Despite growing excitement around artificial intelligence in imaging, only about 30% of radiologists are currently using the technology as part of regular clinical work, according to survey results released Tuesday.
Larger practices are more likely to deploy AI than their smaller counterparts, with most tapping it for aid in interpretation. Intracranial hemorrhage, pulmonary emboli and mammographic abnormalities are some of the most common targets, the American College of Radiology’s Data Science Institute found in its recent poll of the field.
“Concerns over inconsistent performance, potential decrease in productivity, and lack of reimbursement have made the actual use of AI in clinical practice modest,” first author Bibb Allen Jr., MD, chief medical officer of the institute, and colleagues wrote April 20 in JACR. “And perhaps not unexpectedly, larger practices, many of which may be academic centers with dedicated data science research activity, are more likely to use AI than smaller ones,” they added.
For the Data Science Institute’s inaugural annual survey, Allen et al. sent a brief electronic survey to ACR members via email, with info gathered between April and May of 2020. All told, 1,427 radiologists responded to the initial survey, while 366 who said they used AI in practice filled out a follow-up survey
Along with larger entities, about 20% of practices with fewer than five members said they use AI. Out of those not deploying the technology, 80% said they “see no benefit.” One-third of respondents said they cannot justify the expense, or that the purchasing decision was beyond their control. Some expressed concerns about decreased productivity. More than 70% of respondents hold no plans to pay for AI, while 20% see themselves doing so within the next five years.
“Radiologists using AI believe AI adds value to their clinical practice, but due to inconsistent performance, they would not trust AI for autonomous clinical use,” the authors noted. “As barriers to implementation are mitigated, there is opportunity for future growth of AI in clinical practice,” they added.
You can read much more about the survey in the Journal of the American College of Radiology here.