Free-breathing cardiac cine methods are needed for pediatric and chronic obstructive pulmonary disease (COPD) subjects. Multi-slice acquisitions can offer good blood-myocardium contrast, and hence preferred over 3D methods. Current approaches independently recover the slices, followed by post-processing to combine data from different slices. In this work, a deep manifold learning scheme is introduced for the joint alignment and reconstruction of multi-slice dynamic MRI. The proposed scheme jointly learns the parameters of the deep network as well as the latent vectors for each slice, which captures the motion induced dynamic variations, from the k-t space data of the specific subject.
This abstract and the presentation materials are available to members only; a login is required.