Feasibility of omega-3 fatty acid fraction mapping using chemical shift encoding-based imaging at 3 T

The aim of this work is to develop an ω-3 fatty acid fraction mapping method at 3 T based on a chemical shift encoding model, to assess its performance in a phantom and in vitro study, and to further demonstrate its feasibility in vivo.

A signal model was heuristically derived based on spectral appearance and theoretical considerations of the corresponding molecular structures to differentiate between ω-3 and non-ω-3 fatty acid substituents in triacylglycerols in addition to the number of double bonds (ndb), the number of methylene-interrupted double bonds (nmidb), and the mean fatty acid chain length (CL). First, the signal model was validated using single-voxel spectroscopy and a time-interleaved multi-echo gradient-echo (TIMGRE) sequence in gas chromatography-mass spectrometry (GC-MS)-calibrated oil phantoms. Second, the TIMGRE-based method was validated in vitro in 21 adipose tissue samples with corresponding GC-MS measurements. Third, an in vivo feasibility study was performed for the TIMGRE-based method in the gluteal region of two healthy volunteers. Phantom and in vitro data was analyzed using a Bland-Altman analysis.

Compared with GC-MS, MRS showed in the phantom study significant correlations in estimating the ω-3 fraction (p < 0.001), ndb (p < 0.001), nmidb (p < 0.001), and CL (p = 0.001); MRI showed in the phantom study significant correlations (all p < 0.001) for the ω-3 fraction, ndb, and nmidb, but no correlation for CL. Also in the in vitro study, significant correlations (all p < 0.001) between MRI and GC-MS were observed for the ω-3 fraction, ndb, and nmidb, but not for CL. An exemplary ROI measurement in vivo in the gluteal subcutaneous adipose tissue yielded (mean ± standard deviation) 0.8% ± 1.9% ω-3 fraction.

The present study demonstrated strong correlations between gradient-echo imaging-based ω-3 fatty acid fraction mapping and GC-MS in the phantom and in vitro study. Furthermore, feasibility was demonstrated for characterizing adipose tissue in vivo.