Use this URL to cite or link to this record in EThOS:
Title: Genetic algorithms for medical image analysis
Author: Delibasis, Konstantinos K.
Awarding Body: University of Aberdeen
Current Institution: University of Aberdeen
Date of Award: 1995
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
This work attempts the formulation of a number of computer vision problems, often emerging when processing medical images, as optimisation problems. The ability of Genetic Algorithms, a global optimisation technique, to efficiently and reliably perform the required optimisations is assessed. Results are quantified and compared to other more established methods. The problem of anatomical object detection and extraction of their shape from 3D medical images is considered first. Two geometric primitives with parametrically controllable shape are used for geometric modelling and an adequate objective function is introduced to be optimised over the shape parameters. GAs are then employed to optimise the objective function. Quantified assessment of the results using two different anatomical objects is produced and comparisons to interactive object segmentation are made. The problem of texture based segmentation is also considered. The detection of texture is formalised as a problem of designing a mask that exploits relationships between the spectra of different classes of texture. Results are produced in the case of artificial patterns, natural texture and texture present in medical images, including modalities like MRI and X-rays. The results of the segmentation are compared to other better established texture discrimination techniques. Finally, the problem of noise suppression is formulated as a problem of stack filter configuration, a broad family of non-linear filters. Results are produced for different types of noise, including additive uncorrelated noise as well as multiplicative or gaussian and poisson noise. Results from the application of the configured filter are compared to those of other digital filters, commonly used for noise reduction.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available