Learn to Track: Deep Learning for Tractography

Philippe Poulin (1) , Marc-Alexandre Côté (1) , Jean-Christophe Houde (1) , Laurent Petit (2) , Peter F Neher , Klaus H Maier-Hein , Hugo Larochelle , Maxime Descoteaux (1)

1 SCIL - Sherbrooke Connectivity Imaging Lab [Sherbrooke]
2 GIN - [GIN] Grenoble Institut des Neurosciences

We show that deep learning techniques can be applied successfully to fiber tractography. Specifically, we use feed-forward and recurrent neural networks to learn the generation process of streamlines directly from diffusion-weighted imaging (DWI) data. Furthermore, we empirically study the behavior of the proposed models on a realistic white matter phantom with known ground truth. We show that their performance is competitive to that of commonly used techniques, even when the models are used on DWI data unseen at training time. We also show that our models are able to recover high spatial coverage of the ground truth white matter pathways while better controlling the number of false connections. In fact, our experiments suggest that exploiting past information within a streamline's trajectory during tracking helps predict the following direction.

Domaines: Neurosciences

Philippe Poulin, Marc-Alexandre Côté, Jean-Christophe Houde, Laurent Petit, Peter F Neher, et al.. Learn to Track: Deep Learning for Tractography. 2017. ⟨hal-03007276⟩