The following is a summary of “Longitudinal Assessment of interstitial lung abnormalities on CT in Patients with COPD Using Artificial Intelligence-based Segmentation: A Prospective Observational Study,” published in the April 2024 issue of Pulmonology by Shiraishi et al.
Interstitial lung abnormalities (ILAs) detected via computed tomography (CT) imaging have been implicated in influencing clinical outcomes among individuals with chronic obstructive pulmonary disease (COPD). Yet, a standardized method for their quantification still needs to be discovered. This prospective observational study explored the feasibility of employing artificial intelligence (AI)-based segmentation to identify ILAs across two distinct COPD cohorts accurately.
Utilizing the Fleischner Society definition as a visual diagnostic criterion, ILAs were identified and characterized in patients with COPD . The AI-based segmentation method facilitated the precise delineation of ground-glass opacities, reticulations, and honeycombing, with their respective volumes aggregated to yield the percentage ratio of interstitial lung disease-associated volume to total lung volume (ILDvol%). Through comprehensive analysis of cross-sectional data from discovery and validation cohorts, the study established an optimal ILDvol% threshold for ILA detection, demonstrating robust sensitivity and specificity upon validation. Furthermore, a longitudinal assessment of ILDvol% revealed a significant association with ILA development over five years, even after adjusting for pertinent covariates such as age, sex, body mass index (BMI), and smoking history.
The findings underscore the potential utility of AI-based CT quantification in facilitating the identification and longitudinal monitoring of ILAs in patients with COPD. By offering a reproducible and objective methodology for ILA assessment, this approach holds promise for enhancing diagnostic precision and prognostic insights in managing COPD-associated interstitial lung abnormalities.
Source: bmcpulmmed.biomedcentral.com/articles/10.1186/s12890-024-03002-z
