Quantification and Analysis of Lung Involvement by Artificial Intelligence in Patients with Progressive Pulmonary Fibrosis Treated with Nintedanib

Background and Objectives: Progressive pulmonary fibrosis (PPF) presents significant clinical challenges due to irreversible lung damage and declining respiratory function. Nintedanib has demonstrated antifibrotic effects, yet there is a lack of sensitive tools to assess treatment efficacy quantitatively. This study evaluated the potential of artificial intelligence (AI)-powered quantitative computed tomography (QCT) to monitor lung changes and predict treatment outcomes in patients with PPF undergoing nintedanib therapy. Materials and Methods: This retrospective study analysed 37 patients diagnosed with PPF who were treated with nintedanib for one year. AI-powered QCT was performed using the 3D Slicer software version 5.2.2, which quantified lung infiltration, collapse, and vessel volumes. These data were then correlated with pulmonary function tests. Receiver operating characteristic (ROC) analysis was used to assess baseline AI-powered QCT predictors for progression. Results: AI-powered QCT demonstrated a significant reduction in post-treatment right lung infiltration (5.56 ± 3.08 cm3 to 4.88 ± 2.77 cm3, p = 0.041), whereas total lung infiltration decreased non-significantly. Functional parameters, including forced vital capacity (FVC) and diffusion capacity for carbon monoxide (DLCO), showed no significant changes. ROC analysis identified a baseline infiltrated lung volume greater than 21.90% as predictive of continued disease progression (AUC = 0.767; sensitivity, 91.70%; specificity, 68.00%). Conclusions: AI-powered QCT identified diverse parenchymal responses to nintedanib in PPF and showed preliminary prognostic value for clinical trajectory. Imaging biomarkers enhance functional measures and may reveal early treatment effects. Prospective, multicentre validation is necessary to confirm usefulness and establish actionable thresholds for clinical application.