Loughborough University
Browse
- No file added yet -

Medical image processing: applications in ophthalmology and total hip replacement

Download (7.17 MB)
thesis
posted on 2013-03-04, 16:29 authored by Nesreen Otoum
Medical imaging tools technologically supported by the recent advances in the areas of computer vision can provide systems that aid medical professionals to carry out their expert diagnostics and investigations more effectively and efficiently. Two medical application domains that can benefit by such tools are ophthalmology and Total Hip Replacement (THR). Although a literature review conducted within the research context of this thesis revealed a number of existing solutions these are either very much limited by their application scope, robustness or scope of the extensiveness of the functionality made available. Therefore this thesis focuses on initially investigating a number of requirements defined by leading experts in the respective specialisms and providing practical solutions, well supported by the theoretical advances of computer vision and pattern recognition. This thesis provides three novel algorithms/systems for use within image analysis in the areas of Ophthalmology and THR. The first approach uses Contourlet Transform to analyse and quantify corneal neovascularization. Experimental results are provided to prove that the proposed approach provides improved robustness in the presence of noise, non-uniform illumination and reflections, common problems that exist in captured corneal images. The second approach uses a colour based segmentation approach to segment, measure and analyse corneal ulcers using the HVS colour space. Literature review conducted within the research context of this thesis revealed that there is no such system available for analysis and measurement of corneal ulcers. Finally the thesis provides a robust approach towards detecting and analysing possible dislocations and misalignments in THR X-ray images. The algorithm uses localised histogram equalisation to enhance the quality of X-ray images first prior to using Hough Transforms and filtered back projections to locate and recognise key points of the THR x-ray images. These key points are then used to measure the possible presence of dislocations and misalignments. The thesis further highlights possible extensions and improvements to the proposed algorithms and systems.

History

School

  • Science

Department

  • Computer Science

Publisher

© Nesreen Otoum

Publication date

2013

Notes

A Doctoral Thesis. Submitted in partial fulfilment of the requirements for the award of Doctor of Philosophy of Loughborough University.

EThOS Persistent ID

uk.bl.ethos.572549

Language

  • en