Yue Hu1, Sajan Goud Lingala2,
1Electrical & Computer
Engineering, University of Rochester, Rochester, NY, United States; 2Biomedical
Engineering, University of Rochester, Rochester, NY, United States
We consider the problem of free-breathing high-resolution structural cardiac MRI. To overcome the limitations with conventional navigator pulses, we propose to reformulate the structural problem as a dynamic one by recovering a 2D+time dataset from under-sampled k-t data. We use our recently proposed k-t SLR scheme to estimate the principal temporal bases of the data, which enables data sharing between heartbeats and facilitate high-resolution dynamic dataset recovery. The structural image at any respiratory phase can be obtained from the recovered dynamic data. Experiments on free-breathing cardiac MRI data showed the feasibility of the proposed scheme in obtaining high fidelity reconstructions.