Microarray Image Processing Using Expectation Maximization Algorithm and Mathematical Morphology

Abstract

Image processing is an important aspect of microarray experiments. Spots segmentation, which is to distinguish the spot signals from background pixels, is a critical step in microarray image processing. After analyzing other means of microarray segmentation, a new method based on Expectation Maximization (EM) algorithm, mathematical morphological filtering and morphological processing is presented. And its corresponding theory and realizable steps are introduced in this paper. Simulations show that the new method for spot image segmentation has better performance than most common ways, such as the ScanAlizeTM method and GenePixTM method. The results of experiments, which are computationally attractive, have excellent performance and can preserve structural information while efficiently suppressing noise in DNA microarray data.