Methodological considerations in region of interest definitions for paraspinal muscles in axial MRIs of the lumbar spine.

BACKGROUND: Magnetic Resonance Imaging (MRI) is commonly used to assess the health of the lumbar spine and supporting structures. Studies have suggested that fatty infiltration of the posterior lumbar muscles is important in predicting responses to treatment for low back pain. However, methodological differences exist in defining the region of interest (ROI) of a muscle, which limits the ability to compare data between studies. The purpose of this study was to determine reliability and systematic differences within and between two commonly utilized methodologies for ROI definitions of lumbar paraspinal muscle. METHODS: T2-weighted MRIs of the mid-L4 vertebrae from 37 patients with low back pain who were scheduled for lumbar spine surgery were included from a hospital database. Fatty infiltration for these patients ranged from low to high, based on Kjaer criteria. Two methods were used to define ROI: 1) segmentation of the multifidus and erector spinae based on fascial planes including epimuscular fat, and 2) segmentation of the multifidus and erector spinae based on visible muscle boundaries, which did not include epimuscular fat. Total cross sectional area (tCSA), fat signal fraction (FSF), muscle cross sectional area, and fat cross sectional area were measured. Degree of agreement between raters for each parameter was assessed using intra-class correlation coefficients (ICC) and area fraction of overlapping voxels. RESULTS: Excellent inter-rater agreement (ICC > 0.75) was observed for all measures for both methods. There was no significant difference between area fraction overlap of ROIs between methods. Method 1 demonstrated a greater tCSA for both the erector spinae (14-15%, p < 0.001) and multifidus (4%, p < 0.016) but a greater FSF only for the erector spinae (11-13%, p < 0.001). CONCLUSION: The two methods of defining lumbar spine muscle ROIs demonstrated excellent inter-rater reliability, although significant differences exist as method 1 showed larger CSA and FSF values compared to method 2. The results of this study confirm the validity of using either method to measure lumbar paraspinal musculature, and that method should be selected based on the primary outcome variables of interest.

Contribution of Lumbar Spine Pathology and Age to Paraspinal Muscle Size and Fatty Infiltration.

STUDY DESIGN: Retrospective chart analysis of 199 individuals aged 18 to 80 years scheduled for lumbar spine surgery. OBJECTIVE: The purpose of this study was to quantify changes in muscle cross-sectional area (CSA) and fat signal fraction (FSF) with age in men and women with lumbar spine pathology and compare them to published normative data. SUMMARY OF BACKGROUND DATA: Pathological changes in lumbar paraspinal muscle are often confounded by age-related decline in muscle size (CSA) and quality (fatty infiltration). Individuals with pathology have been shown to have decreased CSA and fatty infiltration of both the multifidus and erector spinae muscles, but the magnitude of these changes in the context of normal aging is unknown. METHODS: Individuals aged 18 to 80 years who were scheduled for lumbar surgery for diagnoses associated with lumbar spine pain or pathology were included. Muscle CSA and FSF of the multifidus and erector spinae were measured from preoperative T2-weighted magnetic resonance images at the L4 level. Univariate and multiple linear regression analyses were performed for each outcome using age and sex as predictor variables. Statistical comparisons of univariate regression parameters (slope and intercept) to published normative data were also performed. RESULTS: There was no change in CSA with age in either sex (P > 0.05), but women had lower CSAs than men in both muscles (P < 0.0001). There was an increase in FSF with age in erector spinae and multifidus muscles in both sexes (P < 0.0001). Multifidus FSF values were higher in women with lumbar spine pathology than published values for healthy controls (P = 0.03), and slopes tended to be steeper with pathology for both muscles in women (P < 0.08) but not in men (P > 0.31). CONCLUSION: Lumbar muscle fat content, but not CSA, changes with age in individuals with pathology. In women, this increase is more profound than age-related increases in healthy individuals. LEVEL OF EVIDENCE: 3.