Radiologists look at and analyze images, tasks computer algorithms can do. This is fertile soil for artificial intelligence (AI) — enough so that some predict that AI will replace radiologists.
Emory bioethicist John Banja says: don’t believe the hype. AI will generate tools radiologists will want to use, he says. But human experts will have plenty to do, including making sure that the algorithms are properly vetted and trained on appropriate data.
“We already know what a lot of the ethical issues are going to be…informed consent, privacy, data protection, ownership, all that kind of stuff,” Banja recently told Health Imaging. “What we need to do is drill down to the next level, especially the practice level.”
Banja has received a grant from the Advanced Radiology Services Foundation to support a series of podcasts with radiologists over the next two years. He will be teaming up with Emory radiologist Rich Duszak, a specialist in health policy, and Norm Beauchamp, medical dean at Michigan State.
Banja and Duszak are still planning podcast sessions and lining up interviews, but they said the first episode will be on “AI hype”, and the second will cover standard of care/medical malpractice, with future issues on FDA standards.
Duszak comments on how radiologists need to take control of the algorithms in this video.
Also, with radiology chair Carolyn Meltzer, Banja recently published a review on ethics related to radiology and AI, exploring issues such as selection bias and stretching algorithms too far.
Google’s Project Nightingale highlights concern about big data companies having access to large numbers of patient records – including radiology images.