Vector edge operators for cDNA microarray spot localization

This paper introduces a vector-based framework for edge detection and spot localization in cDNA microarray data. Since cDNA microarray images can be viewed as vector fields, both their spectral and spatial characteristics should be used to determine edges, discontinuities and structural elements. Building upon the powerful nature of nonlinear operators, the proposed vector edge operators can effectively localize microarray spots outperforming the commonly used scalar edge detectors. Moreover, due to the utilization of the principle of robust statistics, vector edge detectors are relatively immune to the noise present in microarray images. Simulation studies reported in this paper indicate that the proposed framework yields excellent performance and it can be readily incorporated in the cDNA microarray processing pipeline.