The effect of the number of fibers in tractography reconstruction of white matter bundles.

Abstract

The study of white matter (WM) through diffusion Magnetic Resonance Imaging (dMRI) is crucial to obtain a better understanding of human brain connections and functions, at a macroscopic level. A large number of works have focused on long range brain connections, while recently, several studies have also analyzed superficial WM connectivity. In recent years, with the massive use of HCP database, and its processing with known softwares like DSI Studio and MRtrix, it is necessary to evaluate the influence of tractography parameters on the reconstruction of fiber bundles and further analyses. We study the effect of the number of fibers, for whole brain tractography, on the reconstruction of deep and superficial WM bundles based on their segmentation using multi-subject bundle atlases. For DSI Studio (deterministic algorithm), a value of 1M fibers could reconstruct most of deep white matter (DWM) bundles, while a value of 1.5M was required for superficial white matter (SWM) bundles. In the case of MRtrix (probabilistic algorithm), a value of 3M fibers was found to be suitable for the study of both kinds of fibers. Furthermore, we found the tracking of SWM bundles to be more sensitive to several parameters than DWM, for DSI Studio.