Precise knowledge of the mesencephalic brain regions made possible neurosurgical procedures such as deep brain stimulation (DBS) of the subthalamic nucleus to treat motor symptoms in Parkinson's disease. Since then, area of interest for DBS is growing and point out new brain areas located in the brainstem¹. However little is known about connections and spatial orientation of brainstem pathways due to a lack of precise atlases based on myelin-stained sections.

Imaging studies using diffusion MRI have shown the potential of this technique to identify structures and tracts in the basal ganglia² and the brainstem³. Brainstem fiber tracts were indeed highlighted with ex vivo MRI acquisition at 11.7T (scanning time: 62 hours) providing a spatial resolution of 0.255 mm³.

Achieving such a resolution in vivo is not realistic, even with ultra-high field MR scanners. Nevertheless, a technique known as super-resolution track-density imaging (TDI) was developed⁴ and validated⁵ as a tool to gain spatial resolution using post-processing methods.

We used diffusion MRI acquired at 3T (1.25 mm isovoxel size) from one healthy individual included and preprocessed in the Human Connectome Project⁶. Probabilistic tractography was run in subject-specific native space using MrTrix software package and its constrained spherical deconvolution technique by randomly seeding throughout the brainstem (segmented with the FreeSurfer recon-all pipeline). The relevant parameters were: number of tracks = 10 000 000, maximum angle between steps = 45°, 0.1 mm step-size, any track with length < 6.25 mm was discarded, l_max = 10, termination criteria: exit the brain or when the FA amplitude was < 0.1.

We applied the SIFT method⁷ to the reconstructed fiber tracks in order to improve the biological accuracy of the reconstruction (from 10 000 000 to 4 000 000 streamlines).

Directionally-encoded colour (DEC) super-resolution TDI maps were generated by calculating the number of tracks in each element of a grid. A salient point of the TDI mapping is that the grid element is made smaller than the acquired voxel size, generating a final map at a much higher resolution than the original DWI data. In our study, we used a 1.25 mm isotropic source diffusion data to calculate a 0.2 mm isotropic TDI map with the display of local fiber directionality.

To evaluate the improvement provided by TDI in comparison to fractional anisotropy (FA), we draw two inclusions ROIs on the same FA and TDI slices and then reconstructed a branch of the pontocerebellar fibers in each case.

Fig. 1 and Fig. 2 show the DEC super-resolution TDI maps and corresponding conventional FA maps. A clear amelioration of the spatial resolution is achieved by super-resolution, as well as the emphasis on different contrast informations. Major pathways were identified by their inferior-superior orientations (blue in Fig. 1) including the corticospinal tract, medial lemniscus and medial longitudinal fasciculus. TDI maps from Fig. 1 provided striking contrasts to delineate the fiber bundles of the corticospinal tract (blue) from the pontocerebellar fibers (red). Fig. 2 further illustrates the anatomical complexity of the fiber tracts shed light on TDI maps at different sagittal plans. The superior cerebellar peduncle was distinctly delineated with its anterio-posterior orientation (green).

Fig. 3 shows that the reconstructed branch of the pontocerebellar fibers by segmenting ROIs on TDI instead of FA allows more precise and detailed results. In particular, the reconstructed tract from the FA ROIs clearly deviates to a more rostral branch of the pontocerebellar fibers while TDI provides an anatomically consistent tract.

With this study we demonstrated that super-resolution TDI maps can provide accurate and useful anatomical information with an exquisite level of details in the brainstem region from in vivo 3T diffusion data. Therefore, anatomists can used these maps as a 3D microscope to explore the brainstem complex organization by identifying subject-specific tracts in healthy humans.

This technique will be a useful resource for clinical research applications once the scanning time needed will be acceptable for patients. In addition, this method will be used in our future studies in order to guide seed/target ROI definition in tractography investigations of the brainstem, to help in the stimulation of white matter pathways with DBS and to better understand neurostimulation interventions in treating a broad range of neurological disorders.