Meeting Banner
Abstract #3813

S-SPIRiT: An Iterative/Shrinkage Approach to SPIRiT for Real-Time Cardiac MRI

Samuel T. Ting1, Rizwan Ahmad1, Yu Ding1, Hui Xue2, Lee C. Potter1, Orlando P. Simonetti1

1The Ohio State University, Columbus, OH, United States; 2Siemens Corporate Research, Princeton, NJ, United States

We propose shrinkage SPIRiT (S-SPIRiT), an application of the fast iterative shrinkage-thresholding algorithm (FISTA) to SPIRiT that results in an L1-regularized implementation of SPIRiT that is more efficient than typical nonlinear conjugate gradient (NLCG) approaches and exhibits robustness to suboptimal parameter tuning and presence of noise. This approach may be especially applicable to cardiac magnetic resonance imaging, where kernel mismatch due to breathing motion can impact image quality.

Keywords

achieve achieved adaptive additional allowing amount arbitrary axial axis beck beyond blood bottle bottom breathing calculation cardiac challenging channels coefficients column combination complex compression computational computing conditions conjugate consistency consistent constraints converged core corporate cost criterion degree denoting domain dynamically efficient error exercise fast formulation frame frames free full function gradient healthy implementation improved improvement in vivo initialize intensive inverse issues iteration iterations iterative kernel linear lung matrix measured meet memory minimizes myocardium need next noise noisy nonlinear operation operator optimal optimized parallel particular performance phantom pool post potter practical problem problems produced providing random randomly real reconstructed reconstruction reconstructions reduced reduces regularization regularized required requirements respectively rest retrospectively robustness sampling scenario scenarios searches self sense shorter shrinkage significantly slice solutions solver space spatial spirit step stopping stress studies suboptimal surpasses susceptible systole taking temporal terms theory thresholding took towards tune tuning typical typically unchanged uniformly updated variance variation view visibly wavelet whereas