Optical and digital methods for the analysis of microscopic tissue sections

This thesis presents optical and computerbased methods for the analysis of microscope specimens that are in the form of thin sections that have been mounted on standard glass slides. Three methods are discussed whereby image and photometric information can be obtained - television scanning, microphotometric scanning and micro-spectrophotometry. The limitations of these methods are analysed and some representative results are given. The thresholding of digital image data with the assistance of neighbourhood operators is examined, with particular reference to the Laplacian and Gradient operators. A method for the statistical analysis of tissue-section textures is presented and is shown to be applicable the discrimination of two pathological classes of human prostate tissue. The effectiveness of the method is evaluated by the use of a computed Mahalanobis distance for the various statistics used.