Structure Tensor Informed Fiber Tractography (STIFT) by combining gradient echo MRI and diffusion weighted imaging.
Kleinnijenhuis M., Barth M., Alexander DC., van Cappellen van Walsum AM., Norris DG.
Structural connectivity research in the human brain in vivo relies heavily on fiber tractography in diffusion-weighted MRI (DWI). The accurate mapping of white matter pathways would gain from images with a higher resolution than the typical ~2mm isotropic DWI voxel size. Recently, high field gradient echo MRI (GE) has attracted considerable attention for its detailed anatomical contrast even within the white and gray matter. Susceptibility differences between various fiber bundles give a contrast that might provide a useful representation of white matter architecture complementary to that offered by DWI. In this paper, Structure Tensor Informed Fiber Tractography (STIFT) is proposed as a method to combine DWI and GE. A data-adaptive structure tensor is calculated from the GE image to describe the morphology of fiber bundles. The structure tensor is incorporated in a tractography algorithm to modify the DWI-based tracking direction according to the contrast in the GE image. This GE structure tensor was shown to be informative for tractography. From closely spaced seedpoints (0.5mm) on both sides of the border of 1) the optic radiation and inferior longitudinal fasciculus 2) the cingulum and corpus callosum, STIFT fiber bundles were clearly separated in white matter and terminated in the anatomically correct areas. Reconstruction of the optic radiation with STIFT showed a larger anterior extent of Meyer's loop compared to a standard tractography alternative. STIFT in multifiber voxels yielded a reduction in crossing-over of streamlines from the cingulum to the adjacent corpus callosum, while tracking through the fiber crossings of the centrum semiovale was unaffected. The STIFT method improves the anatomical accuracy of tractography of various fiber tracts, such as the optic radiation and cingulum. Furthermore, it has been demonstrated that STIFT can differentiate between kissing and crossing fiber configurations. Future investigations are required to establish the applicability in more white matter pathways.