Globally, stroke is a leading cause of death and disability, and accurate and timely diagnosis of large vessel occlusion (LVO) is essential for positive outcomes. We present our robust deep learning-based method for detecting both internal carotid artery (ICA) and middle cerebral artery (MCA) large vessel occlusions (LVO) from computed tomography angiography (CTA) scans. Our proposed two LVO detection models achieved an overall combined accuracy of 90.9% with a sensitivity of 89.8% and specificity of 91.4%. Further, the proposed model is lower in computational complexities and produces the results in less than 40 seconds, validating its adequacy for deployment in the clinical workflow.
