Marco Reisert1, Irina Mader2, Constantin Anastasopoulus2, Susanne Schnell1, Valerij Kiselev1
Local fiber tracking approaches are based on the 'walker' principle, the fibres are reconstructed path-by-path by small successive steps along the tracts. On the other hand global ideas try to reconstruct all fibres at once by optimizing a certain global objective. Local algorithms are fast but suffer from accumulated errors. Global methods have a more sound foundation but are very complex to optimize. This abstract presents an approach, which fuses both ideas while keeping their advantages. The experiments show that the approach is orders of magnitude faster than recent global approaches while improving the detection performance.