AI improves radiology workflow by standardizing reporting recommendations
In a profession where success hinges on quality imaging surveillance, a lack of universally accepted guidelines often gives way to inconsistent reporting in radiology rooms—but a group of clinicians thinks artificial intelligence could be cleaning up workflow.
“Inadequate imaging surveillance has been identified as the most significant contributor to abdominal aortic aneurysm (AAA) rupture,” first author Danny C. Kim, MD, and colleagues wrote in the Journal of the American College of Radiology. “Radiologists can contribute value to patient care and reduce morbidity and mortality related to AAA by incorporating evidence-based management recommendations.”
Of course, guidelines for radiologists exist, the authors wrote—the American College of Radiology and Society of Vascular Surgery have both published professional recommendations—but clinicians aren’t incorporating them in the impression of their reports.
“The challenges lie in achieving 100 percent radiologist compliance to incorporate the recommendations and ensuring that the patient is notified by their provider, the follow-up examination is scheduled and the patient returns for an imaging test that may be scheduled three to five years in the future,” Kim et al. said. “To address these barriers, radiology quality and informatics leads have harnessed IT solutions to facilitate integration of content, communication of results and patient follow-up.”
Half of the researchers hail from Johns Hopkins Radiology in Baltimore, and said their institution uses voice recognition software to level radiologists’ management recommendations. Johns Hopkins uses a program called PowerScribe, which generates a set of “pick lists," or standardized recommendations, for radiologists when they enter an incidental finding.
Researchers at the University of Maryland took that a step further, according to the study, by applying natural language processing to the work. With this, the radiologist can simply dictate findings to a computer, which in turn transcribes that text and generates appropriate recommendations.
New York University took a more team-based approach to the problem, merging faculty from both its radiology and vascular surgery departments to outline a set of standard practices, guidelines and follow-up recommendations for both branches. Recently, NYU Langone Health’s electronic medical record system was optimized to automatically identify radiology reports that contain any of these accepted guidelines.
“Creating standardized recommendations for radiology reports across medical centers is critical for radiology to demonstrate value by effectively advancing the guidance provided in the ACR White Papers,” Kim and co-authors wrote.
The authors said more streamlined management of radiologic findings would mean fewer unnecessary imaging exams, less exposure to harmful radiation and would protect patients from delayed imaging, surgery and biopsies.
“Regardless of practice venue, we encourage radiology quality leadership to follow suit by creating standardized report recommendations in consensus with their referring physician groups and developing performance feedback reports to drive adherence,” Kim and colleagues said. “Collaboration with referring physicians is critical, because other professional societies also have best practice guidelines to manage incidental findings and abnormalities requiring surveillance.”
After graduating from Indiana University-Bloomington with a bachelor’s in journalism, Anicka joined TriMed’s Chicago team in 2017 covering cardiology. Close to her heart is long-form journalism, Pilot G-2 pens, dark chocolate and her dog Harper Lee.