Cagdas Bilen1, Ricardo Otazo2, Daniel K. Sodickson2, Ivan Selesnick1, Yao Wang1
1Department of Electrical Engineering, Polytechnic Institute of NYU, Brooklyn, NY, United States; 2Bernard and Irene Schwartz Center for Biomedical Imaging, NYU School of Medicine, New York, NY, United States
Video coding techniques such as motion compensation has been proposed to exploit temporal redundancy and improve compressed sensing reconstructions of undersampled dynamic MRI data. Many of these methods require reference frames and/or fully sampled low pass k-space data which limits the acceleration factor. We propose a regularization framework with motion compensating prior that adaptively estimates the motion field during the reconstruction iterations with no need for reference frames or fully sampled k-space data.