Workshop on Introduction to DTI using FSL

As water moves freely in cerebrospinal fluid, isotropically along small distances in neuronal cell bodies, but freely along axons direction in white matter, water movement

can then be used to detect white matter and quantify its main properties. This can be done by fitting a diffusion tensor on the diffusion-weigted images, which comes under the

general term of diffusion tensor imaging (DTI). Those values are then usually used to compute fractional anisotropy (a value characterizing water movement direction in voxels),

or to fit more complex models capable of reconstructing white matter fiber directions and even fiber crossings. Diffusion tensor values fitted at each voxels are then built

upon to reconstruct whole fiber tracts, to quantify the effect of learning on white matter structure, or even to compare white matter across groups (e.g. aging subjects vs controls).

This can be done with a large number of MRI analysis packages available online.

In this workshop, we will introduce DTI analysis using the FSL library which differentiates itself by a Bayesian probabilistic approach in fitting parameters at each voxel

and reconstructing fiber tracts. This package is often cited as being more sensitive than more deterministic approaches used by other packages, and can be of special use to

researchers working on brain areas related to language. More precisely, we will start by presenting the basics of diffusion-weighted imaging and the general theory leading to fiber tract

reconstruction with FSL. We will then look at how to set up and perform a DTI analysis with FSL, and then how to interpret their results.