Improving Diffusion Tensor Imaging Segmentation Through an Adaptive Distance Learning Scheme

Abstract

Similarity of diffusion tensors is a crucial point in several applicationslike segmentation, group statistical analysis, etc. The selectionof the most suitable measure, for a given task, is not always clearand often done by trial and error. We present a proof of conceptof an initially flexible learning scheme that infers the measureor combination of measures that achieves the best discriminationbetween a user selected Region of Interest and a representative setof tensors of the whole volume. The results demonstrate the method'spotential to infer the ideal parameters for a task specific segmentationalgorithm, region growing.