Selection of Arterial Input Function for Postprocessing of Cerebral CT Perfusion in Chronic Unilateral High-grade Stenosis or Occlusion of the Carotid or Middle Cerebral Artery

Abstract

We evaluated the effect of the arterial input function (AIF) on computed tomography perfusion (CTP) in patients with unilateral high-grade stenosis or occlusion in the carotid artery or middle cerebral artery without acute stroke. CTP datasets were retrospectively postprocessed using the same venous output function and different AIF selections: the second segment of the anterior cerebral artery (A2 AIF), the second segment of the middle cerebral artery (MCA) on the lesion side (affected M2 AIF), and M2 on the contralateral side (nonaffected M2 AIF). We measured CTP values in the region of interest (ROI) in the bilateral MCA territory and evaluated the lesion-to-contralateral ratios. The mean and standard deviations of cerebral blood flow (CBF) on the normal side were similar to previously reported data only when using "non-affected M2 AIF." Selecting an "affected M2 AIF" overestimated the CBF and shortened the mean transit time (MTT) in normal and lesion areas. Selecting an "A2 AIF" may cause overestimation of CBF in the normal side in patients with nonaffected-side A1 hypoplasia or occlusion. The sensitivity of the CBF ratio or MTT ratio to detect these unilateral cerebrovascular diseases was 100% using "nonaffected M2 AIF for bilateral MCA ROIs" and 70% (CBF ratio) and 90% (MTT ratio) using "respective AIF." The use of "nonaffected AIF for the bilateral MCA ROIs" was found to be the best of these AIF-ROI combinations in patients with chronic unilateral carotid or M1 severe stenosis or occlusion.