Robust Extraction of Fiber Skeleton Based on Whole Fiber Tensor Information and Active Contour Method

Wu Li¹, Xiaoping P. Hu¹

In DTI data analysis fiber-based quantitative analyses are becoming more widely used for cross subject comparison. To detect the fiber trajectory, tractography methods or skeleton extracted from FA are usually used. However, fiber tracking methods track fibers according to local diffusion orientation, which are highly sensitive to noise and leading to cumulative errors. We proposed an approach to extract fiber trajectory based on an active contour model, taking into account both fiber regularity and measured DTI data. Results show that this method leads to improved fiber trajectory, which can potentially improve the accuracy of fiber based analysis.