Next: Bibliography Up: Level Set Modeling and Previous: Model Properties


Conclusions

In this paper we have developed a computational pipeline for DT-MRI level set modeling and segmentation. We proposed a new rotationally invariant anisotropy measure, which does not require eigenvalue computations. We used the invariants to generate scalar volumes that characterize the total diffusivity and diffusion anisotropy of a DT-MRI scan of a human brain. Applying level set modeling and segmentation techniques to the derived scalar volumes we created geometric models of specific brain structures, e.g. the ventricles, corpus callosum, and the internal capsul. The geometric models were then used for quantitative analysis, including volume and surface area calculations.

We would like to thank Dr. J. Michael Tyszka, Dr. Miriam Scadeng and Dr. David Dubowitz for helping us to identify the 3-D structures extracted from the DT dataset. Dr. Jason Wood developed the Iris Explorer modules used to produce part of the results in the paper. This work was supported by National Science Foundation grants #ACI-9982273 and #ASC-89-20219, the National Institute on Drug Abuse, the National Institute of Mental Health and the NSF, as part of the Human Brain Project, Office of Naval Research Volume Visualization grant #N000140110033, and the National Library of Medicine ``Insight'' Project #N01-LM-0-3503.The first DT-MRI dataset is courtesy of the University of Utah SCI Institute, the second dataset is courtesy of Dr. Mark Bastin, University of Edinburgh, UK. Finally, we would like to thank our reviewers for a very detailed review and multiple valuable suggestions.



Next: Bibliography Up: Level Set Modeling and Previous: Model Properties
Leonid Zhukov 2003-09-14