cDNA microarray image segmentation using root signals

Abstract

A vector processing based framework suitable for cDNA microarray image segmentation is introduced and analyzed in this paper. By using nonlinear, generalized selection vector filters the framework proposed here classifies the cDNA image data as either microarray spots or image background. The solution converges to a root signal that represents the segmented cDNA microarray image with the regular spots ideally separated from the background and with their coloration uniquely described by dominant color vectors. It will be demonstrated that the framework readily unifies image denoising, enhancement, data normalization, irregular spot rejection, and spot segmentation in one processing step delivering excellent performance at reasonable computational cost.

A Data-Adaptive Approach to cDNA Microarray Image Enhancement

“…The component x (r,s)1 indicates the R channel while x (r,s)2 indicates the G channel. The two channels are combined to form the cDNA vector x (r,s) in a two-dimensional vector space [2,5]. Thus, the microarray data form a two-channel, Red-Green (RG) image [9,10].…”

“…A number of noise sources mostly in the form of photon noise, electronic noise, laser light reflection and dust on the slide contribute to impairments and defects [2]. These imperfections along with the background fluorescence introduce into the image considerable variability in intensity both within and between the individual spots, as shown in Fig.1.…”

“…cDNA microarray technology [1,2,3] is an advanced tool used to analyze changes in genome-wide patterns of gene expression in different populations of cells. Based on the widely used two-color Cy3/Cy5 system [4,5], the control DNA sequence is usually fluor-tagged with green (Cy3), whereas the experimental sequence of the patient is tagged with red (Cy5).…”

“…Each spot contains multiple copies of a single DNA sequence. The spots occupy a small fraction of the image area and they have to be individually located and isolated from the image background prior to the estimation of its mean intensity [1,2,6].…”

“…In particular, non-specifically bounded DNA or dye molecules and the natural fluorescence of the glass slide result to a substantial noise floor in the microarray image. Since any errors or noise introduced usually limit the performance of the essential analysis procedure, if not defeat it altogether, image processing is necessary in order to eliminate processing errors from propagating further down the processing pipeline to the gene expression analysis tasks [1,2,6,8]. Due to thousands of spots, the microarray image enhancement procedure, such as the one presented in this paper, should be fully automated.…”