Optimal Channel Selection for Respiratory Self-Gating Signals

INTRODUCTION: Motion due to respiration poses a significant challenge for MR imaging of the abdomen and thorax. However, it is possible to compensate for the motion, e.g. by accepting data from only a certain respiration phase, when the temporal phases during the acquisition are known. Traditionally, the respiration phases are obtained using respiration bellows, but the clinical acceptance of such devices is very low. Self-gating techniques determine this information directly from the measurement data, in contrast to navigator techniques that perform additional intermediate navigation scans. Because the navigation scans interrupt the steady state of the magnetization and because navigators prolong the overall scan time, self-gating is favorable for T1-weighted gradient-echo (GRE) sequences. Radial k-space trajectories are particularly well suited because they continuously sample the center of k-space, which allows for detection of respiratory motion. Several algorithms have been proposed for the signal extraction where in most methods a suitable coil element has to be selected manually to get reasonable results. This is time-consuming for modern MRI systems with high number of receive channels, and infeasible for clinical routine applications. Here, we propose a fully automated procedure to generate proper self-gating signals for 3D stack-of-stars GRE sequences with golden-angle sampling, which are comparable to the signal from a respiratory bellows.