The American College of Radiology Data Science Institute (ACR DSI) now offers a catalog of 111 FDA-cleared medical imaging artificial intelligence (AI) algorithms (class II) searchable by company, subspecialty, body area, modality and date cleared. Carefully designed for radiologists, the catalog dramatically reduces the time required to accomplish the formerly complex task of sorting through the algorithms available to those in medical imaging.
“As practicing radiologists, we are interested in AI solutions that could be a good match for our future clinical practice,” said Christoph Wald, MD, Ph.D., MBA, FACR, chairman of the department of radiology at Lahey Hospital & Medical Center in Boston, Massachusetts, and chair of the ACR Commission on Informatics. “Like many institutions, we are looking for ways AI can support the changing clinical landscape where new treatments are emerging, such as in liver-directed therapy and emerging neurotherapies. The ACR DSI catalog not only shows us what’s commercially available, it levels the playing field by bringing all related products into view.”
Searches for algorithms on FDA’s site are cumbersome at best, according to Keith Dreyer, DO, PhD, FACR, ACR DSI chief science officer and chief imaging information officer at Mass General Brigham, making the new tool a big asset to those looking for AI solutions for their clinical practice. “Adding search features in the ACR DSI tool quickly shows you what is available and applicable. This guides radiologists in creating a short list of models and streamlines the process for vetting those options,” he said.
“Until now, you have needed an understanding of what you were looking for to find AI solutions for your institution. We’ve now shortened and simplified the process for radiologists of all practice sizes to understand what is available to them,” said Bibb Allen Jr., MD, FACR, ACR DSI chief medical officer and diagnostic radiologist at Grandview Medical Center.
The new search features were added based on feedback from ACR members in an AI use survey conducted in 2020 that suggests 30% of radiology practices use some form of AI. Searches can now be conducted for all applicable algorithms, or simply to bring up algorithms cleared during a given time period. Many of the models match ACR DSI Define-AI use cases and are linked under related use cases. Clicking on individual models takes users directly to the FDA summary letter for more details.
An ACR Editorial Board guides and reviews the catalog content to ensure it is useful for the community and potential purchasers of commercial algorithms. Members include Wald, Dreyer and Allen, as well as Sheela Agarwal, MD, MBA, ACR senior scientist and digital medical advisor at Bayer HealthCare; Judy Gichoya, MD, MS, assistant professor, interventional radiology and informatics at Emory University: and Jay W. Patti, MD, chief radiology informatics officer at Mecklenburg Radiology Associates.
On March 3, 2021 from 7 to 8 p.m. EST, the webinar, “FDA-Cleared AI Algorithms for Medical Imaging: Explore the Latest AI Tools,” will share best practices for vetting commercial algorithms using the catalog. For more information, visit the ACR DSI website.
