AI-driven radiology reading worklist management reduces report turnaround for patients positive for pulmonary embolism (PE) on CT pulmonary angiography (CTPA), according to a July 26 study published in the American Journal of Roentgenology.
The study findings could translate to better patient outcomes, wrote a team led by Kiran Batra, MD, of the University of Texas Southwestern Medical Center in Dallas.
"Timely intervention, such as initiation of anticoagulation, is critical to optimize clinical outcomes among patients with acute PE," the group explained. "Therefore, it is important that positive CTPA results be communicated to ordering physicians in a timely manner."
Time is of the essence for patients with acute PE, Batra and colleagues wrote. AI has been used to automate the reprioritization of radiologist worklists for detecting intracranial hemorrhage, resulting in reduced report turnaround times; the investigators sought to explore whether AI could have the same effect on reports regarding acute PE on CTPA.
The team conducted a study that included 2,501 CTPA examinations of 2,197 patients who underwent the scans during a pre-AI period (October 2018 to March 2019, 1,335 exams) and a post-AI period (October 2019 to March 2020, 1,166 exams). Exams were performed during regular working hours; the AI algorithm prioritized the CTPA exams on radiologists' reading worklists if CTPA identified acute PE.
The group used electronic medical record data and dictation timestamps to assess the following metrics for both before and after implementation of the AI tool:
- Wait time (time from exam completion to the initiation of the report)
- Read time (time from initiation of the report initiation to its availability)
- Report turnaround time (sum of wait and read times)
The authors found a frequency of acute PE in the pre-AI period of 15.1% and a frequency of 12.3% during the post-AI period. After the AI worklist tool was implemented, it reprioritized 148 exams (12.7%).
The team found the AI worklist tool had the following effect on workflow measures:
| Effect of AI tool on worklist measures for PE-positive CTPA exams | |||
| Measure | Pre-AI | Post-AI | Mean difference |
| Mean report turnaround time | 59.9 minutes | 47.6 minutes | 12.3 minutes |
| Mean wait time | 33.4 minutes | 21.4 minutes | 12 minutes |
There was no significant difference in mean read time in either period: pre-AI read time was 26.5 minutes and post-AI read time was 26.3 minutes.
The bottom line? Using AI to make reading CTPA exams that have identified PE improves patient care.
"By assisting radiologists in providing rapid diagnoses, the AI tool has potential for enabling earlier interventions for acute PE," the authors concluded.
The complete study can be found here.
