Artificial Intelligence-Parallel Stroke Workflow Tool Improves Reperfusion Rates and Door-In to Puncture Interval

Area of Science:

• Neurology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Large-vessel occlusion (LVO) stroke requires rapid intervention.
• Viz.ai's AI software (Viz LVO) automates LVO detection from CT angiography.
• The software aims to expedite stroke team alerts for suspected LVOs.

Purpose of the Study:

• To evaluate the impact of Viz LVO software on door-in to puncture time.
• To assess if AI-driven triage improves treatment efficiency in comprehensive stroke centers (CSCs).

Main Methods:

• Retrospective chart review comparing pre- and post-AI implementation periods (Nov 2016 - May 2020).
• Analysis of door-in to puncture times for consecutive stroke patients at a CSC.
• Examination of demographics and outcomes using a prospectively collected database.

Main Results:

• The mean door-in to puncture time significantly decreased by 86.7 minutes after Viz LVO implementation (206.6 vs. 119.9 minutes, P<0.001).
• The rate of successful reperfusion (mTBI 2B-3) significantly improved in the post-AI cohort (P=0.036).
• Study included 86 patients pre-AI and 102 patients post-AI.

Conclusions:

• Viz LVO software incorporation significantly reduced treatment times in CSCs.
• The AI tool was associated with significantly higher rates of adequate reperfusion.
• Further prospective, multicenter studies are recommended to validate these findings in larger cohorts.