Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.617006
Title: Tissue type analysis of brain tumours using multimodal MRI
Author: Raschke , Felix
ISNI:       0000 0004 5348 3720
Awarding Body: St George's, University of London
Current Institution: St George's, University of London
Date of Award: 2014
Availability of Full Text:
Full text unavailable from EThOS. Please contact the current institution’s library for further details.
Abstract:
The aim of this thesis is to investigate the spatial extent of brain tumours by developing new methods combining the metabolic information provided by low resolution 1 H magnetic resonance spectroscopic imaging (MRSI) data with high resolution structural information from diffusion tensor imaging (DTI). In chapter 3 and chapter 4, the spectral analysis tool LCModel is used to decompose single voxel (SV) MRS data of common adult and childhood brain tumours into the most likely tissue class respectively. Classification according to the highest estim,ated tissue proportion suggested comparable performance to published specialised pattern recognition methods and emphasises the flexibility of the method. In chapter 5, the LCModel tissue type analysis is refined and extended to decompose short echo MRSI data of glioma patients into normal and abnormal tissue proportions. Spatial assessment revealed metabolic low grade characteristics around most grade IV glioblastomas as potential tumour infiltration. Several visualisation techniques explored reveal heterogeneous infiltration patterns varying across patients. In chapter 6 a radial choline-to-N-acetyl-aspartate index (rCNI) is presented as an alternative method for the delineation of brain tumour MRSI exams using the choline to NAA ratio. Both simulations and analysis of real 1.5T and 3T glioma data suggest a higher specificity at similar sensitivity for rCNI over the previously published CN!. The final chapter 7 presents a novel method for the combination of the LCModel tissue type information from chapter 5 with DTI and conventional MRI data. The tissue type information is used to sample high confidence regions of tumour and normal brain and extract the underlying DTI and MRI information and create tissue specific probability density distributions. Additional distributions from areas of tumour infiltration and vasogenic oedema were sampled from glioma and metastasis data respectively. Bayes' theorem was then used to calculate probability maps of the different tissue types which show promise for tumour grading and the differentiation of vasogenic oedema and tumour infiltration.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.617006  DOI: Not available
Share: