Semi-automated mapping of occluded arterial segments in acute large vessel stroke from computed tomography angiography

Abstract

Catheterisation of occluded artery is the most crucial step in the mechanical thrombectomy of large vessel occlusions. Being a blind procedure, potential major untoward events could be anticipated at this stage. In this study, we evaluated the feasibility of reconstruction of the occluded vascular segment from CT angiography, exploiting the attenuation difference between the clot and adjacent brain parenchyma. Twenty-five patients with large vessel occlusion who achieved TICI2B/3 recanalization were identified. The semi-automated reconstruction was done using an open-source 3D segmentation software by two observers. The generated arterial anatomy was categorized as congruent, if all the major arterial segments namely, internal carotid artery, M1 middle cerebral artery, bifurcation, and major divisions matched with post thrombectomy angiograms. The discrepancy of one or more than one arterial segment was classified as mild or total mismatch respectively. Congruent arterial mapping was possible in 88% and 92 % patients for observer 1 and observer 2 respectively, while mild mismatch was noted in two patients for both the observers. Mismatch mostly occurred in superior division for both observers(100% vs 66.6%). The interrater agreement was good (0.77) and the accuracy was not influenced by the length of the clot. The median time for reconstruction was 4 minutes for both the observers (p - 0.21). Delineation of the anatomy of the occluded major intracranial artery is possible and corresponds with the actual anatomy. It could be used as a guide during thrombectomy procedures.