Kernel low-rank regularization: an efficient approach for recovering dynamic images on a manifold

The main focus of this abstract is to introduce an efficient algorithm to recover a free breathing and ungated cardiac MR image series from highly undersampled measurements. The main contributions are (i) a kernel low-rank algorithm to estimate the manifold structure (Laplacian) from noisy navigator signals, (ii) a fast algorithm that uses the Laplacian basis functions to recover the data from highly undersampled measurements. The utility of the algorithm is demonstrated on radial acquisitions from patients with congenital heart disease; the results show that the framework is a promising alternative to self-gating methods.