Date of Award
Master of Applied Science
Faculty of Science and Technology
Associate Professor James Cross
Dr Inta Adams
In this thesis the expression digital morphometry collectively describes all those procedures used to obtain quantitative measurements of objects within a two-dimensional digital image. Quantitative measurement is a two-step process: the application of geometrical transformations to extract the features of interest, and then the actual measurement of these features. With regard to the first step the morphological filters of mathematical morphology provide a wealth of suitable geometric transfomations. Traditional radiometric and spatial enhancement techniques provide an additional source of transformations. The second step is more classical (e.g. Underwood, 1970; Bookstein, 1978; and Weibull, 1980); yet here again mathematical morphology is applicable - morphologically derived feature parameters. This thesis focuses on mathematical morphology for digital morphometry. In particular it proffers a taxonomy of morphological filters and investigates the morphologically derived feature parameters (Minkowski functionals) for digital images sampled on a square grid. "As originally conceived by Georges Matheron, mathematical morphology concerns the analysis of binary images by means of probing with structuring elements [typically convex geometric shapes]" (Dougherty, 1993, preface). Since its inception the theory has been extended to grey-level images and most recently to complete lattices. It is within the very general framework of the complete lattice that the taxonomy of morphological filters is presented. Examples are provided to help illustrate the behaviour of each type of filter. This thesis also introduces DIMPAL (Mehnert, 1994) - a PC-based image processing and analysis language suitable for researching and developing algorithms for a wide range of image processing applications. Though DIMPAL was used to produce the majority of the images in this thesis it was principally written to provide an environment in which to investigate the application of mathematical morphology to Alzheimer's disease research. Alzheimer's disease is a form of progressive dementia associated with the degeneration of the brain. It "is the commonest type of dementia and probably accounts for half the dementia of old age" (Forsythe, 1990, p. 21 ). Post mortem examination of the brain reveals the presence of characteristic neuropathologic lesions; namely neuritic plaques and neurofibrillary tangles. They occur predominantly in the cerebral cortex and hippocampus. Quantitative studies of the distribution of plaques and tangles in normally aged and Alzheimer brains are hampered by the enormous amount of time and effort required to count and measure these lesions. Here in a morphological algorithm is proposed for the automatic segmentation and measurement of neuritic plaques from light micrographs of post mortem brain tissue.
Mehnert, A. (1994). Digital Morphometry : A Taxonomy Of Morphological Filters And Feature Parameters With Application To Alzheimer's Disease Research. Retrieved from https://ro.ecu.edu.au/theses/1468