Radiology leaders optimistic about AI but see cost as a key hurdle

Radiology department chairs are enthusiastic about artificial intelligence in diagnostic imaging but see cost as a key obstacle to its implementation, according to new research released Friday.

Meanwhile, leaders within the specialty expressed minimal concern about the advent of generative AI in medicine and fears this emerging technology could replace workers. The findings are derived from a small survey of 43 academic department chairs, published in the Journal of the American College of Radiology. 

Asked to share their AI wish lists, leaders labeled image acquisition, radiologist workflows, and post-processing of scans as top priorities. 

“Based on our survey, chairs appear optimistic and generally prepared to navigate this early phase of AI adoption in radiology practice,” Elizabeth S. Burnside, MD, MPH, a breast imager and associate dean with the University of Wisconsin–Madison’s school of medicine, and co-authors wrote Jan. 10. Leaders plan to do so, they added, “by focusing on implementation goals important for their constituency, tackling challenges across the tame/wicked spectrum, and monitoring constituencies for replacement anxiety. Future research into key concepts will be crucially important for effective implementation of AI in practice.”

Researchers conducted the web survey in October 2023, targeting Society of Chairs of Academic Radiology Departments members. They contacted a total of 112 individuals (for a response rate of 38%), hoping to tease out their AI outlook and plans for implementation. Burnside and colleagues found no statistical difference between chairs’ general views around AI versus generative artificial intelligence (i.e., ChatGPT or other programs used to create images or other new content). 

Chairs anticipate using this technology to bolster quality and efficiency in their departments (100%), reduce burnout (95%), cut costs (51%), improve healthcare equity (63%), and optimize image acquisition (42%). They also perceived that both radiologists (84%) and technologists (88%) on their staffs are not particularly worried about AI taking over their work. About 72% of chairs said they believe radiology trainees are “not at all” or “a little” worried about being replaced by artificial intelligence. 

Burnside and co-authors offered a free-text space for leaders to further expound on their answers. Several respondents highlighted cost as a critical aspect of decision-making, using phrases such as “too expensive,” “funding is the biggest limitation,” and “high interest, but hospital has shown little interest supporting.” Chairs emphasized the importance of potential AI purchases addressing today’s practice pressures. Given the “national workforce shortage,” for instance, one chair said they will “prioritize AI that increases efficiency.”

“Despite the availability of AI to support a plethora of tasks throughout the imaging value chain, chairs appear to be focused, with slightly less unanimity, on tasks that most closely impact their constituency: radiologists, technologists and scribes,” the authors noted. 

The pre-proof study is limited by its small response rate, lack of demographic details on its respondents, and other factors. Burnside and colleagues also noted that radiology department chairs do not always have control of AI implementation, with health systems or other stakeholders sometimes steering the process. They see avenues for further exploration of some of the survey’s findings. 

“For example, the amount of specialized knowledge that leaders need to make informed/optimal decisions about AI implementation, the balance [of] AI implementation tradeoffs such as image quality vs. radiation dose or expense vs. benefits that chairs should attempt to achieve,” the authors wrote. “Purposeful investment in education and research to support leadership decision-making in diagnostic radiology will likely accelerate the positive contributions AI will make to our field in the future,” they concluded.