Measuring Thigh Cross-sectional Areas From CT Scans

Abstract

PURPOSE: Muscle and fat areas in thigh CT scans are important outcomes in aging and exercise research. NIH Image J and SliceOmatic software programs are often used to quantify these areas but three methodological issues commonly exist between references: the tissue density range used to define muscle and fat varies between studies, areas of intermediate density are omitted, and the handling of areas of matching density (i.e. skin and bone marrow) is unreported. Thus, the purpose of this analysis was to validate methods for using these programs while assessing the effects of these three issues on the results. METHODS: CT scans of the mid-thigh were analyzed for two cohorts based on gender (57F, 65-86 yrs v. 44M, 64-91 yrs) and the effects of resistance exercise training (12 wks, 26M, 62-77 yrs) and detraining (26 wks, subset 16M, 62-75 yrs). The total thigh was segmented into seven regions based on Hounsfield Units range: fat of normal (NDF) and high-density (HDF); muscle of low (LDM), normal (NDM), high (HDM), and very high density (VHDM); and bone. RESULTS (mean, SD, cm2): In the first cohort, male thigh total area was smaller (207 ± 27 v 223 ± 42, P=0.02) and contained less NDF (54 ± 19 v 118 ± 38, P<0.0001) but more muscle of all densities (e.g. NDM 108 ± 19 v 69 ± 14, P<0.0001) than female thigh. These results were from Image J and the effect of skin was not considered. However, the results were strongly correlated (R2=0.99) with those from SliceOmatic even if skin was included as subcutaneous fat. In the second cohort, resistance training (pre v post) increased thigh size (220 ± 32 v 229 ± 30, P<0.0001) and muscle content of all densities (e.g. NDM 107 ± 20 v 111 ± 22, P=0.02) except VHDM. Detraining (trained v detrained) decreased thigh size (230 ± 34 v 222 ± 39, P=0.0007) and content of NDM (115 ± 24 v 106 ± 21, P<000.1) and HDM (2.5 ± 1.8 v 1.8 ± 1.6). These results were from SliceOmatic and were not affected by skin fat and bone marrow. CONCLUSIONS: Image J and SliceOmatic generate similar results for measurement of muscle and fat of all densities in the thigh. SliceOmatic is the more capable program but Image J is sufficient if areas with density similar to muscle and fat are constant and do not affect the results. Funding: VA RRD Merit RX001203 (RAD) and NIH NIA R01 AG046920 (CAP and MMB).