Comparison of Computed Tomography Perfusion and Magnetic Resonance Imaging Perfusion-Diffusion Mismatch in Ischemic Stroke

Abstract

Background and Purpose— Perfusion imaging has the potential to select patients most likely to respond to thrombolysis. We tested the correspondence of computed tomography perfusion (CTP)-derived mismatch with contemporaneous perfusion-diffusion magnetic resonance imaging (MRI). Methods— Acute ischemic stroke patients 3 to 6 hours after onset had CTP and perfusion-diffusion MRI within 1 hour, before thrombolysis. Relative cerebral blood flow (relCBF) and time to peak of the deconvolved tissue residue function (Tmax) were calculated. The diffusion lesion (diffusion-weighted imaging) was registered to the CTP slabs and manually outlined to its maximal visual extent. Volumetric accuracy of CT-relCBF infarct core (compared with diffusion-weighted imaging) was tested. To reduce false-positive low CBF regions, relCBF core was restricted to voxels within a relative time-to-peak (relTTP) >4 seconds for lesion region of interest. The MR-Tmax >6 seconds perfusion lesion was automatically segmented and registered to CTP. Receiver-operating characteristic analysis determined the optimal CT-Tmax threshold to match MR-Tmax >6 seconds. Agreement of these CT parameters with MR perfusion-diffusion mismatch in coregistered slabs was assessed (mismatch ratio >1.2, absolute mismatch >10 mL, infarct core <70 mL). Results— In analysis of 49 patients (mean onset to CT, 213 minutes; mean CT to MR, 31 minutes), constraining relCBF <31% within the automated relTTP perfusion lesion region of interest reduced the median magnitude of volumetric error (vs diffusion-weighted imaging) from 47.5 mL to 15.8 mL ( P <0.001). The optimal CT-Tmax threshold to match MR-Tmax >6 seconds was 6.2 seconds (95% confidence interval, 5.6–7.3 seconds; sensitivity, 91%; specificity, 70%; area under the curve, 0.87). Using CT-Tmax >6 seconds “penumbra” and relTTP-constrained relCBF “core,” CT-based and MRI-based mismatch status was concordant in 90% (kappa=0.80). Conclusions— Quantitative CTP mismatch classification using relCBF and Tmax is similar to perfusion-diffusion MRI. The greater accessibility of CTP may facilitate generalizability of mismatch-based selection in clinical practice and trials.