A New Gridding Technique for High Density Microarray Images Using Intensity Projection Profile of Best Sub Image

Abstract

As the technologies for the fabrication of high quality microarray advances rapidly, quantification of microarray data becomes a major task. Gridding is the first step in the analysis of microarray images for locating the subarrays and individual spots within each subarray. For accurate gridding of high-density microarray images, in the presence of contamination and background noise, precise calculation of parameters is essential. This paper presents an accurate fully automatic gridding method for locating suarrays and individual spots using the intensity projection profile of the most suitable subimage. The method is capable of processing the image without any user intervention and does not demand any input parameters as many other commercial and academic packages. According to results obtained, the accuracy of our algorithm is between 95-100% for microarray images with coefficient of variation less than two.  Experimental results show that the method is capable of gridding microarray images with irregular spots, varying surface intensity distribution and with more than 50% contamination.

Keywords: microarray, gridding, image processing, gridding accuracy