Challenges for Non-Invasive Brain Perfusion Quantification Using Arterial Spin Labeling

Abstract

Arterial Spin Labeling (ASL) sequences for perfusion Magnetic Resonance Imaging (MRI) have recently become available to be used in the clinical practice, offering a completely non-invasive technique for the quantitative evaluation of brain perfusion. Despite its great potential, ASL perfusion imaging still presents important methodological challenges before its incorporation in routine protocols. Specifically, in some pathological conditions in which the cerebrovascular dynamics is altered, the standard application of ASL may lead to measurement errors. In these cases, it would be possible to estimate perfusion, as well as arterial transit times, by collecting images at multiple time points and then fitting a mathematical model to the data. This approach can be optimized by selecting a set of optimal imaging time points and incorporating knowledge about the physiological distributions of the parameters into the model estimation procedures. In this study, we address the challenges that arise in the measurement of brain perfusion using PASL, due to variations in the arterial transit times, by estimating the errors produced using different types of acquisitions and proposing methods for minimizing such errors. We show by simulation that multiple inversion time ASL acquisitions are expected to reduce measurement errors relative to standard approaches. In data collected from a group of subjects, we further observed reduced inter-subject variability in perfusion measurements when using a multiple versus single inversion time acquisitions. Both measurement errors and variability were further reduced if optimized acquisition and analysis techniques were employed.