Tractography Clustering for Fiber Selection in ROI-based Diffusion Tensor Studies

Estimating white matter ber pathways from a diffusion tensor MRI dataset has many important applications in medical research. Even after the definition of white matter ROIs, the precise selection of appropriate fibers of interest for further analysis in population studies is often a time consuming and error prone task. Tractography segmentation methods based on pairwise distances between fibers are computationally expensive and have limitations addressing this problem. This paper proposes a new method to automatically perform fiber tract selection for ROI-based tractography analysis. The tractography segmentation is based on the representation of the fibers by means of their orientation at the closest point to the centroid of the associated ROI. Results and comparisons on two real-world fiber tract selection problems show the proposed method to constitute a simple yet effective tool for addressing this important problem.