Human brain tractography: A DTI vs DKI comparison analysis

Abstract

Water diffusion can be measured using Diffusion Tensor Imaging (DTI). This technique estimates molecular diffusion using a Gaussian model which can account for anisotropy arising, for example, from the presence of myelin sheath. The main direction of diffusion can also be estimated from DTI and used to compute diffusion tracts (tractography), a good tool to analyse the structure of white matter brain pathways. Nevertheless, DTI has limitations, such as ignoring the non-Gaussian properties of biological tissues and the inability to resolve intra-voxel fibre crossings that may lead to the reconstruction of anatomically inaccurate tracts. Diffusion Kurtosis Imaging (DKI) was introduced to ameliorate these problems. By abandoning the Gaussian model, DKI acts as an extension of DTI, allowing the computation of the same scalar parameters as DTI as well as providing measures of tissue heterogeneity and resolving multiple tracts within each voxel[1]. The purpose of this study is to analyse the DKI performance in tractography comparing to the DTI approach. DKI data was acquired from 3 subjects (1 male, mean age: 39±14 years) using Philips Achieva® 3.0T including diffusion weighted (DW) Single Shot Echo Planar images using 64 directions with b-values 0, 1000 and 2000 s/mm2, reconstruction matrix 256×256, slice thickness 1.5 mm, TE/TR 64/7703 ms, FOV 240×240 mm2, 60 slices. DTI data was extracted from DKI considering only b-values 0 and 1000. All DTI and DKI based tracts were computed using DKIu, a Matlab toolbox created by Neto-Henriques et. al., using three different tractography approaches: DTI, Laz-DKI[2] and KT-DKI[1]. The structures considered were the Internal Capsule (IC) and the Corpus Callosum (CC). Using TrackVis for visualisation, DKI tractography was shown to enable improved fibre crossing resolutio n (Figure 1), consistent in all three subjects. The number of streamlines computed in the CC was higher with the DKI approach than with DTI, but lower in the IC. DKI-based tractography appears to successfully address some of the limitations of DTI, as it resolves crossing fibres, providing more useful information regarding the brain's microstructure. This is particularly important in pre-surgical planning and in identifying brain lesions and pathologies.