Deep learning parametric response mapping from inspiratory chest CT scans: a new approach for small airway disease screening

This study proposes a deep learning model to evaluate Small Airway Disease (SAD) using only a single inspiratory CT scan, unlike conventional Parametric Response Mapping (PRM) which requires both inspiratory and expiratory CT scans. The research team used a Generative Adversarial Network (GAN) to generate an expiratory CT from an inspiratory CT, and then used this with the real inspiratory CT to predict PRM. The model was evaluated on 537 participants with normal spirometry, and the results showed that the predicted PRM had a high correlation with the ground truth PRM (emphysema r=0.97, fSAD r=0.64) and effectively stratified SAD (AUC 0.78-0.84). The reliability of the ground truth PRM used in the study was validated by comparison with the commercial software Aview (Coreline Soft). This technology has the potential to enable large-scale early screening for SAD by reducing radiation exposure and costs.