Noise Estimation and Reduction in Microarray Images

Lots of error sources affect the microarray image quality, especially the noise. An image may contain different type noises which will produce distinct influence on image processing, so it doesn’t need to remove all. This paper analyzed the affection of different noises to automatic gridding and proposed grid line number for quantitive evaluation. A new algorithm for noise reduction was developed, which included two parts: edge noise reduction and highly fluorescence noise reduction. Edge detection was executed on the vertical and horizontal projections of microarray image. Highly fluorescent noise was removed by linear replace, which is an easy and fast means. The algorithm was implemented and compared to other common noise reduction methods. Experiment results show the feasibility of the proposed approach.