Segmentation algorithms for spotted microarray images

LARESE, MÓNICA G.; GÓMEZ, JUAN C.

Rosario, Argentina

Otro; XIII Reunión de Trabajo en Procesamiento de la Información y Control (RPIC 2009); 2009

Laboratorio de Sistemas Dinámicos y Procesamiento de la Información, FCEIA, UNR

  <!-- @page { size: 8.5in 11in; margin: 0.79in } P { margin-bottom: 0.08in } -In spotted microarray analysis, image segmentation is a fundamental task which allows to extract the spot areas in order to measure the gene hybridization ratios. In the last years, many segmentation procedures were proposed to segment microarrays, including traditional techniques from the image processing field. Accuracy of the segmentation stage is crucial in order to obtain reliable hybridization ratios. The purpose of this paper is to evaluate the performance of a new segmentation scheme based on Gaussian Mixture Models clustering in combination with Markov Random Fields (MRFs), and quantitatively compare it with some of the existing segmentation techniques. As ground truth is unknown when it comes to performance evaluation in microarray image segmentation, a state of the art database consisting of simulated microarray images with realistic characteristics is used. Two quantitative measures are computed to assess the accuracy, namely the probability of error (PE) and the discrepancy distance (DD), allowing to compare the different strategies under consideration.