Parallel Particle Advection and Lagrangian Analysis for 3D-PLI Fiber Orientation Maps

3D-Polarized Light Imaging (3D-PLI) is a recent neuroimaging technique which is able to record the orientation of nerve fibers in unprecedented detail, reaching micrometer scale. Particle advection, which streamline tractography is based upon, is an established tool to study the structure of fiber tracts within the brain. This work presents a distributed particle advection pipeline for 3D-PLI fiber orientation maps, which are time-invariant vector fields describing the dominant orientation of nerve fibers at each point. Due to the high resolution, 3D-PLI creates large datasets ranging from few gigabytes up to hundreds of terabytes, which is addressed through a data-parallel approach to advection. Furthermore, we propose Lagrangian analysis of time-invariant brain data through computation of Finite-Time Lyapunov Exponents (FTLE), which may aid identification of macroscopic fiber bundles.