Improved reference tracts for unsupervised brain white matter tractography

Abstract

Neighbourhood tractography aims to automatically segment equivalent brain white matter tracts from diffusion magnetic resonance imaging (dMRI) data in different subjects by using a “reference tract” as a prior for the shape and length of each tract of interest. In the current work we present a means of improving the technique by using references tracts derived from dMRI data acquired from 80 healthy volunteers aged 25–64 years. The reference tracts were tested on the segmentation of 16 major white matter tracts in 50 healthy older people, aged 71.8 (±0.4) years. We found that data-generated reference tracts improved the automatic white matter tract segmentations compared to results from atlas-generated reference tracts. We also obtained higher percentages of visually acceptable segmented tracts and lower variation in water diffusion parameters using this approach.