Evaluating the Efficacy of Artificial Intelligence-Based Models in Large Vessel Occlusion Detection: A Systematic Review
Anusuiya Bhorkar
Dominican Academy, New York, U.S.
Publication date: July 10, 2026
Dominican Academy, New York, U.S.
Publication date: July 10, 2026
DOI: http://doi.org/10.34614/JIYRC2026I21
ABSTRACT
During an ischemic stroke, the brain ages approximately 3.6 years for every hour post-onset, making closing the gap between stroke onset and treatment time imperative. Large vessel occlusions (LVOs) account for a significant proportion of ischemic strokes worldwide, making rapid LVO detection critical. Recent advancements in artificial intelligence (AI) have enabled the development of models to detect LVOs, yet questions remain about their effectiveness. This systematic review synthesizes evidence from selected studies evaluating the performance of AI-based models for LVO detection. The reviewed literature suggests that AI has strong potential to support stroke diagnosis, particularly with convolutional neural network (CNN) architectures, and can aid healthcare professionals. However, current models demonstrate inconsistent LVO detection accuracy across different arteries and are unable to achieve the reliability required for autonomous clinical integration. This review will investigate the selected studies to assess areas for positive performance and necessary future improvement.
During an ischemic stroke, the brain ages approximately 3.6 years for every hour post-onset, making closing the gap between stroke onset and treatment time imperative. Large vessel occlusions (LVOs) account for a significant proportion of ischemic strokes worldwide, making rapid LVO detection critical. Recent advancements in artificial intelligence (AI) have enabled the development of models to detect LVOs, yet questions remain about their effectiveness. This systematic review synthesizes evidence from selected studies evaluating the performance of AI-based models for LVO detection. The reviewed literature suggests that AI has strong potential to support stroke diagnosis, particularly with convolutional neural network (CNN) architectures, and can aid healthcare professionals. However, current models demonstrate inconsistent LVO detection accuracy across different arteries and are unable to achieve the reliability required for autonomous clinical integration. This review will investigate the selected studies to assess areas for positive performance and necessary future improvement.
