Section: New Results

Image processing on Diffusion Weighted Magnetic Resonance Imaging

Clustering and classification ofwhite matter fibers from Diffusion Tensor Imaging (DTI)

Participants : Meena Mani, Christian Barillot.

This project can be broken down into three major aspects:

• Spectral clustering of DTI fibers using different distance metrics. Two distance metrics, a mean closest point (MCP) and a barymetric distance were found to give the best clustering results on data sets such as the corpus callosum.

• Spectral clustering using the Nystrom approximation to handle large data sets. This included an investigation of the error involved when the Nystrom approximation is used.

• Clustering and statistical quantitative analysis of DTI fibers using a comprehensive Riemannian framework that allows for a joint analysis of features in a consistent manner. This work was done with Professor Anuj Srivastava at Florida State University. A paper based on this work was submitted to the ISBI 2010 conference.