Meeting Banner
Abstract #1675

Probabilistic Connectivity in Fibre Tractography

Parya MomayyezSiahkal1, Kaleem Siddiqi1

1School of Computer Science, Centre for Intelligent Machines, McGill University, Montral, Quebec, Canada

We introduce a probabilistic connectivity index between two regions, based on diffusion MRI, by using a stochastic nonlinear differential equation to model the Brownian motion of water molecules in a medium. The model is linked to the physical basis of the diffusion process and leads to promising results on the MICCAI 2008 Fibre cup phantom. Our experiments yield highly curving fibre tracts without the need to impose thresholds on curvature or torsion or to eliminate false positives. An additional benefit is the algorithm's low computational complexity and the fact that its parameters are data-driven and are selected automatically.