Improving Diffusion Tensor Imaging Segmentation Through an Adaptive Distance Learning Scheme

In segmentation techniques for Diffusion Tensor Imaging (DTI) data, the similarity of diffusion tensors must be assessed for partitioning data into regions which are homogeneous in terms of tensor characteristics. Various distance measures have been proposed in literature for analyzing the similarity of diffusion tensors (DTs) [1], but selecting a measure suitable for the task at hand is difficult and often done by trial-and-error. We propose a novel approach to semiautomatically define the similarity measure or combination of measures that better suit the data. We us a linear combination of known distance measures, jointly capturing multiple aspects of tensor characteristics, for comparing DTs with the purpose of image segmentation. The parameters of our adaptive distance measure are tuned for each individual segmentation task on the basis of user-selected ROls using the concept of Kernel Target Alignment [2]. The results of the presented method can then be used in any segmentation algorithm as, for example region growing.

@InProceedings { RVTH09,
  author       = "Rodrigues, P. and Vilanova, Anna and Twellmann, T. and Haar Romeny, B.M. ter",
  title        = "Improving Diffusion Tensor Imaging Segmentation Through an Adaptive 	Distance Learning Scheme",
  booktitle    = "Proceedings 17th Scientific Meeting, International Society for Magnetic 	Resonance in Medicine, Honolulu",
  pages        = "1431",
  year         = "2009",
  url          = "http://graphics.tudelft.nl/Publications-new/2009/RVTH09"
}