Abstract
The microarray technology was born for dealing with a lot of genes. It is developed to reveal various genes related to fatal diseases what people want to discover. The purpose of this thesis is to implement a system for dealing with microarray image pattern analysis fast, accurately and repeatedly. And the system provides the following function: (i) Detect differentially expressed genes. (ii) Select a subset of genes which best distinguishes different diseases. (iii) Visualizing our experimental results by the dendrograms. In this thesis, we use two different methods, Otsu and GMM to do gene expression computation. And then we use LOWESS and Moving Average Filtering to do normalization. Finally, we use Fisher linear discriminant for finding differentially expressed genes and check the results by K-means and Hierarchical clustering.