Continuous table movement MRI in a single breath-hold: Highly undersampled radial acquisitions with nonlinear iterative reconstruction and joint coil estimation

INTRODUCTION Continuous table movement MRI (TimCT ) is an emerging technique for a variety of clinical applications [1, 2]. In particular, rapid short-TR two-dimensional imaging of the body in transverse orientation features optimum scan time efficiency with seamless coverage beyond the physical coronal FOV determined by the magnet and the gradient hardware [3]. Furthermore, state-of-the-art implementations support data acquisition with RF phased-array surface coils, while dynamically selecting a subset of receive coils close to the isocenter of the magnet. Although these conditions establish the basis for accelerating data acquisition with parallel imaging, the achievable acceleration factors in a clinical setting are not yet sufficient to scan an extended FOV in a single breath-hold. Radial scanning in combination with innovative iterative image reconstruction, however, promises significantly higher acceleration factors. Moreover, such methods allow for a precise calculation of coil sensitivity profiles which for continuously moving table MRI represents a unique challenge: apart from moving to a new slice position for each section, also the subset of receive coil elements might have been changed. In this work, such problems are overcome by a nonlinear iterative image reconstruction technique with joint estimation of coil sensitivities [4, 5]. A preliminary application deals with continuously moving table MRI based on highly undersampled radial acquisitions. The scan efficiency and image quality of this concept was demonstrated in 5 healthy volunteers for MRI of a large abdominal FOV in a single breath-hold.