Abstract TP289: Workflow Time Reduction to Reperfusion in Anterior Large Vessel Occlusion Using a Non-Contrast CT Based Screening Solution

Abstract

The purpose of this study is to evaluate the reduction in time from hospital arrival to endovascular treatment (EVT) when using a non-contrast CT (NCCT)-based artificial intelligence (AI) solution to screen and notify clinicians of patients with emergent large vessel occlusion (ELVO). In real-world clinical practice, NCCT is primarily used to determine the presence of cerebral hemorrhage. However, if ELVO-suspected patients can be screened and flagged at the NCCT stage, it may improve clinical outcomes by reducing treatment delays. To evaluate clinical effectiveness, we compared the time from emergency room (ER) arrival to reperfusion before and after implementation of the AI-based triage and notification system. Patients aged over 19 years who visited a thrombectomy-capable stroke center with acute stroke symptoms and underwent EVT were included. Post-AI group data were prospectively collected after the implementation of the AI solution (May 1, 2022 – December 31, 2023), while a control group (Pre-AI) was retrospectively selected (May 1, 2020 – April 30, 2022) using 1:3 propensity score matching based on age, sex, and NIHSS score. The primary endpoint was the time from ER door to EVT. Secondary endpoints included time intervals from ER door to CT scan, CT scan to stroke team treatment (STT), and STT to EVT. Time differences between groups were analysed using an unpaired t-test with Welch's correction. A total of 25 Pre-AI cases and 70 Post-AI cases were analysed. The primary outcome showed a significant reduction in time from ER to EVT in the Post-AI group (147.7 ± 31.6 min) compared to the Pre-AI group (174.7 ± 75.0 min, p = 0.0155). Among the secondary endpoints, only the time from CT scan to STT was significantly shorter in the Post-AI group (20.2 ± 7.9 min vs. 35.4 ± 41.3 min, p = 0.0043). These findings demonstrate that early screening and clinician notification of ELVO patients via an AI solution at the initial stage of the clinical workflow can significantly reduce time to EVT and may positively impact patient outcomes. This study demonstrated how AI improved the hyperacute endovascular treatment workflow, by showing the impact on reducing door-to-reperfusion time. It will be particularly valuable in remote regions where clinical experts may be limited.

Ähnliche Arbeiten