Magnetic resonance imaging for glioma molecular classification

Thust, Stefanie Catherine (2024) Magnetic resonance imaging for glioma molecular classification. Doctoral thesis, University of East Anglia.

Preview

PDF Download (49MB) | Preview

Abstract

This doctoral thesis discusses a programme of magnetic resonance imaging (MRI) research for predicting molecular characteristics in glioma, a malignant brain tumour, carried out over 8 years.

In 2016, a new classification system mandated the integration of genetics into glioma tissue diagnosis. My research is centred around identifying imaging biomarkers of group specific tumour mutations.

Chapter 1 summarises a pan European survey to investigate conventional and advanced glioma imaging practices in clinical application. Based on responses from 220 hospital institutions, anatomical and diffusion-weighted imaging were judged to be essential, subsequently formulated as consensus guidance. A second survey on advanced MRI practices corroborated the findings.

Chapter 2 reports on research to identify anatomical MRI features associated with glioma molecular characteristics. Candidate morphologies were shaped by testing recently published visual criteria in a pilot study, and by a literature search. Some features, including diffusion, showed limited reproducibility by qualitative inspection.

Chapter 3 discusses apparent diffusion coefficient (ADC) measurements to distinguish early aggressive cancer stages from slower growing gliomas. The accuracy of ADC results differed by tumour contrast enhancement pattern. The performance and interobserver agreement were tested for different ADC parameters.

Chapter 4 describes research into stratifying visual features by reproducibility and combining these with ADC values and age to achieve a prediction of glioma genetic status. An accurate logistic regression model was established and validated in a new glioma MRI data set.

Chapter 5 details the investigation of histogram methods, including a new software (TexRAD), for genotyping of gliomas. No superior diagnostic yield was identified from histogram analysis compared to regional diffusion measurements. Using TexRAD software did not outperform logistic regression in single MRI sequence assessment.

Chapter 6 discusses ongoing research developments and relates the thesis to recently published data on glioma imaging.

Item Type:	Thesis (Doctoral)
Uncontrolled Keywords:	Publication
Faculty \ School:	Faculty of Medicine and Health Sciences > Norwich Medical School
Depositing User:	Chris White
Date Deposited:	07 May 2025 13:58
Last Modified:	08 May 2025 07:44
URI:	https://ueaeprints.uea.ac.uk/id/eprint/99192
DOI:

Downloads

Actions (login required)

View Item