Diffusion Tensor Imaging for Dementia

Ii. Basics Of Dti And Tractography

Water-diffusion anisotropy (directionality) in the white matter of the brain is defined on the basis of axonal alignment (Wiegell, 2000). Water preferentially diffuses in a direction parallel to the axon’s longitudinal axis but is relatively restricted in the perpendicular axis. This phenomenon can be represented mathematically by the so called diffusion ellipsoid, or tensor (Figure 1).

Figure 1.

Diffusion ellipsoids (tensors). When there is no directionality, the fractional anisotropy (FA) is zero (spherical). A typical tensor of a white matter bundle will have the shape of a cigar. When there are crossing fibers, the ellipsoid becomes flattened, resulting in “pancake” tensors (lower left).

The tensor has three eigenvalues. The long one pointing along the axonal direction is λ1, and the two small axes have lengths λ2 and λ3 (Figure 2). The diffusivity along the principal axis λ1 is also called longitudinal, axial, or parallel diffusivity.

Figure 2.

Diffusion constants of a given ellipsoid are shown in this figure. λ1 represents diffusivity in the longest axis of this tensor. The v1 represents the vector orientation of λ1.

The tensors of cerebral white matter can be reconstructed to track three-dimensional macroscopic fiber orientation in the brain. The translation of the longest axis of the tensor (v1) into neural trajectories can be achieved by various algorithms (Figure 3 and Figure 4).

Figure 3.

Tracking starts at a pixel (or ROI). The FACT program tracks the ellipsoids as long as the adjacent vectors are strongly aligned.

Figure 4.

When vector orientation becomes random, as judged quantitatively by the inner products of these vectors, tracking is terminated. The program also terminates when the diffusion ellipsoids approach a spherical shape.

Iii. Limitations Of Tractography

Perhaps the most important limitation of tractography is that it has not yet been fully validated. Attempts to validate this technique have been made in the past (Qazi, 2009; Lin, 2001; Ciccarelli, 2003; Parker, 2002; Okada, 2006), and most of these efforts have been based on comparisons of tractographic images and known neuroanatomy. A study that evaluated deterministic tractography in patients who underwent intra-operative electrophysiological tests indicated that tractography appears to underestimate fiber tracts (Kinoshita, 2005). Thus, this tool has to be used with caution, knowing that we are observing only a fraction of reality.