Search

Separation of front ground from the background of the spot is the secondary step in microarray investigation. Microarray spots are useful in determining the differential gene expression of given sample. In this article, a new approach for spot segmentation of microarray images using global intensity based model is proposed. A global intensity based model is used to resolve the issues like curve peak producing, curve breaking and combining etc. Before applying proposed method on spot, circular mask is calculated and edges of each spots are identified. Initially orthogonal projection from every edge point to every other point is drawn so that region of interest of spot identified. Next using external force energy function is used to cover all border points. The presented work is found to be accurate when compared to the methods like Fuzzy C-means, Morphological segmentation.

Microarrays are an influential, maximum throughput tool which has motivated researchers to investigate thousands of genes during the last decade. The large quantity of information presented by microarray images necessitates implementation of automated techniques for proficient processing of microarray images, in order to draw an improved biological conclusion. The majority of the techniques discussed in the literature demand manual intervention, which unavoidably reduces the efficiency and reproducibility of the automation process. The presented novel approach mainly focuses on gridding of noisy microarray images automatically. The projection profiles of the binarized image are obtained. The unduly nonuniform distance between grid lines in noisy microarray images are corrected using the coefficient of variation (CV) of successive differences in the gridding locations. Noisy microarray images from various databases were acquired and tested with the presented approach. Keywords: Microarray, Gridding, Binary image, Projection profile, Coefficient of variation