Fat-water separation in low SNR MRI using Graph Cut and mixed magnitude-complex re-fitting with Tukey's Bi-weighting

Motivation: Optimal fat-water separation in single-shot Multi-Echo MR (ME-MR) on a 3Tesla scanner with large echo-spacing is needed for accurate fat quantification with a reduced Specific Absorption Rate. However, the penalty of increased noise makes this challenging. Goal(s): This research proposes a novel method for optimal fat-water separation with low SNR single-shot 3T ME-MR with large echo-spacing. Approach: The proposed method combines the Graph-Cut and Mixed Magnitude-Complex fitting methods, introducing Tukey's bi-weighting during optimization for coping with noise. Results: Liver Proton-Density-Fat-Fraction% obtained from the proposed method correlated better with the gold standard than the widely used Graph-Cut method in these low SNR ME-MR. Impact: The Proton-Density-Fat-Fraction maps obtained from fat-water separation with ME-MR could be of diagnostic and prognostic value in many diseases. This study proposes a novel method for handling fat-water separation in noise-prone single-shot ME-MR with large echo-spacing.