An Efficient Clahe-based, Spot-adaptive, Image Segmentation Technique for Improving Microarray Genes’ Quantification

An efficient spot-adaptive segmentation technique was developed by suitable combining in a cascade mode the benefits of image enhancement (Contrast Limited Adaptive Histogram Equalization technique (CLAHE)) and image segmentation (Seeded Region Growing technique (SRG)) in order to improve genes’ quantification in microarray images. Microarrays utilized for evaluation purposes comprised 7 publicly available images. Initially, an image griding algorithm was employed to divide the image into rectangular image-cells. Subsequently, CLAHE was applied on each individual image-cell, initial SRG-seed was set at the image-cell’s center, and SRG-threshold was estimated from the image-cell’s background. The spot’s boundary was referred to the corresponding cell spot in the original image and the spot’s intensity was evaluated. Extracted intensities were comparatively evaluated against a well-established commercial software package (MAGIC TOOL) employing the Jeffrey’s divergence-metric. The metric of the spot-adaptive segmentation technique was about double as compared to MAGIC TOOL’s metric, with differences ranging between 1.23 and 5.21 in the processed images. Regarding processing time, the proposed method required half the time of MAGIC TOOL’s (211 secs against 487 secs) to process the same cDNA image on the same computer.