A Microarray Image Analysis System Based on Multiple-snake

Microarray technology is a powerful tool that allows scientists to study expression levels for thousands of genes simultaneously. The technology has been useful in many applications, e.g., disease diagnosis, drug discovery, and gene functional study. In this paper, we present a microarray image analysis system which works specifically on nylon membrane microarray. These membrane microarray images have problems that are different from glass slide microarray images. Some of the problems are that spot sizes are very small due to the low image resolution, spots could be merged into one another, images could be noisy, and that spots could occur in various sizes. The system has been developed to handle (i) automatic image alignment and gridding, (ii) spot contour detection, and (iii) intensity measurement. The alignment and gridding system is automated with possible gridding provided for microarray images. In spot contour detection, we apply the multiple-snake method, which is the high-level segmentation method, to automatically extract the contours of multiple spots. In intensity measurement, different ways to estimate the intensity are used and compared. In the experiments, various designs of microarray images have been tested. The reliability of the system is determined by comparing the results of duplicated pairs of spots. We tested robustness of the system with a set of noisy microarray images at different percentages of Gaussian noise. We also tested the system with glass slide microarray images, and the results are very encouraging.