Combined Blood:Brain Partition Coefficient and Perfusion Assessment using QUASAR

E. T. Petersen, K. C. van de Ven, I. Zimine, X. Golay Neuroradiology, National Neuroscience Institute, Singapore, Singapore, Biomedical Engineering, Eindhoven University of Technology, Eindhoven, Netherlands, Singapore Bioimaging Consortium, ASTAR, Singapore, Singapore INTRODUCTION: Perfusion is vital for the homeostasis and thereby survival of an organ and as such is a very important physiological parameter. MRI based perfusion methods include, among others, fully non-invasive arterial spin labeling (ASL) techniques. ASL is useful in the clinical evaluation of vascular diseases and it has gained great popularity within the basic neuroscience, mainly due to its potential for measuring quantitative perfusion. Also the ability of doing repetitive scans is an important asset of this technique for functional neuro-imaging. However, a few problematic factors still exist regarding the absolute quantification of cerebral blood flow (CBF). These include the estimation of the blood equilibrium magnetization M0b, arrival time and shape of the bolus. While different methods have dealt with the latter two [1,2], correct estimation of M0b still remains a source of error. Basically, it is needed in order to scale the acquired ∆M images from the control and label experiment, in order to obtain the flow. Unfortunately, acquiring the value is not trivial, neither within the ASL experiment itself nor in a separate scan. This is mainly due to partial volume effects with the surrounding tissue and pulsating behavior of the blood. In practice, M0b is most often extracted from the sagittal sinus or by using the blood-brain partition coefficient λ, which is defined as the ratio between water content in blood and tissue (M0b = M0t/λ). The former suffers from partial volume problems as well as the fact that R2* differs between the intended arterial blood and measured venous blood magnetization [1]. In single inversion time-point experiments, the signal from the control experiment is often taken as M0t after correction for the TR and expected or measured T1 of the tissue. M0b is then extracted assuming the average partition coefficient for the brain λ = 0.9 [3]. However, the water density varies from region to region and in between tissues (λwm = 0.82, λgm = 0.98 [3]), thereby violating the assumption of a single partition coefficient for all tissues. Here we propose a method for estimating λ on a voxel by voxel basis, based on the multi TI data acquired using the QUASAR sequence [1].