Abstract #2401

Schmidt M, Fraser K

A two-point region-growing phase correction algorithm was recently proposed for fat-water separation, based on detecting large phase shifts at the boundary between water-dominant and fat-dominant regions. This work investigates the performance of the phase correction method in the presence of noise and magnetic field inhomogeneity, and proposes two fundamental modifications to the algorithm: (i) to start the region-growing process with images of lower resolution, and (ii) to apply low-pass filters to the final phase-corrected map prior to calculating separate water and fat images. The resulting method is accurate, faster and less sensitive to noise.