American College of Radiology expands tools to help practices evaluate imaging AI
As artificial intelligence rapidly proliferates across medical imaging, the American College of Radiology is rolling out a growing suite of resources aimed at helping practices make informed decisions about adopting and monitoring these tools.
Radiology Business spoke with Christoph Wald, MD, MBA, vice chair of the ACR Board of Chancellors, professor of radiology at Mayo Clinic, to find our more about the resources for practices to better evaluate artificial intelligence imaging algorithms.
"We have a repository of all the pixel-based AI that is FDA cleared in the United States for perusal by our members and radiologists in the country to make informed purchasing decisions,” he said
The rapidly evolving AI landscape at the annual meeting of the Radiological Society of North America highlights the sheer scale and complexity facing providers. Hundreds of vendors now offer AI-enabled products, spanning both FDA-cleared clinical tools and non-regulated applications focused on workflow and reporting. The FDA has cleared more than, 1,400 algorithms for clinical use, and 80% are for medical imaging. But radiologists need a way to be able to sift through all these algorithms too make informed purchasing choices and understand what is out there, Wald said.
The repository focuses on image-based, or “pixel-based,” AI algorithms—tools that analyze imaging data to assist with triage or diagnosis. But Wald emphasized that innovation extends well beyond image interpretation.
“Here at the show, you see a dichotomy between image-based AI, which helps either triage patients that may have a positive finding or may help make a diagnosis. And then non-pixel-based AI, which is probably equally exciting for us in radiology practice. These generate impressions, use large language model-based report generation, AI for asset management, and AI to get a patient optimally through your practice. A lot of workflow tools that use AI to optimize outputs. So a whole spectrum of products coming at us in radiology practice that are helping us do our job,” Wald explained.
As radiology AI expands, the ACR has built out infrastructure through its Data Science Institute, launched in 2017. This includes a portfolio of programs to guide evaluation and implementation of these algorithms into clinical practice.
“We've seen an increased adoption of AI. We founded the Data Science Institute, which is the organizational body in the ACR that tries to help our members get ahead. We've cataloged all the AI and we've increasingly realized that AI does not necessarily work the way in your practice as it did in the pre-market testing," Wald said.
The ACR program helps practices responsibly adopt AI, to make a careful selection of these products and helps them do acceptance testing. The college also created Assess AI registry, designed to track AI results, de-identified reports, and collect DICOM image data. Wald said a portal shows practices how well the AI is working over time and helps detect drift in algorithms.
A key focus of these efforts is enabling ongoing performance monitoring in real-world clinical environments. Changes in imaging equipment, protocols, or patient populations can all impact how an algorithm performs after deployment.
Dave Fornell has covered healthcare for more than 17 years, with a focus in cardiology and radiology. Fornell is a 5-time winner of a Jesse H. Neal Award, the most prestigious editorial honors in the field of specialized journalism. The wins included best technical content, best use of social media and best COVID-19 coverage. Fornell was also a three-time Neal finalist for best range of work by a single author. He produces more than 100 editorial videos each year, most of them interviews with key opinion leaders in medicine. He also writes technical articles, covers key trends, conducts video hospital site visits, and is very involved with social media. E-mail: [email protected]