Clinically Meaningful MRI Perfusion Abnormalities in Acute Stroke: Comparison of Analytic Techniques
Background: MRI perfusion imaging (PWI) can demonstrate hemodynamic abnormalities in acute stroke. The volume of hypoperfusion derived from calculated perfusion parameter maps has been used to predict tissue at risk for infarction and to identify presumptive ischemic penumbra. It is unclear how best to distinguish true tissue at risk from benign hypoperfusion. A first step toward this goal is identifying clinically significant PWI abnormalities in stroke patients. Our purpose was to evaluate four different perfusion parameter maps to determine which algorithm best correlates with clinical severity. Methods: Twenty patients were retrospectively selected from our database. Selection criteria included 1) acute hemispheric lesion, 2) MRI within 24 hours of symptom onset, and 3) no history of prior stroke. Perfusion maps were derived using four different algorithms to estimate relative mean transit time (rMTT): 1) cerebral blood volume (CBV) / cerebral blood flow (CBF), 2) CBV / peak of the concentration-time curve, 3) time to peak (TTP), and 4) ratio of the 1st / 0th moment of the transfer function (first moment method). Abnormal perfusion volumes were derived from ever-increasing thresholds of rMTT delay relative to normal contralateral tissue. The volumes at each delay threshold were correlated with National Institutes of Health Stroke Scale (NIHSS) for each algorithm. Results: Significant correlations between hypoperfusion volumes and NIHSS were found for all algorithms. The first moment method had the highest correlation (r = 0.76) and the correlations for this method were independent of the delay threshold used to derive the volumes. For the other algorithms, the best correlations were observed for volumes including only voxels with delays of 4 seconds or greater. Conclusions: This analysis suggests that the first moment method may have advantages over the others in determining the correlation of hypoperfusion volume to NIHSS. Further analyses correlating acute hypoperfusion volumes to final infarct volumes may help refine the choice of best analytic method for determining clinically relevant PWI abnormalities.