Date of Award

Spring 2010

Document Type


Degree Name

Doctor of Philosophy (PhD)


Computational Analysis and Modeling

First Advisor

Andrei Paun


The goal of this dissertation is to build a better segmentation method for DNA microarray image processing. Segmentation is a partitioning process used to separate a spot area from a non-spot area in DNA microarrays. It directly affects the accuracy of gene expression analysis in the data mining process that follows. A number of DNA microarray segmentation methods have been proposed in the area, but even modern segmentation methods seem to have accuracy problems. In this dissertation, I will present a segmentation method based on the Active Contours Without Edges (ACWE) algorithm and apply it to two types of DNA microarrays, complementary DNA (cDNA) and Affymetrix GeneChip. Several adjustments will be applied to the original ACWE method to use it more efficiently in the microarray processing area.

As a secondary research objective, I will improve the ACWE method by using higher order schemes in finite difference method for solving the partial differential equation (PDE). The original ACWE method used the associated Euler-Lagrange partial differential equation for the Lipschitz function Φ. It used the lower order finite difference schemes to solve the PDE. The improved ACWE method defines the higher order finite difference schemes to increase the accuracy of segmentation.

Various experimental results will be presented to show that the ACWE method is more efficient than other DNA microarray image segmentation methods.

Statistical analysis is performed to compare the newly proposed method with the previously best methods in this area. Experimental results will also be presented to show that the improved ACWE method has more accurate segmentation results than the ACWE method.