Skip to main content

Can We Get Around the Crossing Fiber Problem by Increasing Spatial resolution?

Posted by on Thursday, September 1, 2016 in Crossing Fibers, Diffusion Tensor Imaging, Diffusion Weighted MRI.

Kurt G. Schilling, Vaibhav Janve, Yurui Gao, Iwona Stepniewska, Bennett A. Landman, Adam W Anderson. “Can We Get Around the Crossing Fiber Problem by Increasing Spatial resolution?” In Proceedings of the ISMRM Workshop on Diffusion MRI. Lisbon, Portugal, September 2016. Oral presentation.

Full text: PDF

Abstract

It is now widely recognized that voxels with crossing fibers and complex configurations present a challenge for diffusion MRI (dMRI) reconstruction and fiber tracking algorithms. This “crossing fiber” problem has been estimated to affect anywhere from 30% [1] to 90% [2] of white matter (WM) voxels, and it is generally assumed that increasing the spatial resolution as much as possible will minimize the number of voxels containing multiple fiber populations. Here, using both dMRI and histology from the same brain, we investigate the prevalence of this crossing fiber problem across a range of spatial resolutions.

Figure 1 shows the results of ST analysis in a region with in-plane fiber crossings. We find that, in both MRI and histology, the percent of voxels with crossing fibers actually increases as we increase the resolution. The middle and bottom rows highlight “single” fiber regions in red and “crossing” fibers as green, and qualitatively show that many regions with a “single” fiber population at a coarse resolution exhibit crossing patterns when processed at a finer resolution.
Figure 1 shows the results of ST analysis in a region with in-plane fiber crossings. We find that, in both MRI and histology, the percent of voxels with crossing fibers actually increases as we increase the resolution. The middle and bottom rows highlight “single” fiber regions in red and “crossing” fibers as green, and qualitatively show that many regions with a “single” fiber population at a coarse resolution exhibit crossing patterns when processed at a finer resolution.