Assessment of microvascular volume in smokers using AI - PROJECT SUMMARY Chronic obstructive pulmonary disease (COPD) and emphysema are globally significant chronic lower respiratory diseases that represent a major cause of morbidity and mortality worldwide. COPD is diagnosed via spirometry, which detects airflow obstruction that is not completely reversible, while emphysema is defined on pathology as the permanent enlargement and destruction of alveolar walls and can also be measured on computed tomography (CT). Despite advancements, pharmacologic therapies for COPD lack efficacy in altering disease progression or improve mortality. Recent CT imaging studies have highlighted the significance of pulmonary vascular endothelial damage in the pathogenesis of these diseases. Evidence suggests that the pulmonary vasculature may be significantly compromised in patients with COPD and emphysema, indicating a fundamental role of pulmonary vascular injury in disease etiology and progression. However, despite advancement in macro-vascular imaging, understanding of the microvasculature’s role in COPD progression remains limited. While new imaging modalities have emerged for detecting microvasculature perfusion, their availability is often limited in large-scale epidemiological studies, hindering accurate investigations into disease etiology and progression. To address this gap, we developed an AI-based algorithm for robust assessment of vascular perfusion from single energy non-contrast CT scans, which enables the definition of novel image-based markers of pulmonary microvasculature alterations in large epidemiological studies. The proposed research, carried out via a secondary analysis of subjects in the COPDGene cohort, aims to develop and validate a quantitative marker of pulmonary microvascular injury and to explore the correlation between microvascular volume loss and the longitudinal progression of COPD and emphysema in smokers with and without COPD. In Aim 1, we will extend the validation of our AI-based method for generating perfusion maps from single energy non-contrast CT scans. Our preliminary analysis on a non-COPD specific dataset demonstrates the effectiveness and reliability of our approach. As part of Aim 1, we will further validate this technique through a comparative analysis between our synthesized perfusion maps and dual energy CT perfusion imaging. This analysis will be conducted using subjects diagnosed with COPD, both with and without PH, for which single energy non-contrast CT images, DECT scans, and perfusion maps are available. In Aim 2, we will estimate perfusion information from subjects in the COPDGene Phase 2 study to define a quantitative marker of microvascular volume. This marker will serve as a tool to investigate the relationship between pulmonary vascular damage and clinical outcomes, COPD, and emphysema development. Through these efforts, we aim to create an image-based quantifiable feature of pulmonary vascular injury, facilitating new research studies on the etiology and evolution of COPD in smokers.
