The radiology profession is a famously paranoid lot, often worried about encroachment on imaging from non-radiologists, competition from teleradiology, reimbursement squeezes, and more. Auntminnie.com is a good place to go to observe how these worries play out.
The latest article, Will machine learning turn radiologists into losers? is a case in point. It reports on a New England Journal of Medicine article that asserts that machine learning will replace radiologists. Images will be sent straight to algorithms, bypassing radiologists completely, they say. There are differences of opinion on how soon this will happen, but an appreciation that once the electronic tools are good enough they will be preferred.
As a patient I’m all in favor of faster, more accurate, and cheaper interpretation of images and if that means there’s no future role for radiologists, so be it. But actually what I hope is that radiologists start to assert themselves as diagnostic quarterbacks, helping to organize and analyze information from pathology, genomics, lab tests and physical examinations. They can work with teams of clinicians in new ways, to speed diagnosis and treatment decisions.
I am aware that some enlightened radiology leaders are already thinking in these terms. I hope the fears spurred by the development of machine learning will accelerate the movement.
Image courtesy of Stuart Miles at FreeDigitalPhotos.net