Quantification of muscle fat in patients with low back pain: Comparison of multi-echo MR imaging with single-voxel MR spectroscopy

Purpose: To compare lumbar muscle fat-signal fractions derived from three-dimensional dual gradient-echo magnetic resonance (MR) imaging and multiple gradient-echo MR imaging with fractions from single-voxel MR spectroscopy in patients with low back pain. Materials and This prospective study had institutional review board ap-Methods: proval, and written informed consent was obtained from all study participants. Fifty-six patients (32 women; mean age, 52 years 6 15 [standard deviation]; age range, 20-79 years) with low back pain underwent standard 1.5-T MR imaging, which was supplemented by dual-echo MR imaging, multi-echo MR imaging, and MR spectroscopy to quantify fatty degeneration of bilateral lumbar multifidus muscles in a region of interest at the intervertebral level of L4 through L5. Fat-signal fractions were determined from signal intensities on fat-and water-only images from both imaging data sets (dual-echo and multi-echo fat-signal fractions without T2 ^* correction) or directly obtained, with additional T2 ^* correction, from multi-echo MR imaging. The results were compared with MR spectroscopic fractions. The Student t test and Bland-Altman plots were used to quantify agreement between fat-signal fractions derived from imaging and from spectroscopy. Results: In total, 102 spectroscopic measurements were obtained bilaterally (46 of 56) or unilaterally (10 of 56). Mean spectroscopic fat-signal fraction was 19.6 6 11.4 (range, 5.4-63.5). Correlation between spectroscopic and all imaging-based fat-signal fractions was statistically significant (R² = 0.87-0.92; all P <.001). Mean dual-echo fat-signal fractions not corrected for T2^* and multi-echo fat-signal fractions corrected for T2^*significantly differed from spectroscopic fractions (both P <.01), but mean multi-echo fractions not corrected for T2^* did not (P =.11). There was a small measurement bias of 0.5% (95% limits of agreement:26.0%, 7.2%) compared with spectroscopic fractions. Conclusion: Large-volume image-based (dual-echo and multi-echo MR imaging) and spectroscopic fat-signal fractions agree well, thus allowing fast and accurate quantification of muscle fat content in patients with low back pain.