REDUCING THE BURDEN OF CANCER WITH ARTIFICIAL INTELLIGENCE: IMAGE-BASED MODELS FOR BREAST CANCER DETECTION, ADVANCED STAGE RISK, AND CROSS-MODALITY VISUALIZATION
Date
2023
Authors
Contributor
Advisor
Department
Instructor
Depositor
Speaker
Researcher
Consultant
Interviewer
Narrator
Transcriber
Annotator
Journal Title
Journal ISSN
Volume Title
Publisher
Volume
Number/Issue
Starting Page
Ending Page
Alternative Title
Abstract
Medical imaging provides a non-invasive method for obtaining valuable information used to detect cancer and assess cancer risk. Constant improvements to imaging technologies have resulted in more available information as well as new types of information that can be tied to biomarkers of cancer. Our central hypothesis is that only a portion of the imaging information related to breast cancer risk and detection is currently being utilized due to the limitations of deriving and testing preconceived imaging biomarkers of simple constructs. The objective of this dissertation is to develop artificial intelligence (AI) and machine learning (ML) models to fully interrogate medical images for additional information related to cancer detection and risk. In this work, we also develop methods to address limited labeled training data, which is a common problem in ML. Our specific aims and outcomes are as follows: 1) Identify compositional imaging biomarkers related to breast cancer detection; 2) Investigate imaging features associated with risk of advanced-stage breast cancer; 3) Implement pretraining methods to improve modeling with limited labeled medical images for body composition. Novel contributions to the field resulting from each aim are as follows: 1) Malignant lesions were confirmed to have unique compositional signatures which can be used to improve detection specificity; 2) Mammographic images were found to contain advance stage cancer risk information beyond just breast density; 3) Self-supervised pretraining with domain-specific data helped us overcome limitations stemming from the size of our labeled dataset. Both a more quantitatively accurate image generation model and a more accurate dual-energy X-ray absorptiometry image analysis model resulted from this study aim.
Description
Keywords
Medicine, Artificial intelligence, Medical imaging, 3D Body Scan, Breast Cancer, Deep Learning, DXA, Imaging Biomarkers, Self-Supervised Learning
Citation
Extent
124 pages
Format
Geographic Location
Time Period
Related To
Related To (URI)
Table of Contents
Rights
All UHM dissertations and theses are protected by copyright. They may be viewed from this source for any purpose, but reproduction or distribution in any format is prohibited without written permission from the copyright owner.
Rights Holder
Local Contexts
Email libraryada-l@lists.hawaii.edu if you need this content in ADA-compliant format.