Automated Large Artery Occlusion Detection in Stroke: A Single-Center Validation Study of an Artificial Intelligence...

Area of Science:

• Neurology
• Medical Imaging
• Artificial Intelligence

Background:

• Timely identification of large vessel occlusions (LVOs) in acute ischemic stroke (AIS) is crucial for effective treatment.
• Delays in LVO recognition can lead to poorer patient outcomes due to limited expert availability.

Purpose of the Study:

• To evaluate the diagnostic performance of an artificial intelligence (AI) algorithm for detecting LVOs in AIS patients.
• To assess the AI's accuracy in identifying occlusions within specific cerebral arteries.

Main Methods:

• Retrospective analysis of 610 computed tomography angiographies (CTAs) from AIS patients.
• The Viz-LVO Algorithm® was used to detect LVOs from the internal carotid artery terminus (ICA-T) to the Sylvian fissure.
• Performance was compared to human expert grading using test characteristic analysis and Cohen's kappa.

Main Results:

• The AI algorithm achieved 87.9% accuracy (87.6% sensitivity, 88.5% specificity) for ICA-T and/or MCA-M1 LVOs (AUC 0.88).
• Cohen's kappa was 0.74, indicating substantial agreement.
• For secondary analysis including MCA-M2 occlusions, accuracy was 82.7%.

Conclusions:

• AI-based LVO detection offers a rapid and accurate method for identifying critical findings in AIS.
• Automated analysis facilitates prompt notification to emergency teams, supporting timely treatment decisions.
• This AI tool can expedite reperfusion therapies or transfers to specialized stroke centers.