Hojin Kim1,2, Kyung Sung1, Brian Andrew Hargreaves1
1Department of Radiology, Stanford University, Stanford, CA, United States; 2Electrical Engineering, Stanford University, Stanford, CA, United States
In the least-squares fat/water separation techniques, the residual or cost runction that is minimized contains exactly one or two local minimum, depending on the relative amount of fat and water, and water-fat phase difference. Separation algorithms attempt to find which minimum provides true field-map, but may converge to the incorrect local minimum. Based on this principle, this work proposes a robust field-map estimation technique by tracking two minima at each pixel through region growing process and suggesting more secure way of determining an initial seed for region growing.