Validation of CT perfusion based vascular territory mapping: correlation to visual pial grading and outcome measures.

Abstract

Background/Aims: Vascular territory mapping (VTM) software estimates which intracerebral vessel provides peak arterial flow to a brain voxel. We hypothesize that the VTM algorithm may correlate to visual measurements of leptomeningeal grading and stroke outcome measures. Method(s): Consecutive patients with acute ischemic stroke and M1 MCA occlusion were recruited from the INSPIRE database (International Stroke Perfusion Imaging Registry). VTM software assigned regions of the brain to an estimated feeding vessel (ACA, MCA, PCA). Whole brain dynamic CTA was used to detect retrograde filling from pial arteries in the MCA territory. Collateral grading scores were based on origin (ACA/PCA) and extent of retrograde flow. Patients were pooled to assess relationships between VTM variables, collateral grades, and clinical outcomes. Result(s): The final dataset included 115 patients with M1 occlusion. 64% had higher posterior collateral grading scores, 26% had equal grading and 25% had higher anterior grading. With regards to posterior collateral grade, ischaemic core and VTM-oPCA were significantly associated with decreased posterior collateral grade. An increase in VTM-oPCA was associated with an odds decrease in posterior collateral grade of 0.984 (95% CI, 0.99-1.0), p=0.001. VTM variables were not predictive of anterior collateral grade or stroke outcome measures. Conclusion(s): VTM volumes did not correlate with visualized collateral vessels as hypothesized. An unanticipated inverse association between VTM-oPCA and posterior collateral grade was identified. It is unlikely that VTM volumes correlate with directional visual pial collateral grade.