Meeting Banner
Abstract #4249

Dynamic Cardiac MRI Reconstruction with Weighted Redundant Haar Wavelets

Jun Liu1, Jeremy Rapin1, Ti-chiun Chang1, Alban Lefebvre1, Michael Zenge2, Edgar Mueller2, Mariappan S. Nadar1

1Siemens Corporate Research, Princeton, NJ, United States; 2Siemens AG, Healthcare Sector, Erlangen, Germany

High temporal resolution is often desired in Cardiac Magnetic Resonance Imaging (CMRI). Parallel imaging enables the reconstruction with a reduced the number of acquired frequencies, hence accelerating the acquisition, and spiral sampling scheme is widely used in CMRI. In this work, a new approach was proposed for dynamic cardiac MRI reconstruction by applying an L1 regularization based on the weighted 3D redundant Haar wavele for incorporating the smoothness in both spatial and temporal directions. Experiments conducted on a CMRI data with only 15 spokes for each temporal phase demonstrated the effectiveness of the proposed approach.

Keywords

able accelerating account achieved achieves acquisition acquisitions addition although applied applying artifact artifacts assessment assoc band beck cardiac cardiovascular channels clinical coefficients coil component concatenation conducted consecutive constant contrary corporate correspond demands denoted denotes desired details dimension dynamic effective effectively effectiveness efficiently employed enables estimation excellent exhibit favorable fidelity field fine frequencies frequency function good healthy help hence implemented imposed include incorporating independent intrinsic invasive little many matrices meanwhile noise obvious operator optimization optimizing parallel penalization penalized penalty piecewise pixel plot plots prior problem process product promising properties proposed proven pulse radial reasonably reconstruction reconstructions reduced redundant regularization regularizing repetition represent resolution resolved respect respectively sake samples sampling scanner scheme sector sense several severely similarity since smooth smoothness solve solved space spatial spokes statistical streaking strongly structure summary suppress system take target technology temporal tensor term third treated type verify vertical view visually volunteer wavelet wavelets widely width wise written yield