International Conference "From Scientific Computing to Computational Engineering"
(EN)
The objective of this work was to investigate the segmentation capability of Gaussian Mixture Models
(GMMs) on microarray images, using wavelet based image enhancement techniques. A Simulated Microarray
image of 200 spots was produced using a Microarray Scan Simulator. The image was pre-processed using 3
different daub4 wavelet based filters, developed in MATLAB. The detail coefficients of the wavelet transform
were processed up to scale 3 using Hard, Soft and Sigmoidal Thresholding. An automatic gridding process was
developed and applied on the preprocessed microarray image with the purpose of identifying the spots. The
GMMs algorithm was then applied to each spot in order to discriminate foreground from background. The
segmentation capability of the GMMs was evaluated by calculating the segmentation matching factor for each
spot. Optimal segmentation results were obtained by pre-processing the microarray image by all the waveletbased
filters. Wavelet based preprocessing was found to improve the GMMs segmentability, in discriminating
the spot’s foreground from background.
(EN)
