3D Gd-Enhanced Carotid MRA with High Spatial and Temporal Resolution: An Application of a New Class of K-space Sampling Theorems

Introduction: In principle, gadolinium-enhanced carotid MRA has well-documented advantages over methods based on time-offlight contrast. Unfortunately, conventional Fourier imaging methods fail to achieve adequate spatial resolution because data acquisition is limited to the brief first pass of contrast material. However, the conventional Fourier sampling theorem does not exploit the strong prior knowledge that is available; the high-contrast edges in carotid MRA images are known to be confined to small, easily predetermined portions of the total field of view (FOV). Multiple Region MRI (mrMRI [1,2]) utilizes this type of prior knowledge to reconstruct images from relatively sparse k-space sampling patterns that can be acquired in significantly less time than the conventional ones. Importantly, if the sampling pattern is chosen wisely, mrMR image reconstruction does not amplify noise, just as in conventional Fourier imaging. This paper describes how mrMRI can be used to perform 3D contrast-enhanced carotid MRA (mrMRA) with high spatial resolution. Temporal resolution was high enough to avoid bolus timing and to facilitate the use of temporal filtering techniques that improve SNR and arterial contrast by using the entire series of 3D “snapshot” scans.