Diffusion tensor imaging: Influence of segmentation on fiber tracking in the supraspinatus muscle-An inter-operator reliability analysis.

The ability of muscle to generate force depends on its architecture and health condition. MR-based diffusion tensor imaging of muscle (mDTI) is an innovative approach for showing the fiber arrangement for the whole muscle volume. For accurate calculations of fiber metrics, muscle segmentation prior to tractography is regarded as necessary. Since segmentation is known to be operator dependent, it is important to understand how segmentation affects tractography. The aim of this study was to compare the results of deterministic fiber tracking based on muscle models generated by two independent operators. In addition, this study compares the results with a segmentation-free approach. Fifteen subjects underwent mDTI of the right shoulder. The results showed that mDTI can be successfully applied to complex joints such as the human shoulder. Furthermore, operator segmentation did not influence the results of fiber tracking and fascicle length (FL), fiber volume (FV), fractional anisotropy (FA), axial diffusivity (AD), radial diffusivity (RD), and mean diffusivity (MD) showed excellent intraclass correlation estimates (≥ 0.975). As an exploratory approach, the segmentation-free fiber tracking showed significant differences in terms of mean fascicle length. Based on these findings, we conclude that tractography is not sensitive to small deviations in muscle segmentation. Furthermore, it implies that mDTI and automatic segmentation approaches or even a segmentation-free analysis can be considered for evaluation of muscle architecture.