In this thesis we present a method for the segmentation of nuclei in hematoxylin and eosin stained pathology slide images, as well as all the relevant mathematical background of the techniques used. Also, we present three methods for automated region on interest detection and compare each method's ability to identifying regions of high nuclear density. By applying both region of interest detection and nuclear segmentation, we attempt to identify any features which could help differentiate between cases of prostate adenocarcinoma, ranging from moderate to more severe. It was found that simple morphological features (such as nuclear size and shape) were insufficient metrics with which to differentiate tumors with similar Gleason scores.
