An NMF based approach for addressing DNA microarray sub-grid images

Microarray technology is a very powerful tool. In one experiment, it gives researchers information about the function of thousands of genes by analyzing the scanned images of DNA microarray. Image analysis is an important step to quantify the genes' expression levels and to give accurate information to researchers. In general, addressing, segmentation and intensity extraction are the three main steps which are done during the analysis of the scanned images. In this paper we propose a new algorithm for accurate and automatic addressing of DNA microarray sub-grids images that is crucial for high-throughput analysis. The proposed algorithm uses the non-negative matrix factorization (NMF) and circular binary segmentation (CBS) to subdivide the image into small blocks. The algorithm uses Permutation tests to find out the blocks which have spots. The experimental results using real DNA microarray images show that our algorithm is a promising and has high accuracy.