ACR expands pilot program designed to help radiologists create AI

The American College of Radiology (ACR) has expanded its ACR AI-LAB pilot program geared toward helping radiologists develop AI models without the use of coding language.

According to an ACR statement, the program now includes seven total institutions. The initial institutions—Massachusetts General Hospital and The Ohio State University—will work alongside Lahey Hospital and Medical Center, Emory University, The University of Washington, the University of California San Francisco and Brigham and Women’s Hospital.

All seven locations will share AI models created at their institution, testing platforms on local patient data and fine-tuning AI based on individual results. At the heart of the program is the ability for radiologists to adjust AI models without making line-by-line changes to code. This has never been done in radiology at such a large scale, according to the statement.

“ACR AI-LAB has kicked off a very exciting era of AI democratization, making it possible for health care institutions and industry to build customized AI models for investigative purposes without coding and without moving image data off premises,” said Keith Dreyer, DO, PhD, ACR DSI chief science officer, in the ACR statement. “Soon all institutions interested in participating in the AI democratization revolution will have the opportunity to get involved.”

Nvidia is also providing it Clara AI software toolkits to help with annotation, transfer learning and “pipeline integration.” Nuance will help integrate AI platforms for radiologists.

“Today marks a major step in accelerating the development of AI for medical imaging. We know algorithms can underperform when deployed at sites where they weren’t trained,” said Bibb Allen Jr., MD,  ACR DSI chief medical officer. “Now, radiologists in the pilot program will have access to AI algorithms developed outside their institutions in order evaluate a model’s performance using their own data and, as necessary, retrain the algorithm using their local data to enhance its performance."