Use this URL to cite or link to this record in EThOS:
Title: Combining global and local information for the segmentation of MR images of the brain
Author: Hutton, Chloe Anne
ISNI:       0000 0001 3585 5066
Awarding Body: University of London
Current Institution: University College London (University of London)
Date of Award: 1999
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Magnetic resonance imaging can provide high resolution volumetric images of the brain with exceptional soft tissue contrast. These factors allow the complex structure of the brain to be clearly visualised. This has lead to the development of quantitative methods to analyse neuroanatomical structures. In turn, this has promoted the use of computational methods to automate and improve these techniques. This thesis investigates methods to accurately segment MRI images of the brain. The use of global and local image information is considered, where global information includes image intensity distributions, means and variances and local information is based on the relationship between spatially neighbouring voxels. Methods are explored that aim to improve the classification and segmentation of MR images of the brain by combining these elements. Some common artefacts exist in MR brain images that can be seriously detrimental to image analysis methods. Methods to correct for these artifacts are assessed by exploring their effect, first with some well established classification methods and then with methods that combine global information with local information in the form of a Markov random field model. Another characteristic of MR images is the partial volume effect that occurs where signals from different tissues become mixed over the finite volume of a voxel. This effect is demonstrated and quantified using a simulation. Analysis methods that address these issues are tested on simulated and real MR images. They are also applied to study the structure of the temporal lobes in a group of patients with temporal lobe epilepsy. The results emphasise the benefits and limitations of applying these methods to a problem of this nature. The work in this thesis demonstrates the advantages of using global and local information together in the segmentation of MR brain images and proposes a generalised framework that allows this information to be combined in a flexible way.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: Biophysics