gms | German Medical Science

Objective: Diffusion tensor imaging (DTI) and (semi-)automated fiber tracking algorithms are techniques for the in vivo identification of parts of fiber tracts in studies of neuronal connectivity. Furthermore, they are used more and more frequently for planning neurosurgical approaches and intraoperative neuronavigation. DTI, however, does not directly render the anatomy of fiber tracts, but reflects the Brownian movement of water molecules within tissue. Therefore, we compared the anatomical accuracy of different fiber tracts in healthy and pathological human brains, which were interactively delineated with the results given by an automated fiber tracking program.

Methods: A healthy female and a male child with a left fronto-medial glioma were investigated at a 3.0 T clinical MR scanner. We obtained a high resolution multi-shot T1-weighted 3-D FFE sequence covering the entire head for anatomical reference. Corresponding single-shot diffusion weighted spin-echo EPI sequences along each of nine different gradient axes were acquired. The fiber tracts were interactively marked in each section on the basis of anatomical landmarks (e.g., lateral ventricles, calcarine sulcus, precentral gyrus, internal capsule) and the diffusion images. Automated fiber delineation was performed by use of the MEDInria tracking program.

Results: Comparing both, the interactively delineated with the tracked fiber tracts, they show considerable differences. E.g., the corticospinal tract can be mapped with high precision interactively. The results show a high correspondence to known anatomy and histological data of that fiber tract, which is also true for the optic radiation. The tracked data reveal only a part of the fiber tracts; the algorithm does not identify the entire tracts. Especially in the corticospinal tract most of the fibers above the internal capsule, running in a sharp curve to the lateral cortical surface were not detected. The algorithm cannot distinguish between anatomically defined fiber tracts, “erroneous” fibers were frequently added to a tract but belonging to another fiber system.

Conclusions: This study shows, that the DTI images offer sufficient quality concerning spatial and grey-value resolution for precise identification of major fiber tracts. We recommend the more time consuming interactive delineation method, especially for neurosurgical purposes, because the fiber tracking algorithms are, till now, not able to detect entire fiber tracts precisely which is necessary for intraoperative fiber tract navigation.