiac anatomy before acting on it.

Open-Source Economics and a National Consortium

About 2.4 million people in the U.S. live with congenital heart disease, and NVIDIA describes this population as historically too rare and too diverse to attract traditional device-company investment at the scale families need. Jolley said the population is too small to support traditional commercial development by normal economics, and that open source defies those economics for small and heterogeneous populations by allowing collaboration and progress without barriers.

The SlicerHeart tools are available at no cost for others to use and extend, and researchers at Stanford and Boston Children’s contribute additional tools alongside CHOP. A national consortium of children’s hospitals is now forming to build next-generation shared modeling infrastructure, with open-source tools linking work across the participating institutions. NVIDIA said its investment and collaboration on open platforms, including MONAI for medical imaging AI, Newton for physics simulation and OpenUSD for 3D interoperability and virtual reality, gives a children’s hospital lab access to tools that would otherwise require a company-sized engineering team to build and sustain.