Reproducibility of diffusion tensor imaging in human forearm muscles at 3.0 T in a clinical setting

The aim of the present study was to evaluate a fast clinical protocol to enable diffusion tensor imaging of the human forearm and assess the reproducibility of six diffusion tensor imaging parameters, i.e., the tensor eigenvalues (lambda(1), lambda(2), and lambda(3)), mean diffusivity, fractional anisotropy, and ellipsoid eccentricity. The right forearms of 10 healthy volunteers were scanned twice, with a 1-week interval. Reproducibility of the diffusion tensor imaging parameters was interpreted using Bland-Altman plots, coefficient of repeatability, repeatability index, and the intraclass correlation coefficient. Analysis was done for three regions of interest: the whole muscle volume, flexor digitorum profundus, and extensor digitorum. The Bland-Altman analysis showed that there is good agreement between the two measurements. Based on the intraclass correlation coefficients, agreement was substantial (0.59 < intraclass correlation coefficient < 0.92) for all six parameters of the whole muscle volume and flexor digitorum profundus but only fair (0.18 < intraclass correlation coefficient < 0.64) for the extensor digitorum. Using a 7 min 40 sec scan protocol, which was well tolerated by the volunteers, the reproducibility of diffusion tensor imaging parameters was demonstrated. However, repeatability varies, depending on the region of interest and diffusion tensor imaging parameters. This should be taken into account when a longitudinal study is designed.