Faculty Host: Dimitris Metaxas

Abstract:

This Talk will concentrate on our image analysis research with emphasis on its application to the two important problem domains, medical and bioinformatics image analysis.

Deformable contours (snakes) are used to extract boundaries of regions and track them while structures in an image undergo nonrigid motion. In this talk, we present two novel frameworks for the snake formulation: spatiotemporal analysis formulation, and multiple snakes formulation. Our formulation has been successfully applied to track tongue and heart wall boundaries from ultrasound imagery. Multiple snake formulation has been applied to extract boundaries of gene spots in microarray images to quantify the expression level.

Our recent work includes spline-based nonrigid motion and point correspondence recovery method for 3D surfaces. This method is based on differential geometry. Shape information is used to recover the point correspondences. In contrast with the majority of shape-based methods which assume that shape (normal, curvature) changes are minimum after motion, our method focuses on the nonrigid relationship between before-motion and after-motion shapes.

Lastly, our work on Protein Docking is presented. This involves finding a binding configuration by which two proteins fit together in 3D space. Protein docking is considered a difficult problem because of (i) huge search space of docking binding sites, and (ii) complexity of the docking formulation. Therefore, computer-aided analysis and prediction of these interactions are becoming increasingly important. We thus study and apply novel techniques based on computer graphics and computer vision to work for the problem of protein docking. Our method(s) reduce search space hugely by docking patch-to-patch, based on high-level geometry information.

For details of other projects, visit: http://vims.cis.udel.edu

Speaker Bio:

Dr. Kambhamettu is an Associate Professor in the Computer and Information Sciences Department and Director of Video/Image Modeling and Synthesis (VIMS) Lab. at the University of Delaware. From fall 1997 to spring 2003, he was Assistant Professor in the same department. From 1994-96, he was a research scientist at NASA-Goddard, where he received the "1995 Excellence in Research Award". He received his Bachelor of Engineering degree in Computer Science and Engineering from Osmania University (India) in 1989, his MS and PhD in Computer Science and Engineering from the University of South Florida in 1991 and 1994 respectively. Dr. Kambhametu received NSF CAREER award in 2000 and is the associate editor of PATTERN RECOGNITION journal. His interests include Computer vision, Medical and Bioinformatics Image analysis, Visualization and Computer Graphics.