Highly efficient motion-corrected simultaneous cardiac PET-MR imaging

Abstract

Abstract-MR-based motion correction of PET data has been proposed as a high-impact application for improving image quality and myocardial tissue quantification in cardiac PET-MR imaging. However, current PET-MR motion compensation schemes acquire only motion information from MR data during PET acquisition time and diagnostic MR images afterwards, significantly increasing total exam time. Here we propose an efficient acquisition and reconstruction scheme for simultaneous cardiac PET-MR imaging that acquires both diagnostic and motion information from MR, with motion information being used to correct both PET and MR data. The proposed MR acquisition scheme includes acquiring low-resolution 2D image navigators at each heartbeat by spatially encoding the startup echoes of an ECG-gated gradient echo sequence, allowing for 100% MR scan efficiency (i.e. all data is accepted for reconstruction, reducing scan time). These 2D image navigators are then used to estimate foot-head and right-left motion of the heart during the respiratory cycle in a beat-to-beat fashion, providing motion estimates to translationally correct and bin the MR data. MR images from each respiratory bin are used to estimate 3D non-rigid motion between respiratory bins. These non-rigid motion fields are then used for correcting both the MR and PET data. Results from healthy volunteers show that motion correction improves MR visualisation of the right and left anterior descending coronary arteries. A simulated PET acquisition showed improvements in contrast of the myocardium and better depiction of papillary muscles when applying motion correction. The presented framework improves image quality and contrast compared with uncorrected images for both MR and PET images.