The microarchitecture specific to nerve tissues causes diffusion anisotropy in the white matter of the brain: water molecule diffusion preferably follows the direction of the fibers and is restricted perpendicularly to the fibers.

By performing diffusion-weighted acquisitions in at least 6 directions (and far more in angular high resolution imaging), it is possible to extract the diffusion tensor which synthesizes all the data.

The diffusion tensor characterizes diffusion: anisotropic coefficient, preferred directions and restrictions in space. Different images will be obtained depending on the complexity of the post-processing of this data :

  • Fractional anisotropy (null when diffusion is isotropic, of increasing value when diffusion becomes anisotropic)
  • Main diffusion direction
  • Fiber tracking


Local modeling of the diffusion phenomenon

The limitations of the tensor model in precisely determining the directions of water molecule displacements in the case of fiber crossing, for example, have led to the development of new models, which require a greater number of measurements :

  • Multitensor model (several diffusion tensors coexisting in a voxel)
  • Q-ball (S-space): requires a large number of acquisitions in different directions, with a constant b-factor
  • Q-space (Diffusion spectrum imaging) (figure 13.11): the ultimate technique, able to describe fiber crossings, but requiring a high number of acquisitions (129 to 515 !) in different directions and with different b-factors to sample the diffusion equivalent of a 3D k-space