Abstract #1535
A Preconditioned ADMM Strategy for Field-Corrected Non-Cartesian MRI Reconstruction
Joshua D. Trzasko 1 , Armando Manduca 1 , Yunhong Shu 1 , John Huston III 1 , and Matt A Bernstein 1
1
Mayo Clinic, Rochester, MN, United States
Sparse reconstruction of non-Cartesian MRI data remains
computationally challenging since multiple gridding
operations must be executed at each iteration of the
reconstruction. Recently, an efficient
alternating-direction-method-of-multiplier (ADMM)
strategy was proposed for sparse MRI reconstruction. For
non-Cartesian MRI, the data fidelity sub-problem must
also be solved iteratively. If off-resonance effects are
accounted for, standard circulant preconditioners cannot
be used to accelerate this task. In this work, we show
that an algebraic reformulation of the ADMM scheme
enables the use of simple but effective diagonal PCs for
non-Toeplitz models, and demonstrate their practical
benefit for undersampled SWIRLS 3D CE-MRA.
This abstract and the presentation materials are available to members only;
a login is required.
Join Here
