Abstract WMP15: Hyperperfusion on Arterial Spin Labeling: Objective Decision Support Using Pattern Recognition

• DOI: 10.1161/str.47.suppl_1.wmp15
• Publication Date: 2021-07-03T12:32:07Z
• Authors: Fabien Scalzo; Songlin Yu; Sagar Patel; David Liebeskind; Danny Wang
• Keywords: Arterial spin labeling; Stroke

Background: Hyperperfusion detected on arterial spin labeling (ASL) images acquired after stroke onset has been shown to correlate with subsequent hemorrhagic transformation (HT). Presence of hyperperfusion is typically by visual review cerebral blood flow (CBF). Such a subjective as it challenged inter-reader variability, noise, and lack standard threshold. We present in this study quantitative detection model that can provide an objective decision support for the interpretation ASL CBF maps rapidly delineate regions. Methods: Cerebral (CBF) acute patients presenting occlusion MCA territory were coregistered standardized atlas space. To achieve reliable hyperperfusion, we formalize problem nonlinear classification relates regional voxel intensity values corresponding binary label (normal or hyperperfused). Our method takes into account healthy contralateral hemisphere its during determination voxel. Each input feature vector combines at interest, matched region, distribution difference between them. associated presence was manually established consensus experts. The predicted regions compared groundtruth two researchers. Results: A total 361 scans collected from 221 (age=72±17 years; 45% males). 76 subsequently used our analysis. An AUC 83±5% reached leave-one-out cross-validation, which corresponds accuracy detecting manual delineation maps. Conclusion: Pattern recognition based regression accurate measure could therefore improve patients.