CoSMo: Compressed Sensing Motion Correction for Coronary MRI

INTRODUCTION: In cardiac imaging, respiratory motion causes image artifacts and ghosting. Over the past two decades, several approaches have been proposed to minimize motion artifacts such as breath-holding, respiratory navigators or self-gating [1]. In high resolution imaging, respiratory navigators using a 2D selective RF excitation typically positioned on the right hemi-diaphragm, have been used to track and gate the respiratory motion. In this technique, the k-space lines acquired immediately after the navigator signal are used for image reconstruction only if the navigator signal is within a pre-defined gating window. The lines acquired outside the gating window are rejected and reacquired in the subsequent cardiac phase. The percentage of the accepted lines is reported as the navigator efficiency, which is commonly between 30% and 60% for 5mm gating windows. Furthermore, toward the end of data acquisition, the long acquisition time and patient movement may cause navigator drifting where the majority of navigator signals fall out of the acceptance window resulting in unsuccessful completion of the imaging. The rejected k-space lines depend on the patient’s respiratory pattern and change during the data acquisition. Image reconstruction without re-acquiring these k-space lines results in incoherent aliasing artifacts. In this study, we investigate the utility of compressed sensing (CS) reconstruction [2,3] in estimating the rejected k-space lines instead of re-acquiring the data, and thus reducing the total acquisition time in coronary MRI.