Radiology experts attempt to predict future of imaging

June 7, 2005
Douglas Page

A panel of radiology seers consulted a collective crystal ball during the closing session of the SCAR meeting and attempted to predict what the future holds for radiology in three, 10, and 25 years.

A panel of radiology seers consulted a collective crystal ball during the closing session of the SCAR meeting and attempted to predict what the future holds for radiology in three, 10, and 25 years.

The panel included Richard Morin, Ph.D., of the Mayo Clinic, Jacksonville, Dr. Paul Chang, University of Pittsburgh Medical Center, and Dr. Eliot Siegel, University of Maryland/Baltimore VA Medical Center. Among the issues the experts addressed was the question of whether "radiologist" will be the most familiar term for those performing medical imaging in the future.

"That's not even true today," Siegel said. "Most images are acquired outside of radiology."

Chang expressed concerns about radiology's future.

"My biggest fear is what my chief orthopedist said: that the principal value of radiology is as a 'safety net.' Radiology has the potential to marginalize itself. Our surgical colleagues aren't turfing it out at 3 in the morning," he said.

Radiology needs to add value to its service, according to Morin.

"It's time for radiology to reexamine the role of medical imagers and how it can be enhanced," he said. "If we pushed structured reporting throughout all of radiology, we could build a national database to predict outcomes. It's time that we really know how good a CT study is in diagnosing early pulmonary disease. That is value added."

The panel also speculated on what portion of exams will be reviewed by computers in the future.

By 2008, 90% of exams will be read by humans, trailing to 50% by 2015 and 45% by 2030, Morin said.

"We will get to some level of imaging where humans will still be necessary, unless there is a tremendous breakthrough in the field of computer-assisted detection," he said.

Computer-assisted diagnosis could be a reality by 2030, although that idea may be anathema to radiologists, he said. But technology is just too good, and humans are too few, particularly when 4000 image data sets are being generated.

"I see a clear trend where machines will be interpreting images, although humans will never be completely out of that loop," Morin said.

Humans will at least be reviewing and signing off on every study 25 years from now, Siegel said, and imaging processing and decision support will converge with computer-assisted diagnosis.

In the current model, a radiologist looks at a mammogram and the computer acts as a second reader.

"I believe that will go away, and computers will become an integral part in providing information to the radiologist at the point of diagnosis," Siegel said.