AI dramatically reduces radiologists’ rate of missed incidental pulmonary embolism on routine CT
Commercially available artificial intelligence software can dramatically reduce radiologists’ rate of missed incidental pulmonary embolism on routine CT scans, according to a new analysis published in Radiology: Cardiothoracic Imaging.
This AI-assisted workflow prioritization tool has demonstrated high diagnostic accuracy while freeing up time in busy providers’ schedules. Timely identification of this concern is crucial, given the increased risk of poor patient outcomes.
“PE is one of the diagnoses that is most commonly missed or delayed by physicians. Detection of [incidental pulmonary embolism] on routine contrast-enhanced chest CT scans can be especially challenging when the IPE is small and isolated,” Laurens Topff, MD, with the Department of Radiology at the Netherlands Cancer Institute, and co-authors wrote April 20. “We demonstrated that commercially available AI software had high diagnostic accuracy in the detection of [incidental] PE on chest CT scans in patients with cancer and was effective in significantly reducing the time to diagnosis of positive examinations compared with the routine workflow in a setting with a backlog of unreported scans,” they added later.
The study sample included a total of 11,736 contrast-enhanced CT scans from 6,447 oncology patients treated at a comprehensive cancer center between 2019-2020. The study was split into three separate periods with (1) radiologists initially having no AI assistance or special instructions while reporting followed by (2) no aid but instructions to screen for possible incidental PE. During this second phase, a staff radiologist performed human triage, immediately reporting those with PE, while the rest were returned to a work list. And finally, (3) rads had help from AI to prioritize their work lists. Researchers left 15-week intervals between the three phases to allow radiologists time to grow accustomed to the software, provided by Israel-based Aidoc.
Prevalence of incidental PE varied between 1% to 1.4% during the three phases. The software logged high diagnostic accuracy for this concern, including 91.6% sensitivity, 99.7% specificity and a negative predictive value of 99.9%. In clinical practice with a backlog of unreported exams, AI was able to reduce the median detection and notification time for incidental PE in flagged scans from “several days” down to just one hour. Meanwhile, the missed rate of IPE dropped from 44.8% down to 2.6% when radiologists were aided by AI. About 37.8% of IPE-positive scans showed emboli in the main or lobar pulmonary arteries, with the benefit of timely assessment and treatment “most evident in these patients,” the authors noted.
Topff et al. cautioned that their study has several limitations, and future studies are needed to investigate the effect of early diagnosis of IPE on morbidity and mortality.
“The intended use of the investigated AI software is limited to workflow triage and not diagnostics,” the authors cautioned. “Automated work list prioritization can assist radiologists, who remain responsible for verifying flagged examinations and must be aware of possible false-negative findings. Our study showed that a considerable number of scans (315 of 4294 [7.3%]) were not analyzed by the deployed AI software due to issues with data retrieval and technical validation, resulting in delayed diagnosis of IPE in these patients. This demonstrates the importance of monitoring and improving the yield of scans analyzed by AI software after deployment.”
The authors also emphasized that they are not affiliated with Aidoc or the vendor industry and had complete control of the data and information submitted.
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.