Abstract
BACKGROUND AND PURPOSE: In ischemic stroke, leptomeningeal collaterals can provide delayed and dispersed compensatory blood flow to tissue-at-risk despite an occlusion and can impact treatment response and infarct growth. The purpose of this work is to test the hypothesis that the inclusion of this delayed and dispersed flow with an appropriately calculated local arterial input function (local-AIF) is needed to quantify the degree of collateral blood supply in tissue distal to an occlusion. MATERIALS AND METHODS: Seven experiments were conducted in a preclinical middle cerebral artery occlusion model. Dynamic susceptibility contrast MRI was imaged and postprocessed to yield quantitative cerebral blood flow (qCBF) maps with both a traditionally chosen single arterial input function applied globally to the whole brain (ie, "global-AIF") and a delay and dispersion corrected AIF (ie, "local-AIF") that is sensitive to retrograde flow. Leptomeningeal collateral arterial recruitment was quantified with a pial collateral score from x-ray angiograms, and infarct growth was calculated from serially acquired diffusion-weighted MRI scans. RESULTS: The degree of collateralization at x-ray correlated more strongly with local-AIF qCBF in the ischemic penumbra (R(2) = 0.81) than with traditional global-AIF qCBF (R(2) = 0.05). Local-AIF qCBF was negatively correlated with infarct growth (slower infarct progression with higher perfusion, R(2) = 0.79) more strongly than global-AIF qCBF (R(2) = 0.02). CONCLUSIONS: In acute stroke, qCBF calculated with a local-AIF is more accurate for assessing tissue status and collateral supply than traditionally chosen global-AIFs. These findings support the use of a local-AIF that corrects for delayed and dispersed retrograde flow in determining quantitative tissue perfusion with collateral supply in occlusive disease.