1 Department of Health Kinesiology and Applied Physiology, Concordia University, 7141 Sherbrooke Street W, SP-165.29, Montreal, QC, H4B 1R6, Canada.
2 Department of Orthopaedic Surgery, University of California, San Francisco, CA, USA.
3 Department of Health Kinesiology and Applied Physiology, Concordia University, 7141 Sherbrooke Street W, SP-165.29, Montreal, QC, H4B 1R6, Canada. maryse.fortin@concordia.ca.
4 PERFORM Centre, Concordia University, Montreal, QC, Canada. maryse.fortin@concordia.ca.
5 Centre de Recherche Interdisciplinaire en Réadaptation (CRIR), Montreal, QC, Canada. maryse.fortin@concordia.ca.

Purpose: The purpose of this study was to evaluate the agreement between paraspinal muscle composition measurements obtained from fat-water images using % fat-signal fraction (%FSF) in comparison to those obtained from T2-weighted magnetic resonance images (MRI) using a thresholding method.

Methods: A sample of 35 subjects (19 females, 16 males; 40.26 ± 11.3 years old) was selected from a cohort of patients with chronic low back pain (LBP). Axial T2-weighted and IDEAL (Lava-Flex, 2 echo sequence) fat and water MR images were obtained using a 3.0 Tesla GE scanner. Multifidus, erector spinae, and psoas major muscle composition measurements were acquired bilaterally at L4-L5 and L5-S1 using both imaging sequences and related measurement methods. All measurements were obtained by the same rater, with a minimum of 7 days between each method. Intra-class correlation coefficients (ICCs) were calculated to assess intra-rater reliability. Pearson Correlation and Bland-Altman 95% limits of agreement were used to assess the agreement between both measurement methods.

Results: The intra-rater reliability was excellent for all measurements with ICCs varying between 0.851 and 0.997. Strong positive correlations indicating a strong relationship between composition measurements were obtained from fat-water and T2-weighted images for bilateral multifidus and erector spinae muscles at both spinal levels and the right psoas major muscle at L4-L5, with correlation coefficient r ranging between 0.67 and 0.92. Bland-Altman plots for bilateral multifidus and erector spinae muscles at both levels revealed excellent agreement between the two methods, however, systematic differences between both methods were evident for psoas major fat measurements.

Conclusion: Our findings suggest that utilizing fat-water and T2-weighted MR images are comparable for quantifying multifidus and erector spinae muscle composition but not of the psoas major. While this suggests that both methods could be used interchangeably for the multifidus and erector spinae, further evaluation is required to expand and confirm our findings to other spinal levels.