Feasibility of whole-brain dynamic contrast enhanced (DCE) MRI using 3D k-t PCA

INTRODUCTION: Quantification of cerebral blood flow using dynamic contrast enhanced (DCE) MRI [1] has several advantages over conventional dynamic susceptibility contrast (DSC) MRI [2]. Most importantly, DCE-MRI avoids the inherent susceptibility artifacts of DSC-MRI, thus allowing for more reliable measurement of the arterial input function (AIF) and quantification of blood brain barrier leakage in brain tumors [3]. However, DCE-MRI currently suffers from low signal-to-noise ratio (SNR) and limited spatial coverage (in the order of 3-4 slices per second). The long acquisition times of DCE-MRI are due to interleaved saturation recovery (SR) preparation pulses, as well as incompatibility with echo planar imaging (EPI) which adds an undesirable degree of T2*-weighting. Theoretically, the SNR of DCE-MRI can be improved by acquiring data in 3D instead of in multiple 2D slices. Another advantage of 3D imaging is that the SR pulses can be skipped provided that the entire volume of interest is excited by each readout RF pulse. However, 3D DCE-MRI requires undersampling of k-space in order to maintain decent temporal resolution. In this study we present and investigate the feasibility of a non-SR-prepared 3D imaging sequence for whole-brain DCE-MRI. Image reconstruction from undersampled k-space data was performed using k-t PCA [3] and partial Fourier imaging, yielding a spatial coverage of 20 slices per second.