University of Houston
Department of Computer Science

In partial fulfillment of the Requirements for the Degree of
Doctor of Philosophy


Alok N. Bandekar
will present his thesis proposal

 

A Hybrid Image Segmentation Framework for Medical Images



Abstract

Medical imaging has experienced an explosive growth over the last few years due to technological advances in imaging modalities. Modern imaging technology allows doctors to diagnose and study diseases early by distinguishing between normal and diseased anatomy. However, with an ever expanding spectrum of treatment and diagnostic options, there is an increased demand for objective and accurate quantitative measures. An accurate segmentation allows for good quantitative and morphological analysis. Thus, the development of computer-assisted methods which will provide unbiased and consistent results with minimal human intervention is highly desirable.

 

This dissertation proposes a hybrid medical image segmentation framework based on hierarchical, multi-class, multi-feature fuzzy affinity. Our framework consists of a training phase and a deployment phase. In the training phase, we propose to compute the first order and second order statistics at a variety of scales covering the entire scale spectrum of the image. In the deployment phase, we propose to initialize the target object seed region using domain specific knowledge and then compute a hierarchical fuzzy connectedness-based object. Finally, this framework will be applied on a number of biomedical domains (i.e., automatic segmentation of confocal data from living neurons, and segmentation of CT and IVUS data).

 

 

Date: Tuesday, October 5, 2004
Time: 12:00 PM
Place: 550-PGH

 
Faculty, students, and the general public are invited.
Thesis Advisor: Prof. Ioannis A. Kakadiaris

Committe Members:

Prof. Nicolaos Karayiannis, Prof. Ioannis Pavlidis, Prof. Ricardo Vilalta, and Prof. George Zouridakis