Are radiologists becoming obsolete?

Radiologists wonder if technology will replace human imaging interpretations.


With the evolution of IBM’s Watson supercomputer and other advancements in artificial intelligence, could machines replace radiologists? That’s a topic to be explored during a session at RSNA 2016, Nov. 27-Dec. 2 at McCormick Place in Chicago.

Eilot Siegel, M.D., a radiologist and professor at University of Maryland School of Medicine, and Bradley J. Erickson, a professor and associate chair for research at Mayo Clinic, will assess issues for colleagues to consider. Radiologists are worried that technology could replace them, acknowledges Siegel. “What they know is what they hear in presentations and read in journals. They are hearing that the threat is machine learning and artificial intelligence.”

In fact, U-Maryland is seeing some students moving away from radiology to other specialties. They can stop worrying, Siegel says. His answer: Stop worrying. “My message simply is that it is complete nonsense that we’ll be replaced any time soon.”

New applications will make radiologists’ work safer and smarter, he says. “Machines can’t do complex image interpretations from typical studies that radiologists read.” At the same time, there are amazing and exciting developments ahead that radiologists can look forward to, he contends.

Erickson will play the opposite role, taking the view that super machines are a threat, and point out advancements in technology such as self-driving cars, next-generation speech recognition and investments in artificial intelligence. And there are other technologies not yet known that will come into radiology in the next 20 years. However, both presenters believe the answer lies somewhere in the middle, Siegel says.

If nothing else, Siegel will argue that even if aliens came with a robot able to interpret images, it still would take many years to test the robot’s proficiencies in multiple imaging modalities and get FDA approval. Further, radiologists have tasks beyond interpreting images such as supervising technicians, working to optimize patient safety and help doctors make better decisions, he adds.

“The whole idea to replace us would only be a subset of what we do. Machines won’t be able to match the human aspect of radiology. Those who believe there is a good chance of replace need to know there is absolutely no chance.” The session, “Elementary, My Dear Watson: Will Machines Replace Radiologists?” is scheduled at 7:15 a.m. on Wednesday, Nov. 30 in Room E450.

More for you

Loading data for hdm_tax_topic #better-outcomes...