Diffusion Tensor Visual Analysis of the Human Brain

Magneto-Resonance (MR) Imaging allows retrieving information about tissues by emitting a magnetic pulse and measuring the electro-magnetic radiation as emitted from hydrogenium atoms. Depending on the structure of this magnetic pulse, different hydrogenium atoms are selected and diverse properties of the tissues, in particular depending on the content of water, can be highlighted. In a more elaborated version, a sequence of signals can be combined with a strong gradient magnetic field to retrieve information about the motion of water molecules along the direction of this gradient. This diffusion of water in tissues is not isotropic in general, and especially neuronal fibers constrain diffusion along their orientation. Thus this image acquisition technique yields best results in brain tissues where the density of neuronal fibers is very high. In its most simple form, anisotropic diffusion is modeled by a symmetric tensor field, i.e. a symmetric 3×3 matrix given at each point in a volume as built from six independent measurements. As there are six times more quantities given per point as in a standard MR image, the question rises how to visually best represent these data.