Radiology professionals from seven top healthcare institutions will use the ACR AI-LAB to demonstrate the process of creating investigational artificial intelligence (AI) models from image data without the use of a programming language. Using an AI model developed at one institution, each of the seven institutions will have the ability to evaluate and optimize the model for their own investigational use.

Based on the recently announced ACR AI-LAB reference architecture, this pilot represents a major milestone in the effort to allow institutions to develop high-quality algorithms that address local clinical needs, some of which may ultimately be made commercially available. In addition to the seven institutions, there are two major technology contributors; NVIDIA is providing software and edge infrastructure, and Nuance is providing last-mile integration to the participating radiologist.

The pilot—originally including Massachusetts General Hospital and The Ohio State University—now also includes Lahey Hospital and Medical Center, Emory University, The University of Washington, the University of California San Francisco and Brigham and Women’s Hospital. NVIDIA will provide its NVIDIA Clara AI software toolkits at no cost to the institutions to perform the annotation creation, transfer learning, and pipeline integration. In addition, Nuance will provide the last-mile technology required to integrate AI for the participating radiologist. Once the pilot is complete, the initiative is anticipated to progressively expand to all institutions interested in participating.

Sharing local AI models from image data between institutions for fine-tuning—while patient information remains securely on site at the originating institution—has not previously been done successfully in radiology at this scale. This is due, in part, to the variability in how medical images are created, including the equipment, software, and protocols used.

The pilot sites will use ACR AI-LAB to evaluate AI developed elsewhere, modifying the investigational algorithms to improve performance based on testing and evaluating them on local patient data. Creating the local AI models will not require ACR AI-LAB users to have programming skills. ACR AI-LAB allows users to adjust and change AI models without having to make line-by-line changes to the underlying code.

Once the pilot is complete, the consortium is anticipated to progressively expand to more institutions and vendors interested in participating. The investigational algorithms resulting from this project will undergo further evaluation and refinement by sites should they pursue commercialization, including obtaining appropriate regulatory clearance or approvals, as applicable.

“Today marks a major step in accelerating the development of AI for medical imaging,” says Bibb Allen Jr., MD, FACR, ACR Data Science Institute Chief Medical Officer. “We know algorithms can underperform when deployed at sites where they weren’t trained. 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.”

The ACR DSI made the initial version of the AI development framework available in April 2019. More information on ACR AI-LAB is available at ailab.acr.org