Automated collateral assessment restricted to the hypoperfused area for distal vessel occlusions in ischemic stroke

Objectives This study aims to: (1) develop and evaluate a quantitative assessment of collateral status in the downstream area of an occluded intracranial artery in acute ischemic stroke and compare this method to middle cerebral artery (MCA)-based assessment; (2) determine the agreement between the automated occlusion-downstream area collateral score (ODACS) and expert raters' assessments, and compare this to inter-rater agreement. Methods Patients from MR CLEAN-NO IV and MR CLEAN Registry with a proximal M1, distal M1, or M2 occlusion were included. Using the hypoperfused area from CT perfusion (CTP) as a proxy for the occlusion-downstream territory and automated vessel segmentations from CT angiography (CTA), ODACS is calculated as the vessel volume ratio between downstream ipsilateral and its contralateral regions. ODACS was compared to a whole MCA-territory approach and evaluated against visual scoring by two expert raters that visually estimated ODACS using CTA and CTP, and their inter-rater agreement. Results The study included 204 patients with a proximal M1 (52%), distal M1 (32%), or M2 (16%) occlusion. ODACS yielded lower collateral scores than MCA-based scoring for all occlusion locations, with larger differences in more distal occlusions. For M2 occlusions, 58% of patients shifted from good (> 50%) to poor (<= 50%) collateral filling of the occluded territory using ODACS. Moderate (weighted Cohen's kappa kappa = 0.45) inter-rater agreement and fair (kappa = 0.35) to moderate (kappa = 0.51) ODACS-rater agreement were observed. Conclusions ODACS yields lower collateral scores compared to MCA-based scoring and is comparable to scores from expert raters.