Visualizing MR diffusion tensor fields by dynamic fiber tracking and uncertainty mapping

Recent advances in magnetic resonance imaging have provided methods for the acquisition of high-resolution diffusion tensor fields. Their 3D-visualization with streamline-based techniques-called fiber tracking-allow analysis of cerebral white matter tracts for diagnostic, therapeutic as well as neuro-scientific purposes. The illusiveness of fiber visualizations and the inability to reliably visualize branching structures are problems still waiting for solutions. In this paper we present an on-the-fly approach to the tracking of branching and crossing fibers by dynamically setting secondary seeds in regions where branching is assumed, thus avoiding computationally intensive preprocessing steps. Moreover, we propose an uncertainty mapping technique that uses color-coding to enrich 3D fiber displays with information on their validity. Probability values for fiber samples are Computed from dataset features as well as characteristics of the tracking process. In contrast to data optimization and pre-processing approaches, our algorithms focus on highly interactive visualization scenarios in collaborative environments. (c) 2006 Elsevier Ltd. All rights reserved.