Radiology department tests ‘on premise’ AI solution with potential to reduce costs, simplify workflows
A radiology department recently explored the use of an “on premise” artificial intelligence solution with the potential to reduce costs and simplify workflows.
Recent technological advancements have allowed for the integration of AI when interpreting CT angiography exams, experts wrote Tuesday in Radiology. Such assistance often is necessary given rising volumes of CCTA scans necessary to assess chest pain—the primary cause of outpatient and ED visits, totaling over 10 million encounters annually.
“However, these systems are shifting toward off-premises solutions, raising concerns about processing delays, data security, and increased costs,” Fernando Kay, MD, associate chief of cardiothoracic imaging at UT Southwestern Medical Center in Dallas, and co-authors wrote May 13. “A new development in this area is an AI software prototype that offers comprehensive CCTA analysis on premises, enhancing control over data management and system integration.”
Kay and colleagues aimed to assess the performance of an on-premise, AI-based coronary artery calcium scoring and CCTA analysis software against expert physicians’ findings. They included consecutive patients undergoing CCTA for coronary analysis at their academic center between 2017 and 2021 across four scanners from three different vendors. The retrospective study utilized AI software provided by Siemens Healthineers (also a co-author of the study, with no direct financial sponsorship of the analysis).
The study included a total of 1,041 CCTA scans from 1,032 patients. On-premises AI was accurate in ruling out obstructive coronary artery disease and achieved “substantial to near perfect agreement” with human experts. This was based on Coronary Artery Disease Reporting and Data System (CAD-RADS) 2.0 categories for stenosis severity and plaque burden, the authors reported.
Although the algorithm showed high negative predictive value (a measure of how accurate a negative test result is), positive predictive value was low (65% for moderate-grade and 39% for high-grade stenosis). While on-premise AI has the potential to reduce certain costs for data storage and migration, it also could increase expenditures elsewhere, the authors cautioned.
“Such a low [positive predictive value] could lead to unnecessary downstream invasive procedures, patient anxiety, increased healthcare costs, and potential procedure-related complications stemming from false-positive CCTA results,” the authors advised. “However, AI-flagged CCTA studies would still undergo expert evaluation in a traditional clinical workflow. Conversely, the high [negative predictive value] offers substantial clinical value by helping to rule out clinically important coronary stenosis, thereby safely triaging patients. This will not only assist interpretation but also expedite discharge for low-risk patients and prioritize high-risk scans for timely expert review.”
Read much more about the results, including a corresponding editorial, in the Radiological Society of North America’s flagship journal.
Marty Stempniak has covered healthcare since 2012, with his byline appearing in the American Hospital Association's member magazine, Modern Healthcare and McKnight's. Prior to that, he wrote about village government and local business for his hometown newspaper in Oak Park, Illinois. He won a Peter Lisagor and Gold EXCEL awards in 2017 for his coverage of the opioid epidemic.