Probabilistic Tracking U-fiber on the Superficial White Matter Surface

Abstract

The short association fibers or U-fibers connect two neighboring gyri and travel in the superficial white matter (SWM) beneath the cortical layer. These U-fibers are essential for the understanding of neurodevelopment and neurodegeneration. However, the complex structures and the high curvature of the U-fibers lead to erroneous streamlines reconstruction of the traditional tractography since the volume-based tractography cannot use the biological characteristic of U-fibers that they tightly beneath the cortical layer. In this work, we proposed a surface-based framework for probabilistic tracking of the U-fibers on the triangular mesh of the SWM. We develop a novel approach to project the fiber orientation distributions (FODs) data onto the tangent space of the SWM surface. With the projected FODs, an advanced probabilistic tracking technique, which regularizes the streamlines based on the intrinsic geometry of the surface, is developed to reconstruct the highly bent U-fibers on the SWM surface. In the experiment, we demonstrate our method based on the high-resolution diffusion imaging data from the Human Connectome Project (HCP). We quantitatively compare the proposed method with state-of-the-art volume-based tractography from MRTrix and another surface-based tractography on the U-fibers of the central sulcus. Moreover, we show the reconstructed U-fibers on the parietal lobe and frontal lobe. The results show that our method outperforms the other two methods and successfully reconstructs the U-fibers on the cortical regions with high intersubject variability.