Quantitative perfusion measurements using pulsed arterial spin labeling: Effects of large region-of-interest analysis

Purpose: To study arterial spin labeling (ASL) MRI techniques and to investigate various problematic issues that still hinder the accurate and robust quantitative analysis of ASL data. Materials and Methods: A pulsed-ASL (PASL) sequence was implemented on a 3-T imaging system and a protocol was developed for the measurement of perfusion based on fitting to a standard kinetic model. Both numerical simulations and multi-inversion time MRI data were analyzed. The effect of fitting a kinetic curve to a large region of interest (ROI) with a distribution of arterial transit times was compared to a pixel-by-pixel (PBP) method. Results: It was found that a significant underestimation of perfusion of approximately 17 +/- 6% (P < 0.001) occurs in gray matter, when comparing an ROI with a PBP analysis over a group of 12 healthy subjects. Conclusion: Analysis of ASL data based on a large ROI may suffer from inaccuracies arising from a distribution of transit times, implying that averaging of ASL kinetic data over such regions should therefore be avoided. When possible, a PBP fit should be performed. (c) 2005 Wiley-Liss, Inc.