| A microarray is used to measure gene expression. This thesis applies maximum-likelihood estimation techniques to the fundamental measurement in microarray analysis—measuring the gene expression ratio from fluorescent microarray images. This is accomplished through the derivation of a statistical model of the intensity of spots in microarray images. The model includes spot pixel intensity distributions, inter-pixel correlations and the intensity distribution of background noise. From this statistical model, an estimator for gene expression ratio has been produced.; The estimator, a pair of polynomial equations solved simultaneously, is compared via simulation to other methods of measuring gene expression ratio. Performance analysis of this estimator indicates a five per-cent improvement in the standard deviation of the measures of gene expression ratio compared to traditional methods. |
