Robustness of fat quantification using chemical shift imaging

The purpose of this study was to investigate the effect of parameter changes that can potentially lead to unreliable measurements in fat quantification. Chemical shift imaging was performed using spoiled gradient echo sequences with systematic variations in the following: two-dimensional/three-dimensional sequence, number of echoes, delta echo time, fractional echo factor, slice thickness, repetition time, flip angle, bandwidth, matrix size, flow compensation and field strength. Results indicated no significant (or significant but small) changes in fat fraction with parameter. The significant changes can be attributed to the known effects of T1 bias and two forms of noise bias.