Though the use of artificial intelligence has been steadily on the rise in radiology over the last decade, it has failed to push radiologists off a metaphorical cliff as was prophesized five years ago.
“There's no ground underneath (radiologists). It's just completely obvious that in five years deep learning is going to do better than radiologists,” IBM CEO Virginia Rometty declared at an information technology speaking event. Those words came five years after Watson, IBM’s question-answering computer, defeated famously decorated Jeopardy champions, Ken Jennings and Brad Rutter.
Soon after, computer scientist and noted neural network leader Geoffrey Hinton then compared the field of radiology to a cartoon coyote who had run off a cliff but had not yet realized there was no longer any ground beneath him. Some even proclaimed that it was time to stop training radiologists because their days of interpreting images were effectively ending in the wake of AI algorithms.
Years later, however, the rug still remains soundly under human feet. To honor the anniversary of radiology being declared a dying profession, Richard B Gunderman MD, PhD, and Aish Thamba, BS, both with the department of radiology at Indiana University School of Medicine, published an editorial in Academic Radiology that discusses how far AI has come while also reminding us of how far it has yet to go.
“In the intervening years, it has become clear that IBM and Hinton overestimated the speed with which Watson and image perception algorithms would overtake physicians, and especially radiologists,” the authors wrote. “Hinton may not have understood the full scope of contributions radiologists make to patient care, supposing that it is confined to image interpretation.”
They pointed to the failed partnership between Watson and MD Anderson Cancer Center, when Watson’s performance could not compare to a physician's when treating patients as individuals. They then go on to highlight the many roles a radiologist fulfills—from caregiver to teacher to mentor and beyond—before acknowledging the biggest edge radiologists have over machines: the ability to see patients as humans, rather than datasets.
“Radiologists must still see patients as more than mere collections of data. The ultimate challenge in improving patient care is not to make patients, diseases, and physicians fit our models, but to make our models fit patients, diseases, and physicians,” the authors concluded.
You can read the full editorial in Academic Radiology.
In addition to her background in journalism, Hannah also has patient-facing experience in clinical settings, having spent more than 12 years working as a registered rad tech. She joined Innovate Healthcare in 2021 and has since put her unique expertise to use in her editorial role with Health Imaging.