Inter-rater Agreement in Three Perfusion-Computed Tomography Evaluation Methods before Endovascular Therapy for Acute Ischemic Stroke

Purpose: There is ongoing debate on which method of perfusion computed tomography (PCT) evaluation in ischemic stroke is the most appropriate for improved selection of patients for endovascular treatment. We sought to test different assessment methods for inter-rater reliability. Methods: Twenty-six patients were enrolled prospectively before endovascular therapy for acute anterior circulation ischemic stroke. Three raters experienced in stroke imaging and blinded to other imaging and clinical information independently analyzed 22 technically successful PCT scans according to 3 prespecified assessment methods applied to cerebral blood flow (CBF)/cerebral blood volume (CBV) and time-to-peak (TTP) maps: (1) visual mismatch estimate (VME), (2) Alberta Stroke Program Early CT Score perfusion method (ASPECTS-PCT), and (3) quantitative perfusion ratios (qPRs): R-CBF, R-CBV, R-TTP. Inter-rater agreement was assessed with Cohen's kappa, intraclass correlation coefficients (ICC), Bland-Altman plots, and global and descriptive statistics. Results: Significant differences between raters were found with VME and ASPECTS-PCT (P < .001) but with qPRs only for CBV (P = .03). Inter-rater agreement for VME was at best moderate by kappa statistics (.51); moderate by ICC for all parametric maps of ASPECTS-PCT (.56-.62), strong for R-TTP (.76), and excellent for R-CBF (.92) and R-CBV (.86). Pairwise comparisons revealed less scattering of individual values with qPRs and less deviation of mean differences from 0, suggesting minor systematic deviation by any 1 rater as compared with VME or ASPECTS-PCT. Conclusion: PCT evaluation methods used before endovascular therapy for acute anterior circulation stroke are subject to substantial inter-rater disagreement. QPRs in PCT evaluation had better inter-rater reliability than the often used VME and ASPECTS-PCT assessment.