CHOP Uses AI to Build Patient-Specific Heart Models in Seconds

Children's Hospital of Philadelphia has deployed an AI-powered modeling system that transforms medical imaging into patient-specific virtual heart replicas, allowing surgeons to rehearse pediatric cardiac procedures before entering the operating room.

The technology addresses a significant clinical need: approximately 1% of all live births involve congenital heart defects, ranging from holes in heart chambers to valve abnormalities. Planning surgical interventions for these conditions traditionally required extensive manual evaluation of imaging data.

Dramatic reduction in planning time

Dr. Matthew Jolley, a pediatric cardiologist and cardiac anesthesiologist at CHOP, said the AI platform has compressed treatment planning from five to six hours down to four to six seconds. The system converts CT scans, MRIs, and ultrasounds into advanced three-dimensional heart models using NVIDIA's MONAI framework, an open-source platform for medical imaging AI.

The virtual replicas enable physicians to compare different medical devices and simulate procedures specific to each patient's anatomy. "We can take those images of a child and turn it into a selection process where you say, okay, I have these five devices, which one is going to match best to that child in the operating room or in the cath lab," Jolley explained. "You go through that virtual rehearsal ahead of time."

Beyond surgical planning

The models serve a dual purpose beyond clinical decision-making. Jolley noted that the visualizations help families better understand planned procedures before they occur, potentially improving informed consent and reducing anxiety around complex cardiac interventions.

CHOP currently processes between 150 and 200 model cases annually. The hospital has expanded the program into a coalition that includes Boston Children's Hospital and Stanford, allowing participating institutions to evaluate patient outcomes across a larger dataset.

Why it matters

Congenital heart defects remain among the most common and serious birth abnormalities, often requiring multiple interventions throughout childhood. AI-powered surgical planning could reduce operative complications by improving device selection and allowing surgeons to anticipate anatomical challenges before the first incision. The multi-institutional approach also creates opportunities for comparative effectiveness research that would be difficult for any single center to conduct alone.

The details were first reported by 6abc Philadelphia.