A New Mahalanobis Distance Measure for Clustering of Fiber Tracts

Cheng Guan Koay¹, Carlo Pierpaoli¹, Peter J. Basser¹

In this work, we present a simple and novel generalization of Mahalanobis distance measure for the dyadics of the eigenvector for the purposes of clustering fiber tracts and fiber orientation. This approach is built upon a series of works by Koay et al. on the diffusion tensor estimation and the error propagation framework. The proposed Mahalanobis distance measure for the dyadics is the ideal measure for clustering of fiber tracts as it does not depend on ad hoc combinatorial optimization that is typical in the eigenvector-clustering techniques, which is due to the antipodal symmetry of the eigenvector.