TITLE: Geometric Tools for Knowledge Discovery from Biomedical Images

ABSTRACT:
Advance in imaging technology has allowed biological structures at all scales - ranging from proteins to organs - to be captured in digital forms. Researchers are often interested in questions like "where is structure X in an image?" or "how do the structures in two images compare?".

The employment of current methodologies are often hampered by practical challenges such as the complexity of the biological structure, low imaging resolution, and noise. In this talk, I will present the novel use of two geometric tools that we found particularly suited for modeling and understanding a class of biological structures in biomedical images. The first tool is the medial skeleton, a shape descriptor that robustly captures and differentiates shape elongations, such as the tubular and plate-like parts commonly found in biological forms. The second tool is subdivision meshes, a hierarchical geometry that has unique advantages for deformable registration and spatial data mining among anatomical images. I will demonstrate the application of these tools in several biomedical projects, including identifying 3D structures of molecular complexes from volumetric reconstructions, mapping gene expression patterns among tissue sections, and monitoring bone density changes in CT volumes.

BIO:
Tao Ju is an Assistant Professor in the Department of Computer Science and Engineering at the Washington University in St. Louis (USA). He obtained his PhD degree in Computer Science from Rice University (Houston, USA) in 2005, and his Bachelor degree from Tsinghua University (Beijing, China) in 2000. His main research area is computer graphics, with focuses on geometric processing, animation, and shape modeling. He is also interested in application of geometric techniques in biomedical modeling and image analysis. His work is supported by funds from NSF and NIH, and he received a NSF CAREER Award in 2009.