Loukas, Constantinos; (2002) Quantitative image analysis of biological factors influencing radiotherapy. Doctoral thesis (Ph.D), UCL (University College London).

Text Quantitative_image_analysis_of.pdf Download (24MB)

Abstract

Image analysis is a rapidly evolving field with growing applications in science and engineering. In cancer research, it has played a key role towards the solution of problems with major diagnostic importance, minimising human intervention and providing vital clinical information. This thesis describes research which has attempted to develop, and make improvements to, image analysis methods in order to assess key biological features in tissue sections of various types of human tumours. The analysed histological material was stained for a variety of markers likely to be involved in the cellular response to radiation treatment. The principal area of work was also in connection with the extraction of substantial information that could be used to select the most appropriate treatment schedule. Task-oriented edge detection, watershed-like, and principal components analysis algorithms were developed for the segmentation of cancer cell-nuclei (both positively stained and non) in large-scale histological images. Probabilistic grey-level clustering and neural network-based methods were employed for detecting blood vessels in single-stained tissue specimens. Colour segmentation techniques based on pattern recognition and fuzzy clustering, combined with texture analysis, were designed for detecting simultaneously hypoxia and blood vessels in double-stained histological material. The algorithms' performance was validated on tissue section images encountered in routine clinical practice. The most successful of the methods studied were also integrated into some specific problem-designed image analysis tools, for facilitating the process of feature extraction.

Download activity - last month

Export as XMLCSVJSON

Download activity - last 12 months

Export as XMLCSVJSON

Downloads by country - last 12 months

Export as CSVXMLJSON

Archive Staff Only

View Item