Abstract: This study evaluated the accuracy of fatty acid composition (FAC) quantification at 5.0 T ultra-high-field MRI by systematically assessing the effects of imaging dimensionality, readout polarity, and echo-time combination, and establishing a protocol for human adipose tissue. Multi-echo gradient-echo imaging was performed in oil phantoms and in vivo. The number of double bonds (ndb) and methylene-interrupted double bonds (nmidb) were stably estimated using a multi-peak fat prior-constrained model, with magnetic resonance spectroscopy as reference. Three-dimensional imaging yielded estimates closer to NMR values, while monopolar and bipolar readouts showed no significant differences. An echo-time combination of TE1<1.9 ms and ΔTE=0.6~1.0 ms significantly improved fitting accuracy. In human abdominal subcutaneous adipose tissue, ndb and nmidb were 2.817±0.067 and 0.858±0.087, respectively, demonstrating good consistency and stability. These results provide a technical basis for accurate FAC quantification at 5.0 T MRI and its application in metabolic disease research.

Key words: chemical shift-encoded MRI, fatty acid composition, multi-echo gradient echo, 5.0?T magnetic resonance, imaging parameter selection