Effect of Fat Spectral Model Parameters on Hepatic Fat Quantification by Multi-Echo Gradient-Echo Magnetic Resonance Imaging

Takeshi Yokoo¹, Mark Bydder¹, Gavin Hamilton¹, Michael S. Middleton¹, Heather M. Patton², Terek Hassanein², Jeffrey B. Schwimmer³, Claude B. Sirlin¹

Recently several groups suggested that accurate estimation of hepatic fat content using gradient-recalled-echo (GRE) MR imaging requires modeling of the signal interference caused by the non-dominant fat peaks as well as the dominant 1.2 ppm peak. Previously our group demonstrated high fat quantification accuracy in human subjects using non-T1-weighted multi-echo GRE imaging with a 3-peak fat spectral model. However, it is conceivable that the fat estimation accuracy depends on the choice of the model parameters. In this study, we perturbed fat spectral model and assessed its effect of fat estimation accuracy, using MR spectroscopy as the reference technique.