After three decades in practice, NYU Langone Medical Center’s chairman of Radiology, Dr. Michael Recht, has gained substantial insight into what the radiology field gets right and what it could improve. That’s why shifting focus to radiology AI a few years back was a natural progression.

Recht is exploring how artificial intelligence can be used to benefit patients and practitioners and ultimately, change how radiology is practiced. The change would rely on technology, but in Recht’s opinion, it’s needed because of technology too.

“I think what I would tell you is when I went into radiology, it was really the center of what you did in medicine. So in all my rotations, when we had questions, we would go down to radiology, look at the films, talk to the radiologists, and we just knew those were the people who could give us a lot of the answers,” he recalled.

Radiologists weren’t just providing information on what was going on with patients he said, but they would also give direction as to what to do next from an imaging point of view, but times have changed and in part due to the introduction of PACS, there’s a perception that there’s not much of a need to visit with radiologists face to face.

Today, when Recht interviews radiology resident applicants, he always asks what kind of experience they have on radiology rounds. They usually answer that they very rarely had rounds or they ask him if he means with an actual radiologist or just reviewing the films in the ward with the attending physicians on their service. “I think that’s really sad and I think AI has a chance to change how we practice radiology and how we integrate with the clinical care teams by allowing us to do a lot of what we do more efficiently and better, freeing up our time from tasks that we currently do, but really don’t need to do. Allowing us to do the things that really drew us into radiology,” he said.

When healthcare AI started, it was about pattern recognition and whether it could really make the diagnosis. Since those days, it’s become clear that AI can impact every aspect of the image chain, from making sure people order the appropriate test, to ensuring exams are protocoled correctly, to data mining the EMR for a more complete picture to use in patient treatment. AI will also be used for quantitative measurements. “I don’t think many radiologists went into radiology to use a ruler or electronic calipers, so this will be welcomed,” Recht said. AI may help to create reports that are patient-friendly instead of most current reports that are more physician-focused.

One of the most exciting and promising uses of AI is in image acquisition and reconstruction. “We’re doing a lot of work in this area on our own, but we also have a research partnership with Facebook. Our goal is to accelerate MR and CT by a factor of 10. “In MR what that means is that we can do imaging in five minutes – as fast as you can do an X-ray. If we can do that, it takes away one of the reasons to use X-ray over MR. The other challenge in using MR is lowering cost. For CT, we know that a chest CT can give more information than a chest X-ray, but we can’t replace all chest x-rays with CT because of radiation dose. However, if we were able to use AI to allow us save time in MR and decrease radiation dose in CT, again we’d be able to significantly change our imaging paradigms.

One way AI won’t impact radiology, in Recht’s opinion, is by replacing radiologists. More people will be getting imaging exams in the coming years and the exams are getting larger. AI will help radiologists keep up without burning out, will be used to help radiologists increase their accuracy, and will give them more opportunities to pursue aspects of their jobs that can best benefit from their experience and expertise. “My daughter is a radiology resident and the way she is going to practice will be incredibly different than the way we practice today, but I am confident AI is going to make her career most exciting and fulfilling,” Recht said.