gms | German Medical Science

Objective: Diffusion tensor imaging (DTI) based fiber tractography (FT) is routinely used in neurosurgical tumor surgery to visualize major white matter tracts such as the corticospinal tract (CST). However, a major limitation of clinically available DTI based FT is the estimation of a single major diffusion direction, not capable of resolving crossing or kissing fibers. Various complex models eventually overcoming such limitations of the DTI based approach exist but haven't found their way into clinical practice. Especially in areas where no single major diffusion direction can be seen a dual-tensor model might be beneficial to overcome limitations of the classical DTI model. However, recently integrated in the clinical practice its capability to improve FT is investigated.

Methods: MRI data of ten healthy subjects were retrospectively analyzed. Data were processed using Brainlab Fiber Tracking Elements version 1.0 for the one-tensor and version 2.0 for the dual-tensor approach (Brainlab, Munich, Germany). FT of the CST was performed applying a region of interest (ROI) based approach, using the same include ROIs (motor cortex, cerebral peduncle) and exclude ROIs (for tract refinement) for both models. Tract volumes as well as the Dice coefficient (DC) as measure of spatial overlap between both approaches were calculated and analyzed. Statistical analysis was performed using a paired t-test.

Results: Mean volumes of the one-tensor compared with the dual-tensor FT for the left CST were 8.15cm^³ ± 1.37cm^³ [5.9 cm^³; 10.1 cm^³] and 6.60cm^³ ± 1.23cm^³ [4.61cm^³; 8.48cm^³], respectively (p<0.01). The mean values of right CST with respect to volumes were 7.80cm^³ ± 1.64cm^³ [5.23cm^³; 10.60cm^³] for the one-tensor FT and 6.33cm^³ ± 1.49cm^³ [4.04cm^³; 9.06cm^³] for dual-tensor FT (p<0.01). The mean DC was 88.11% ± 1.68% on the left [85.89%; 90.97%] versus 88.00% ± 1.89% on the right side [84.28%; 90.70%].

Conclusion: Statistically significant differences were found between both models regarding volume of the reconstructed tracts, with even smaller volume in case of the dual tensor FT approach, whereas the Dice coefficient revealed a good spatial overlap. Even though the dual tensor model offers two possible directions to follow for tractography, only one (with smaller angular difference) is used, which might lead to less widened tracts, and therefore more refined and possibly even more plausible reconstructions, whereas aspects such as its capability of resolving complex structures need to be further investigated.