A pitfall to AI in radiology diagnosis

Radiologists are more likely to align their diagnostic decisions with AI advice and spend less time on decision making, according to a Radiological Society of North America study.

The study, published Nov. 20 in Radiology, had 220 radiologists and internal medicine or emergency medicine physicians read chest X-rays alongside AI advice. Physicians were asked to evaluate eight X-ray cases obtained from Beth Israel Deaconess Hospital in Boston via the open-source MIMI Chest X-Ray Database. Participants were presented with the patient's clinical history, AI advice and X-ray images. The AI provided either correct or incorrect diagnosis with local or global explanations.

The physicians could accept, modify or reject the AI suggestions and were asked to report their confidence level in the findings and the usefulness of the AI advice.

Researchers found that reviewers were more likely to align their diagnostic decision with AI advice, spent less time considering the advice when AI provided a local explanation. 

When the AI was correct, the average diagnostic accuracy among physicians was 92.8% with local explanations. When the AI was incorrect, the physician accuracy was 23.6%.

"When we rely too much on whatever the computer tells us, that's a problem, because AI is not always right," senior author Paul Yi, MD, director of intelligent imaging informatics and associate member in the department of radiology at St. Jude Children's Research Hospital in Memphis, Tenn., said in the news release. "I think as radiologists using AI, we need to be aware of these pitfalls and stay mindful of our diagnostic patterns and training."

Copyright © 2024 Becker's Healthcare. All Rights Reserved. Privacy Policy. Cookie Policy. Linking and Reprinting Policy.

 

Articles We Think You'll Like

 

Featured Whitepapers

Featured Webinars