University of British Columbia
Towards improving prostate cancer diagnosis and treatment with shear wave absolute vibro-elastography and automatic low-dose-rate prostate brachytherapy planning
Abstract
dc:descriptionProstate cancer (PCa) represents a substantial health concern for men, and this thesis focuses on improving its diagnosis and treatment. The first contribution introduces a registration framework that localizes transperineal template-guided mapping biopsy (TTMB) cores in volumetric ultrasound (US) using only standard TTMB information. With an average target registration error of 1.2 mm and ≈97 s of registration time, this framework provides fast and accurate TTMB core localization. The pathology of these registered cores can serve as ground truth for training US-based PCa classifiers. To improve diagnosis, three US systems are developed to perform Shear Wave Absolute Vibro-Elastography (S-WAVE), a technique for quantitative tissue stiffness assessment. The first system, compatible with TTMB, employs multi-frequency transperineal excitation for the first time and correlates with magnetic resonance elastography (MRE) imaging at 96%. Clinical studies revealed a positive correlation between its measurements and PCa detected from histopathology. A PCa classifier, trained using clinical S-WAVE data taken with this system, achieved an AUC of 0.87±0.12. The second system is the first hand-operated 3D S-WAVE system with a transducer-mounted exciter designed for targeting systematic biopsies. When compared to MRE, it achieves a cross-correlation of 99% and 94% on quality assurance phantoms and in vivo human livers, respectively. In the third system, both S-WAVE and strain elastography are implemented for the first time in a commercial microUS system thus enabling the possibility of multi-parametric microUS imaging. For localized PCa, low-dose-rate prostate brachytherapy (LDR-PB) is an effective treatment where small radioactive seeds are permanently implanted within the prostate. The arrangement of these seeds is pre-planned manually, which is an iterative and time-consuming process. To automate this, two new methods based on conditional generative adversarial networks are proposed. Comparable results to the high-quality manual plans were achieved, demonstrating 98.9% coverage of the clinical target volume (CTV) receiving the minimum prescribed dose. Planning times are significantly reduced, with the proposed methods running approximately 7 and 400 times faster.
Degree
thesis:*- Name thesis:degree_name
- Doctor of Philosophy - PhD
- Level thesis:degree_level
- doctoral
- Discipline thesis:degree_discipline
- Biomedical Engineering
- Grantor dc:publisher
- University of British Columbia
- Year dc:date
- 2023
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Aleef, Tajwar Abrar
- Contributors dc:contributor
-
- Salcudean, Septimiu Edmund
- Mahdavi, S. Sara
Rights
dc:rights- Statement dc:rights
-
- Attribution-NonCommercial-ShareAlike 4.0 International
- Language dc:language
- eng
Identifiers
dc:identifier.*- Handle dc:identifier
- http://hdl.handle.net/2429/86967