Quantification of thigh muscle volume in children and adolescents using magnetic resonance imaging

Abstract Estimating muscle volume (MV) using variable numbers of cross-sectional area (CSA) slices obtained from magnetic resonance imaging (MRI) introduces an error that is known in adults, but not in children and adolescents, whereby body sizes differ due to growth and maturation. Therefore, 15 children and adolescents (11 males, 14.8 ± 2.1 years) underwent MRI scans of the right thigh using a 1.5 T scanner to establish this error. A criterion MV was determined by tracing around and summing all CSAs, with MV subsequently estimated using every second, third, fourth and fifth CSA slice. Bland-Altman plots identified mean bias and limits of agreement (LoA) between methods. Error rates between 1.0 and 10.4% were seen between criterion and estimated MV. Additional analyses identified an impact of formulae selection, with a cylindrical formula preferred to a truncated cone. To counter high error between criterion and estimated MV due to the discrepancies in the number of CSA slices analysed, length-matched criterion volumes were established, with reduced error rates (0.5-2.0%) being produced as a result. CSA at 50% thigh-length also predicted MV, producing a high error (13.8-39.6%). Pearson's correlation coefficients determined relationships between error and measures of body size/composition, with all body size/composition measures being correlated (r = -0.78-0.86, p < 0.05) with the error between criterion and estimated MV. To conclude, MV can be accurately estimated using fewer CSA slices. However, the associated error must be considered when calculating MV in children and adolescents, as body size biases estimates.