A Study on Denoising Techniques for Microarray Images Priya Nandihal 1 and Dr. Manjunath .S.S 2 1 Assistant Professor, Dayananda Sagar Academy of Technology and Management Email: talk2priya.nandihal@gmail.com 2 Professor and Head, Dayananda Sagar Academy of Technology and Management Email: mnj_ss2002@yahoo.co.in Abstract— Microarray technology has transformed the field of genomic research by allowing the simultaneous profiling of thousands of genes. The microarray process is based entirely on the accurate extraction of quantitative information from images. Several types of noises are caused by the imperfection in generation of microarray images which affects accurate gene expression profiling. Spot recognition is difficult task as noise sources during image acquisition damages the image. Thus, Denoising is one of the major pre-processing steps in microarray image analysis. This paper presents an overview of some of the popular methods used to denoise microarray images both in spatial and frequency domain. The performance metrics used to measure the image quality after denoising is also discussed. Index Terms— Microarray experiment, spatial filters, transform domain filters, performance metrics. I. INTRODUCTION Microarray technology was invented in 1995 and since that, it has been used as an important technology for gene study. Microarray is a chip that contains abundant Deoxyribonucleic Acid (DNA) sequence with its own unique location for each spot, allowing estimation of expression levels of thousands of genes simultaneously [1]. The importance of microarray is to unveil hidden biology of biological processes, monitoring gene expression levels, and for drug and treatment development; for example, therapeutic drugs for gene expression levels of cancer. By analyzing and comparing normal versus abnormal microarray gene expression profiling, genes involved in that particular disease can be identified. Due to its importance in pharmaceutical and clinical research, many applications for microarray have been developed to analyze them. Microarray can be processed through three steps [2]: (i) gridding, which is a process of assigning the location of each spot, (ii) segmentation, which is a process of grouping the pixels with similar features and finally (iii) information extraction, which calculates red and green foreground intensity pairs and background intensities. These experiments, like any other, are prone to noise. Measurement of gene expression levels can be influenced by the noise introduced to the data during the preparation, hybridization and scanning phase. Additive or multiplicative Gaussian, Poisson and exponential noise models have been used to describe the noise which affects microarray images [3][4][5] Several methods have been proposed for eliminating and reducing the noise [6], [7] in microarray images. Two popular domains are the spatial filtering and the frequency domain approach. In the spatial filtering Grenze ID: 02.ICSIPCA.2017.1.51 © Grenze Scientific Society, 2017 Int. Conf. on Signal, Image Processing Communication & Automation, ICSIPCA