Accelerated 3D Time-Resolved MR Angiography using Cartesian HYPR LR Reconstruction

INTRODUCTION In many MR applications, it is desirable to achieve high spatial and temporal resolution at the same time, such as contrast-enhanced MR angiography (CE-MRA). However, fully-sampled Cartesian acquisition for large matrices is extremely time consuming and precludes the simultaneous achievement of both goals. Several Cartesian undersampling methods, such as CAPR [1], TRIPPS [2] and variable density random sampling [3], combined with view-sharing, parallel imaging, compressed sensing [3] or other reconstruction techniques [4], have been reported to address this problem. Recently, Highly constrained backPRojection (HYPR) [5] and HYPR with Local Reconstruction (HYPR LR) [6] have been shown to achieve high spatial and temporal resolution as well as high SNR using radial acquisition patterns. Here we present a general reconstruction method that fits most Cartesian undersampling schemes and its implementation is straight forward. MATERIALS AND METHODS The proposed method is a Cartesian-based reconstruction method under the HYPR LR framework. Although it is applicable to most Cartesian undersampling techniques mentioned above, here we use TRIPPS acquisition to illustrate its implementation, as shown in Fig. 1. In the TRIPPS approach, the center of the ky-kz plane is fully sampled and points along zig-zag radial lines extend into the outer region, undersampling each time frame. A composite image is generated by summing the k-space data of all or a subset of the time frames, averaging if points