AI in Radiology: Emerging Trends, Current Obstacles and Future Directions

For radiologists who may be apprehensive about the impact of AI, Nina Kottler, MD, MS, FSIIM, FAIM, emphasized the importance of reframing expectations in a recent interview with Diagnostic Imaging.

In a recently issued Future Health Index report on AI in health care, a majority of clinicians credited AI with facilitating improvements in precision, workflow efficiency and work-life balance. However, surveyed clinicians in the report also cited concerns about AI training and overreliance on AI.

While acknowledging a fair amount of hesitancy in the field about AI, Dr. Kottler said at times there is a tendency to “throw the baby out with the bath water” when there is an obvious error with AI use. She noted the importance of context and perspective when evaluating AI tools.

For example, Dr. Kottler raised the example of an AI tool for detecting pneumothorax on chest X-rays that will likely have a better performance in an intensive care unit that it will in an outpatient setting.

“There (are) some basic fundamental things that people have to understand about AI in general. Then the second piece of what they have to know is: how does that very specific AI work? What kind of errors is it making, and where does it do well? Setting that expectation in advance is really important, not only so that we could decrease the sort of automation bias and complacency, the over trust of AI, because we have to show people where it doesn't work, but we also have to mitigate the distrust of AI,” maintained Dr. Kottler, the chief medical AI officer for Mosaic Clinical Technologies.

Dr. Kottler also emphasized that appropriate training in AI for radiologists is clinical training.

“You need clinical people to talk about where it works well and where it doesn't,” added Dr. Kottler. “I will say that in general, people who don't have literacy in AI are going to fall behind. It's going to become a patient safety competency.”

While 39 percent of the surveyed clinicians in the report noted concerns about AI possibly contributing to a loss of clinical skills, Dr. Kottler said there is going to be a certain letting go with the acquisition of new skills and a shift in the roles of radiologists.

“It's not that I don't think about it. I do think about it, and it's important to keep up our skills, but frankly, some of our skills we're going to let go in order to pick up new skills. It's kind of hard to pick up new skills when you're still like managing everything that you're doing now,” posited Dr. Kottler. “Some capability that AI will take over will enable us to do stuff we've never done before, and I think, frankly, that adds more value to the system. So we have to think about it in that more holistic mechanism.”