French radiologists optimistic about AI's benefits, ready to learn more

A majority of French radiologists feel that they don’t know enough about AI, according to a new study published in Diagnostic and Interventional Imaging. But they want to learn more and sound optimistic about the technology’s many benefits.

Researchers from the department of musculoskeletal radiology at the University Hospital of Lille in France surveyed 200 radiologists and 70 radiology residents about the topic in January 2019. The survey included 42 questions and took approximately 5-10 minutes to complete. More than 67% of respondents were men and the mean age was more than 39 years old.

Overall, 68.5% of respondents answered that they do not use AI in their day-to-day work and do not expect that to change within the next year. Just 7% of respondents answered that they use AI on a day-to-day basis, and 24.4% said they “would consider” using AI while working within a year.

When asked about their general knowledge regarding AI, 73.3% of respondents said they “had received insufficient previous information” on the topic. Another 13.7% said they had never been instructed about AI in any way.

In addition, 33% of respondents said they had read no scientific publications about AI, and 52.6% said they had read less than five publications about the topic. Considering all of these answers, the research team determined that 23% of the respondents had “basic knowledge on AI in radiology.”

“Our results clearly show that there is an ongoing lack of training and information about AI among French radiologists, despite the tremendous increase in scientific publications,” wrote corresponding author Thibaut Jacques and colleagues. “The overflow of information and the fast evolution in this field could paradoxically be discouraging individual radiologists, especially those who think that a strong technical background is required.”

While a whopping 94.4% of respondents said they were interested in receiving “generic” training on AI, 69.3% showed interest in “technically advanced training on this subject.” How do they think AI will impact radiology? The most popular answers were lowering medical errors related to radiology (81%), providing faster interpretation times (74.4%) and allowing radiologists to spend more time with patients (52.2%).

“Interestingly, the reduction in interpretation time came in second position of all positive expectations, after the reduction of medical error and immediately followed by the increase of time spent with patients, instead of the increase in the volume of examinations which came last,” the authors wrote. “If it were to be proven again, these results confirm that the main concern of radiologists remains the patients.”