Is the optimal Tmax threshold identifying perfusion deficit volumes variable across MR perfusion software packages? A pilot study.

Abstract

Accurate quantification of ischemic core and ischemic penumbra is mandatory for late-presenting acute ischemic stroke. Substantial differences between MR perfusion software packages have been reported, suggesting that the optimal Time-to-Maximum (Tmax) threshold may be variable. We performed a pilot study to assess the optimal Tmax threshold of two MR perfusion software packages (A: RAPID®; B: OleaSphere®) by comparing perfusion deficit volumes to final infarct volumes as ground truth. The HIBISCUS-STROKE cohort includes acute ischemic stroke patients treated by mechanical thrombectomy after MRI triage. Mechanical thrombectomy failure was defined as a modified thrombolysis in cerebral infarction score of 0. Admission MR perfusion were post-processed using two packages with increasing Tmax thresholds (≥ 6s, ≥ 8s and ≥ 10s) and compared to final infarct volume evaluated with day-6 MRI. Eighteen patients were included. Lengthening the threshold from ≥ 6s to ≥ 10s led to significantly smaller perfusion deficit volumes for both packages. For package A, Tmax ≥ 6s and ≥ 8s moderately overestimated final infarct volume (median absolute difference: -9.5mL, interquartile range (IQR) [-17.5; 0.9] and 0.2mL, IQR [-8.1; 4.8], respectively). Bland-Altman analysis indicated that they were closer to final infarct volume and had narrower ranges of agreement compared with Tmax ≥ 10s. For package B, Tmax ≥ 10s was closer to final infarct volume (median absolute difference: -10.1mL, IQR: [-17.7; -2.9]) versus -21.8mL (IQR: [-36.7; -9.5]) for Tmax ≥ 6s. Bland-Altman plots confirmed these findings (mean absolute difference: 2.2mL versus 31.5mL, respectively). The optimal Tmax threshold for defining the ischemic penumbra appeared to be most accurate at ≥ 6s for package A and ≥ 10s for package B. This implies that the widely recommended Tmax threshold ≥ 6s may not be optimal for all available MRP software package. Future validation studies are required to define the optimal Tmax threshold to use for each package.