Correlation of volumetric mismatch and mismatch of Alberta Stroke Program Early CT Scores on CT perfusion maps

We aimed to determine if volumetric mismatch between tissue at risk and tissue destined to infarct on computed tomography perfusion (CTP) can be described by the mismatch of Alberta Stroke Program Early CT Score (ASPECTS).

Forty patients with nonlacunar middle cerebral artery infarct <6 h old who had CTP on admission were retrospectively reviewed. Two raters segmented the lesion volume on mean transit time (MTT) and cerebral blood volume (CBV) maps using thresholds of >6 s and <2.0 mL per 100 g, respectively. Two other raters assigned ASPECTS to the same MTT and CBV maps while blinded to the volumetric data. Volumetric mismatch was deemed present if >or=20%. ASPECTS mismatch (=CBV ASPECTS - MTT ASPECTS) was deemed present if >or=1. Correlation between the two types of mismatches was assessed by Spearman's coefficient (rho). ROC curve analyses were performed to determine the optimal ASPECTS mismatch cut point for volumetric mismatch >or=20%, >or=50%, >or=100%, and >or=150%.

Median volumetric mismatch was 130% (range 10.9-2,031%) with 31 (77.5%) being >or=20%. Median ASPECTS mismatch was 2 (range 0-6) with 26 (65%) being >or=1. ASPECTS mismatch correlated strongly with volumetric mismatch with rho = 0.763 [95% CI 0.585-0.870], p < 0.0001. Sensitivity and specificity for volumetric mismatch >or=20% was 83.9% [95% CI 65.5-93.5] and 100% [95% CI 65.9-100], respectively, using ASPECTS mismatch >or=1. Volumetric mismatch >or=50%, >or=100%, and >or=150% were optimally identified using ASPECTS mismatch >or=1, >or=2, and >or=2, respectively.

On CTP, ASPECTS mismatch showed strong correlation to volumetric mismatch. ASPECTS mismatch >or=1 was the optimal cut point for volumetric mismatch >or=20%.