Robust Image Processing Techniques for DNA Microarray Analysis

Abstract

DNA microarray technology is a recently developed, rapidly evolving field which analyses cellular data at the genomic level. It is widely used in the analysis of gene expression levels using which gene sequencing and molecular structure can be studied with a high amount of accuracy and clarity. Image processing plays a critical role in the analysis of microarray. It is used to address feature extraction, gene clustering and data mining and thus aid the analysis of differentially expressed genes. Reliable and robust gridding is a critical step in microarray based studies. Automatic gridding techniques have been conventionally applied to suit rectangular spot arrangements. Some microarrays manufactured using recent technologies have a honeycomb or hexagonal arrangement of gene spots where they cannot be placed along definite rows and hence cannot be subjected to rectangular gridding directly. Our algorithm demonstrates a novel method to improve existing automatic gridding techniques such that the algorithm is applicable to both types of spot arrangements. It operates independent of manual intervention, is simple and has the ability to identify every spot individually. Further, the algorithm performs complete analysis of microarray images. It performs spot location, segmentation, intensity extraction and calculates the gene expression for each spot after normalization using Lowess and Quantile techniques.