Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.790891
Title: Investigating tissue heterogeneity using MRI in prostate cancer
Author: Devine, William
ISNI:       0000 0004 8499 9041
Awarding Body: UCL (University College London)
Current Institution: University College London (University of London)
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Abstract:
Multi-parametric MRI, a promising new technique for grading prostate cancer using MRI, classifies a high number of regions as indeterminate. This is a symptom of the wider problem that clinical usage of MRI in prostate cancer only includes basic techniques and does not directly categorise tissue microstructure. This work provides insight into the microstructure of the prostate using a combination of new tissue models and acquisition schemes. Each is tested with the aim of producing a method that is better at detecting and grading prostate cancer. The first section utilises microstructural diffusion models to better quantify tissue heterogeneity in the prostate. The two models investigated provided more information about the heterogeneous nature of the prostate that ADC and showed significant difference between lesions and normal tissue. The next section looks into combining multi-echo T2 (ME-T2) sequences with quantitative tissue modelling called Luminal Water Imaging (LWI). This work produced an optimal LWI fitting technique and acquisition. Then the ability of LWI to detect the PI-RADS v2.0 score of regions of interest was examined, showing that it was able to differentiate between scores better than ADC. This work also showed that LWI can differentiate between tumour and normal tissue with an AUC of 0.81 (p < 0.05) when compared to ADC with an AUC of 0.75 (p < 0.05) in this dataset. The next section further improves the acquisitions using larger datasets. It showed that correcting for imperfect pulse refocusing could improve on the performance of LWI in detecting PCa. This work also showed that fewer echoes could be used in the acquisition. Neural networks were then used to detect and grade prostate cancer using the data points from both multiple b-value diffusion and ME-T2 decay curves. The neural network's ability to distinguish between different PIRADS scores was shown to have an AUC of 0.87 (p < 0.05) using 32-echo data.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.790891  DOI: Not available
Share: