Background and purpose: Perfusion-weighted imaging-derived maps of time-to-maximum (Tmax) are increasingly used to identify the tissue at risk in clinical stroke studies (eg, DEFUSE and EPITHET). Using quantitative positron emission tomography (PET), we evaluated Tmax to define the penumbral flow threshold in stroke patients and compared its performance to nondeconvolved time-to-peak (TTP) maps.
Methods: Comparative perfusion-weighted imaging and quantitative 15O-water PET images of acute stroke patients were analyzed using cortical regions of interest. A receiver-operating characteristic curve analysis described the threshold independent performance of Tmax (area under the curve) and identified the best threshold (equal sensitivity and specificity threshold) to identify penumbral flow (< 20 mL/100 g/min on PET cerebral blood flow). The results were compared with nondeconvolved TTP and other current perfusion-weighted imaging maps using the Mann-Whitney rank-sum test.
Results: In 26 patients (time delay between MRI and PET, 65 minutes), the best threshold for penumbral flow was 5.5 seconds for Tmax (median; interquartile range, 3.9-6.6; sensitivity/specificity, 88%/89%). The area under the curve value was 0.95 (median; interquartile range, 0.93-0.97). Deconvolved Tmax did not perform significantly better than TTP (P = 0.34).
Conclusions: Maps of Tmax detected penumbral flow but did not perform better than the easy-to-obtain maps of nondeconvolved TTP. Thus, "simple" TTP maps still remain suitable for clinical stroke studies if detailed postprocessing is not feasible.