Most radiologists using AI favor ‘home-grown’ algorithms over commercially available tools

Radiologists utilizing artificial intelligence in clinical practice favor algorithms they’ve created themselves compared to commercially available tools, the American College of Radiology reported this week.

Overall, practices were using 40 of the more than 80 algorithms currently cleared by the Federal Food and Drug Administration, along with 27 in-house tools. In aggregate, these “home-grown” solutions had a slightly higher utilization rate (9.8%) compared to any one commercially available algorithm (9%), researchers explained Tuesday in JACR.

FDA-cleared algorithms for screening mammography were used most often, followed by those targeting pulmonary embolus, MR brain exams and brain hemorrhage. The findings are part of the ACR Data Science Institute’s first annual survey.

“Unfortunately, a large majority of the FDA-cleared algorithms have not been validated across a large number of sites, raising the possibility that patient and equipment bias could lead to the inconsistent performance observed by 73% of the survey respondents,” Bibb Allen Jr., MD, CMO of the ACR-DSI and a radiologist at Grandview Medical Center in Alabama, and colleagues explained.

For their survey, the authors sent a short email questionnaire to ACR members, collecting responses between April and May 2020. Overall, 1,427 radiologists responded, with 366 who used AI in their practice completing a follow-up survey.

While the hype around AI in medical imaging continues to grow, only about 30% of rads are currently using such tools as part of their daily clinical work. Larger practices were more likely to use AI than smaller organizations, according to the survey.

Among those not using AI, 80% said they “see no benefit,” and nearly one-third reported they could not justify the expense or that purchasing tools was beyond their control. What’s more, 72% have no plans to buy AI tools while approximately 20% want to do so within the next five years. 

“Understanding how radiologists are using AI … and monitoring these trends over time will be critical to aid the development of AI algorithms that help improve medical care,” the authors explained. “Ultimately, meeting the clinical needs of radiologists will drive innovation … that will provide needed information, can be efficiently implemented into daily workflow, and have the potential to improve the quality and efficiency of patient care.

Read more from the survey published April 20 in the Journal of the American College of Radiology.

Around the web

Its testing slate will kick off with the diagnostic and interventional radiology subspecialty exams in nuclear imaging on Jan. 10. 

EZaccessMD offers teleconsults with board-certified physicians coupled with mobile diagnostics, hoping to help health plan holders limit costs on facility-based visits. 

A new survey of ortho experts offers key clues to help improve musculoskeletal radiology services and interdisciplinary collaboration. 

Psychology researchers have used machine learning to wring useful two-year dementia trajectory predictions from more than 500 potentially contributing risk factors.

Trimed Popup