A novel method to assess pial collateralization from stroke perfusion MRI: subdividing T_max into anatomical compartments

To develop and validate a quantitative and observer-independent method to evaluate pial collateral circulation by DSC-perfusion MRI and test whether this novel method delivers diagnostic information which is redundant to or independent from conventional penumbra imaging by the mismatch approach. We retrospectively identified 47 patients with M1 occlusion who underwent MR diffusion/perfusion imaging and mechanical thrombectomy at our facility. By automated registration and segmentation, T_max delays were attributed specifically to the pial, cortical and parenchymal compartments. The resulting pial volumes at delay were defined as the pial T_max map-assessed collateral score (TMACS) and correlated with gold standard digital subtraction angiography (DSA). Mismatch ratio was assessed by conventional penumbra defining MRI criteria. Strong correlation was found between TMACS and angiographically assessed collateral score (Pearson ρ = -0.74, p < 0.001). In multiple logistic regression, both good collaterals according to TMACS [OR 4.3 (1.1-19, p = 0.04)] and mismatch ratio ≥ 3.5 [OR 12.3 (1.88-249, p = 0.03)] were independent predictors of favourable clinical outcome. Perfusion delay in the pial compartment, as evaluated by TMACS, closely reflects the extent of pial collaterals in gold-standard DSA. TMACS and mismatch ratio were found to be complementary predictors of a favourable clinical outcome, each adding independent predictive information. (orig.)