Automatic Spot Adressing in cDNA Microarray Images

Abstract

Complementary DNA (cDNA) microarrays are a powerful high throughput technology developed in the last decade allowing researchers to analyze the behaviour and interaction of thousands of genes simultaneously. The large amount of information provided by microarray images requires automatic techniques to develop accurate and efficient processing. Each spot in the microarray contains the hybridization level of a single gene. One of the most important features of these images are the regularity and pseudo-periodicity implicit in the spot arrangement. In this paper, an automatic approach based on texture analysis characterization techniques is proposed to localize spots in microarray images. The method estimates the displacement vectors which characterize the texture (i.e. the spot arrangement). This is achieved by means of applying the generalized Hough transform on the 2D autocorrelation function previously segmented via morphological operations. The obtained displacement vectors are used to generate a grid template which overlaps the original image. The Root-Mean-Square-Error (RMSE) between the estimated locations and the ones computed via a semiautomatic tool is calculated to evaluate the accuracy of the process. The method yields promising results.