Abstract

The annihilating filter-based low-rank Hanel matrix approach (ALOHA) is oneof the state-of-the-art compressed sensing approaches that directlyinterpolates the missing k-space data using low-rank Hankel matrix completion.Inspired by the recent mathematical discovery that links deep neural networksto Hankel matrix decomposition using data-driven framelet basis, here wepropose a fully data-driven deep learning algorithm for k-space interpolation.Our network can be also easily applied to non-Cartesian k-space trajectories bysimply adding an additional re-gridding layer. Extensive numerical experimentsshow that the proposed deep learning method significantly outperforms theexisting image-domain deep learning approaches.

 

Quick Read (beta)