RSNA 2016: Radiologists must become as centaurs joined at the midsection to AI

RSNA, Chicago—No two words struck more fear into the hearts of radiologists in 2016 than “artificial” and “intelligence.” The cascading virality of the hashtag #AI only intensified the sense that, slowly but surely, something doomful this way comes.  

More than a few have wondered aloud if big imaging data could indeed continue feeding the deep-learning beast to the point where computing power gets buff enough to elbow radiologists all the way out of the reading room.

“Take a permanent break, human. From now on, I’ll handle this.”

Exhale, rads. It’s not going to happen.

However, AI will challenge radiology to, in effect, cut a deal:

If the specialty will collaborate and cooperate with the technology, the technology will allow radiology to help lead all of healthcare deep into the promising new world of precision medicine.

Such was the reassuring word from Keith Dreyer, DO, PhD, in what was, for this press-pass-bearing attendee, one of the most intriguing presentations at an RSNA that had no shortage of must-see goings-on.    

“Radiologists will be the centaur diagnosticians, allowing machines to make us smarter, help us do more and give us more value,” said Dreyer, who is associate professor of radiology at Harvard Medical School and director of the Center for Clinical Data Science at Massachusetts General Hospital.

Dreyer spoke to a nearly packed house in Chicago’s 4,250-seat Arie Crown Theater during the 102nd Scientific Assembly & Annual Meeting of the Radiological Society of North America.

His use of Garry Kasparov’s famous word picture—the “chess centaur” is half human, half computer—had me hoping the head end turns out to be the human (leading) and the computer the tail (following) rather than the other way around.

Honestly, I’m not sure it makes much difference. It is, after all, only an analogy.

Either way, a new day really is dawning. Dreyer predicted that technological singularity—the point at which computers’ collective “superintelligence” rises above the limits of human comprehension—could arrive right around 2029, just as famously projected by the futurist Ray Kurzweil.

Don’t fret about that day, Dreyer suggested. Embrace it.

“Soon we will be able to create a precision radiology report for all body parts and all examinations,” Dreyer said, advising his fellow radiologists to push for AI that not only automates their processes but also augments their expertise.

“We should use AI to expand our diagnostic and clinical roles,” he said, “serving as our patients’ trusted advisor.”

Which would, one can’t help but notice, bring radiology right back to where it started.