Lung diseases, including chronic obstructive pulmonary disease (COPD), asthma, infections, lung cancer, and tuberculosis, continue to be major contributors to global mortality and healthcare burden. Among these, emphysema—a primary component of COPD affects approximately 200 million individuals worldwide and has a high rate of underdiagnosis in the screening population. The COVID-19 pandemic has further strained global respiratory health, particularly among patients with pre-existing lung conditions, emphasizing the need for enhanced diagnostic techniques.
This thesis of Yeshaswini Nagaraj investigates AI-driven solutions for detecting and analyzing emphysema and COVID-19 using machine learning (ML), radiomics, and deep learning (DL). Focused primarily on computer tomography (CT) imaging, it highlights AI's potential to improve diagnostic accuracy, automate tasks, and support clinical decision-making. The research is organized into distinct sections, with initial chapters addressing emphysema detection and quantification through an AI prototype designed for low-dose CT scans, demonstrating robust performance in early disease detection.
Subsequent chapters shift to AI applications in COVID-19 diagnosis, emphasizing structured reporting systems and multi-modal models that can detect emphysema while triaging COVID-19 cases. Radiomics has also been explored for its ability to quantify emphysema severity through high-dimensional feature extraction. Overall, the research highlights the importance of AI in advancing lung disease detection and management, offering valuable tools for enhanced patient outcomes and future research directions in pulmonary disease analysis.
