CT angiography source-images and CT perfusion: Are they complementary tools for ischemic stroke evaluation? Response to: Reliability of CT perfusion in the evaluation of ischaemic penumbra

We read with great interest the study of Alves et al. 1 demonstrating that the qualitative evaluation of CBV and MTT maps obtained with computed tomography perfusion (CTP) imaging may overestimate the true ischemic penumbra in ischemic stroke. The authors conclude that CBV abnormalities are reliable markers for irreversible ischemia. They demonstrate that CBV may, rarely, overestimate the ischemic “core” and that MTT tends to overestimate the extent of final infarct areas, as these values do not differentiate the true “at risk” penumbra from benign oligemia. The authors also highlight some important methodological considerations related to the use of perfusion analysis software. Software applications from different vendors do not generate equivalent quantitative perfusion results. Caution should thus be exercised when interpreting quantitative/qualitative CTP measures, because these values may vary considerably, depending on the post-processing software. The major mathematical technique utilized for calculation of perfusion parameters is the deconvolution approach. Deconvolution can be performed using several methods. The classic deconvolution method is termed “standard singular value decomposition” (sSVD). This technique is robust, and its results are independent of the underlying vascular anatomy (i.e. intra/extracranial stenoses and/or leptomeningel collaterals). However, sSVD is sensitive to delays in the arrival of the contrast agent. When such delays occur, MTT and CBV are overestimated and CBF is underestimated. The use of abnormal map values may lead to overestimation of the ischemic penumbra by including non-ischemic brain regions into delayed contrast agent arrival calculations. To overcome these difficulties, new deconvolution methods have been developed to minimize the effects of bolus delay and dispersion, including delay-corrected deconvolution (dSVD) and block-circulant deconvolution (bSVD) algorithms. As a general rule, the use of a delay-insensitive deconvolution method is recommended when imaging stroke patients, a population in whom arterial stenoses are common.