Segmentation and Spot Centering for Microarray Image

In this article we explain segmentation method and spot centering for spot array image based on cDNA chip. First phase of conventional microarray chip analysis consists of detecting strong targets, finding their centers, and regression of four-corner coordinates of each subarray from these centers of strong signals. Detecting strong targets is a process of separation between real target spots and other components, e.g., background and noisy objects. We take a conventional segmentation method in other image processing techniques. This method safely extracts spot targets in spot array image. We explain two categories of segmentation, 4-connected and 8-connected searching. Conventional microarray chip analysis systems generally use center of mass as centers of target spots. Center of mass is a nice method in regular shapes of target spots, e.g, circular or elliptical spots. However because there are various shapes of spots in spot array images, center of mass is not sufficient to use in dealing with every spots. In order to overcome this insufficiency we propose two additional centering methods, center of MBR(Minimum Bounding Rectangle) and center of geometry.