Dissertation
CT-based automated measures of lung volume-defined mechanical biomarkers of ribs and their relations with COPD-related respiratory impairment
University of Iowa
Doctor of Philosophy (PhD), University of Iowa
Autumn 2025
Abstract
Chronic obstructive pulmonary disease (COPD) is a progressive respiratory condition characterized by airflow limitation, structural remodeling of the lung, and impaired thoracic mechanics. While lung parenchymal and airway abnormalities have been extensively studied, the rib cage - an essential structure for maintaining breathing mechanics - has received limited attention. There is growing evidence that rib motion and deformation play a critical role in respiratory function and may serve as sensitive biomarkers of COPD-related impairment. Accurate and reproducible measures of rib deformation from medical imaging are of paramount importance for advancing phenotypic characterization and early detection of disease progression in COPD. The primary aim of my PhD research is to develop and validate a novel computational framework for quantifying breathing-related rib deformation (rib Δ-metrics) from paired inspiratory and expiratory CT scans using automated image processing techniques. The first aim of my PhD research is to develop automated computational algorithms to extract rib Δ-metrics. The second aim of my PhD research is to systematically evaluate the performance of rib Δ-metrics. The third aim of my PhD research is to investigate the biological and clinical relevance of rib Δ-metrics using large-scale human datasets.
The proposed framework integrates deep learning–based rib segmentation, automated rib labeling, centerline computation used skeleton-based shortest path algorithms and breathe-related deformation quantification. Three rib "Δ" -metrics were derived: (1) axial areal expansion (∆_(rib"-" A)), reflecting thoracic expansion; (2) craniocaudal (∆_(rib"-" CC)), representing superior–inferior rib motion; and (3) anteroposterior (∆_(rib"-" AP)), capturing anterior–posterior displacement. The segmentation model, based on a three-dimensional (3-D) U-Net architecture, achieved a Dice similarity coefficient exceeding 0.95 and demonstrated robust reproducibility. We also analyzed the correlations of respiratory "Δ" -metrics (rib Δ-metrics combined with airway Δ-metrics and diaphragm Δ-metrics) with demographic, smoking, COPD, the associations of respiratory Δ-metrics with lung health, and physical activity parameters, respiratory Δ-metrics impairments in COPD severity groups and the early respiratory mechanical signals of dyspnea.
This thesis presents a new CT-based computational framework for automated calculation of rib Δ-metrics between inspiratory (total lung capacity) and expiratory (residual volume) phases and analyzes the correlations of respiratory Δ-metrics. The key contributions of this thesis include: (1) development of a deep learning-based multi-phase rib segmentation algorithm; (2) implementation of automated rib labeling and skeleton-based centerline computation; (3) definition and validation of three rib Δ-metrics quantifying thoracic deformation; and (4) statistical characterization of respiratory Δ-metrics clinical and biological relevance in COPD.
Details
- Title: Subtitle
- CT-based automated measures of lung volume-defined mechanical biomarkers of ribs and their relations with COPD-related respiratory impairment
- Creators
- Yan Liu
- Contributors
- Pranav K. Saha (Advisor)Alejandro Freymond Comellas (Committee Member)Kung-Sik Chan (Committee Member)Xiaodong Wu (Committee Member)Syed Ahmed Nadeem (Committee Member)
- Resource Type
- Dissertation
- Degree Awarded
- Doctor of Philosophy (PhD), University of Iowa
- Degree in
- Electrical and Computer Engineering
- Date degree season
- Autumn 2025
- Publisher
- University of Iowa
- Number of pages
- xvi, 102 pages
- Copyright
- Copyright 2025 Yan Liu
- Language
- English
- Date submitted
- 12/08/2025
- Description illustrations
- illustrations (some color)
- Description bibliographic
- Includes bibliographical references (page 92-102).
- Public Abstract (ETD)
Chronic obstructive pulmonary disease (COPD) is a major cause of breathing difficulty and long-term disability. It affects millions of people and is often caused by smoking or long-term exposure to harmful air. One key aspect of how we breathe is the motion of the ribs, which expand and contract along with our lungs. However, rib movement in people with COPD has not been studied in detail, even though it plays an important role in respiratory function.
The first aim of my PhD research is to develop and test new methods that can automatically measure how the ribs move between breathing in and breathing out. These measurements are called rib deformation (rib ?-metrics), and they are computed from CT scans of the chest taken at different lung volumes. The second aim of my research is to carefully evaluate the accuracy and reliability of these automated rib measurements. The third aim of my PhD is to study how these rib movement measurements are related to a person s lung health, symptoms, smoking history, and the severity of their COPD.
The proposed framework integrates deep learning based rib segmentation, automated rib labeling, centerline computation used skeleton-based shortest path algorithms and breathe-related deformation quantification. The segmentation model, based on a U-Net architecture, achieved a Dice similarity coefficient exceeding 0.95 and demonstrated robust reproducibility. We also analyzed the correlations of respiratory ?-metrics (rib ?-metrics combined with airway ?-metrics and diaphragm ?-metrics) with demographic, smoking, COPD, the associations of respiratory ?-metrics with lung health, and physical activity parameters, respiratory ?-metrics impairments in COPD severity groups and the early respiratory mechanical signals of dyspnea.
- Academic Unit
- Electrical and Computer Engineering
- Record Identifier
- 9985134848802771
Metrics
1 Record Views