Purpose: To determine the accuracy of several magnetic resonance imaging (MRI)-based strategies for assessment of quadriceps muscle volume (MV) and changes in MV with training.
Methods: Images were acquired along the length of both thighs from young (26 +/- 3 yr, N= 23) and older (69 +/- 3 yr, N= 24) men and women before and after strength training. The quadriceps cross-sectional area (QCSA) of each section was measured before and after training. MV was directly assessed using all of the sections (each 9-mm thick with a 1-mm gap). Alternative estimates of MV were calculated using increasingly greater intervals between sections: every 1.1 cm (MV2), 3.1 cm (MV4), 5.1 cm (MV6), 7.1 cm (MV8), 9.1 cm (MV10), and a single QCSA (L1). The 95% limits of agreement (LOA, +/- 2 SD) between each alternative measure and the criterion measure MV were determined with Bland and Altman plots. Regression was used to predict MV from L1 and to obtain the standard error of the estimate (SEE).
Results: Before training, the 95% LOA with MV for the alternative measures ranged from 0.7% to 6.36% of MV, and the prediction of MV from L1 yielded a SEE x 2 of 14.1% of MV. For change in the alternative measures, the 95% LOA ranged from 10.3% to 26.3% of the total change in MV, and the prediction of deltaMV from deltaL1 yielded a SEE x 2 of 60% of the change in MV.
Conclusion: Increasingly greater intervals between axial MRI sections result in substantially reduced agreement with a criterion measure of MV. The use of one axial section results in relatively higher error and thus should be used only when large effect sizes are expected.