This abstract is focused on investigating the ability of a local analysis performed on Track Density Imaging (TDI)¹ to reveal subtle differences in the white matter architecture that are not easily detected using conventional diffusion MRI analysis tools.

TDI can provide super resolution images of the white matter. Although it has been employed to investigate anatomical features and histopathologic correlations have been made, to the best of our knowledge it has almost never been used to discover white matter abnormalities related to pathological conditions in group studies (with the exception of Ziegler et al.²).

Group studies that make use of conventional scalar measures (fractional anisotropy, mean diffusivity,…) to perform comparisons accross subjects only exploit one feature of diffusion at a time. We hypothesize that TDI can encompass more features of the diffusion information, and thus will be more sensitive to small local changes in the white matter.

After DWI data acquisition, fiber-tracking was performed using the MRtrix software package (http://www.mrtrix.org). Spherical harmonics were computed from the DWIs using constrained spherical deconvolution³ and fed into the fiber-tracking algorithm in order to model multiple fiber orientations. This way, whole-brain streamline tractography was performed randomly placing a large number of seeds throughout the brain. (500000 fibers were obtained in our experiments). TDI then computes the total number of fiber tracks that lie within each element of a grid, whose size can be smaller than the original acquired voxel size.

In order to perform group comparisons, FA maps of all subjects were also computed and resampled to the same resolution as the TDI maps. Tract-Based Spatial Statistics (TBSS) were then employed, projecting the FA maps from all subjects onto a mean FA skeleton⁴. Comparisons, which are restricted to this FA skeleton, were then made on the FA maps and the TDI maps.

Two different group studies were carried out to illustrate the potential of this approach, using two different datasets:

Experiment 1: 17 healthy controls (10 female, 7 male, 74.5 ± 3.5 years) and 19 patients (12 female, 7 male, 76.1 ± 2.7 years) diagnosed with mild Alzheimer’s disease were selected. Patients were diagnosed according to NINCDS-ADRDA Alzheimer's Criteria. There were no significant age differences between the two groups. Diffusion weighted images were acquired in a GE Signa 1.5 T MRI unit at QDiagnóstica, Valladolid, Spain. The parameters of the acquisition protocol were the following: 25 gradient directions, one baseline volume, b = 1000 s/mm2, 1.015 × 1.015 × 3 mm3 of voxel size,TR = 13,000 ms, TE = 85.5 ms, 256 × 256 matrix, NEX = 2 and 39 slices covering the entire brain. DWIs were processed as previously indicated.

Experiment 2: 18 patients (17 female, 1 male, 36.6 ± 11.8 years) with chronic migraine and 10 patients (6 female, 1 male, 33.5 ± 7.7 years) with episodic migraine were selected, having been diagnosed according to ICHD III edition. There were no significant age differences between the two groups. The parameters of the acquisition protocol and the DWI processing pipeline were identical to those employed in the previous experiment.

No significant differences in the FA maps were found between groups in both experiments.

For the Alzheimer’s disease group study, however, local analysis of TDI was able to locate differences in widespread areas within the white matter. Patients showed a decrease in the track density in the areas depicted in Figure 3. There is extensive literature regarding white matter changes in Alzheimer’s disease, and widespread alterations have been found throughout the white matter, although Mean Diffusivity has shown a higher discrimination power than Fractional Anisotropy.

With regard to the migraine group study, local analysis of TDI located a decrease in the track density at the anterior part of the corpus callosum in patients with chronic migraine with respect to those with episodic migraine (see Figure 4). There is very few literature analyzing differences between different types of migraine patients using dMRI but, remarkably, our results are consistant with those in⁵, where a different approach was employed.

Local analysis of Track Density Imaging can reveal subtle differences in white matter group studies. A processing pipeline combining TDI and TBSS was proposed and tested on two different datasets where conventional TBSS analysis on the FA did not yield any findings. Results showed the proposed approach to be able to detect subtle changes in the white matter, with increased sensitivity with respect to FA analysis. The full potential of local analysis of TDI needs to be further investigated, as well as the interpretation of the differences found.