Fiber Bundle Segmentation using Major Diffusion Orientations in Reduced Position Orientation Space

Esmail Davoodi-Bojd¹, Mohammadreza Nazem-Zadeh², Hamid Soltanian-Zadeh¹, Quan Jiang²

Position Orientation Space (POS) is composed of three dimensions for position and two dimensions for orientation information in each voxel. The POS can be very helpful in low resolution datasets. If two fiber tracts are irresolvable sharing crossing area in three dimensional space, they may be resolvable in five dimensional POS. Since four distinct peaks of the Orientation Distribution Function (ODF) are adequate for extracting the fibers, we propose a novel strategy to the compute major directions from the ODF, then they are employed for segmenting fiber bundles by an algorithm similar to that of Hagmann et al.