Brain fibre tracking improved by diffusion tensor similarity using non-Euclidean distances

Fibre tracking is a non-invasive technique based on Diffusion Tensor Imaging (DTI) that provides useful information about biological anatomy and connectivity. In this paper, we propose a new fibre tracking algorithm, named TAS (Tracking by Angle and Similarity), which is able to overcome the shortfalls of existing algorithms by considering not only the main diffusion directions, but also the similarity of diffusion tensors using non-Euclidean distances. Quantitative comparison is carried out through a collection of simulation experiments using statistics of diffusion tensor anisotropy and volume, and tracking errors. Fibre tracking in Corpus Callosum from a healthy human brain dataset is presented.