To compare different lipid multipeak spectral models to the single-peak model in Dixon-based fat-water separation and to evaluate differences between visually scored magnetic resonance (MR) images and quantitatively assessed fat fractions in muscle of Duchenne muscular dystrophy patients.Materials and Methods:
T1-weighted and 3-point Dixon imaging of the upper and lower leg was performed in 13 Duchenne patients and six healthy controls. Three-, four-, and five-peak lipid spectrum models were compared to a single-peak model and to each other. T1-weighted images were visually scored by two radiologists and quantitative fat fractions were obtained from Dixon images.Results:
Differences between the multipeak spectral models were minimal. The three-peak model was used for subsequent comparisons. Although there was high correlation between quantitative and visual scores, visual scores were consistently higher than quantitative values of the same muscles.Conclusion:
There are minor differences between the various lipid spectral models in terms of quantifying fat fraction in a large number of skeletal muscles in the legs of Duchenne patients and healthy controls. Quantitative 3-point Dixon MRI is more precise and reliable than visual radiological methods for evaluation of fat fractions for potential longitudinal follow-up or therapy evaluation of Duchenne patients. J. Magn. Reson. Imaging 2013;38:619–624. © 2013 Wiley Periodicals, Inc.