Multiresolution field map estimation using golden section search for water-fat separation

Many diagnostic MRI sequences demand reliable and uniform fat suppression. Multipoint water-fat separation methods, which are based on chemical-shift induced phase differences, have shown great success in the presence of field inhomogeneities. This work presents a computationally efficient and robust field map estimation method. The method begins with subsampling image data into a multiresolution image pyramidal structure, and then utilizes a golden section search to directly locate possible field map values at the coarsest level of the pyramidal structure. The field map estimate is refined and propagated to increasingly finer resolutions in an efficient manner until the full-resolution field map is obtained for final water-fat separation. The proposed method is validated with multiecho sequences where long echo-spacings normally impose great challenges on reliable field map estimation.