Robust graph-based tracking through crossing fibre configurations

Graph-based distributed tractography of the brain provides an alternative to streamline approaches. However, graph-based tracking through complex fibre configurations has not been extensively studied and existing methods have inherent limitations. In this study, we discuss these limitations and present a new approach for robustly propagating through fibre crossings, as these are depicted by the Q-ball orientation distribution functions (ODFs). Complex ODFs are decomposed to components representative of single-fibre populations and an appropriate image graph is created. Path strengths are calculated using a modified version of Dijkstra´s shortest path algorithm. A comparison with existing methods is performed on simulated and on human Q-ball imaging data.