Motion-Compensated Image Reconstruction for Magnetic Resonance (MR) Imaging and for Simultaneous Positron Emission Tomography/MR Imaging

In this work, novel algorithms for 4D (3D + respiratory) and 5D (3D + respiratory + cardiac) motion-compensated (MoCo) magnetic resonance (MR) and positron emission tomography (PET) image reconstruction were developed. The focus of all methods was set on short MR acquisition times. Therefore, respiratory and cardiac patient motion were estimated on the basis of strongly undersampled radial MR data employing joint motion estimation and MR image reconstruction. In case of simultaneous PET/MR acquisitions, motion information derived from MR was incorporated into the MoCo PET reconstruction. 4D respiratory MoCo MR image reconstructions with acquisition times of 40 s achieved an image quality comparable to standard motion handling approaches, which require one order of magnitude longer MR acquisition times. Respiratory MoCo PET images using 1 min of the MR acquisition time for motion estimation revealed improved PET image quality and quantification accuracy when compared to standard reconstruction methods. Additional compensation of cardiac motion resulted in increased image sharpness of MR and PET images in the heart region and enabled time-resolved 5D imaging allowing for reconstruction of any arbitrary combination of respiratory and cardiac motion phases. The proposed methods for MoCo image reconstruction may be integrated into clinical routine, reducing MR acquisition times for improved patient comfort and increasing the diagnostic value of MR and simultaneous PET/MR examinations of the thorax and abdomen