4D reconstruction of airway dynamics using endoscopic optical coherence tomography.

Endoscopic optical coherence tomography (OCT) offers non-invasive, high-resolution cross-sectional imaging of the respiratory tract, supporting three-dimensional (3D) airway reconstruction for clinical diagnosis and evaluation of airway diseases. However, conventional 3D reconstruction methods typically overlook respiration-induced motion, thereby limiting their ability to capture the dynamic anatomical changes required for four-dimensional (4D) visualization. We propose a respiration-resolved 4D airway reconstruction method using endoscopic OCT. Repeated OCT scans are acquired at each spatial location to capture structural variations across the respiratory cycle. The collected data are processed through an automatic reconstruction framework that integrates a deep learning-based lumen segmentation algorithm for accurate airway boundary extraction, a respiration correction algorithm that identifies and aligns B-scans corresponding to consistent respiratory phases using lumen contour variation, and a rotation-correction algorithm that aligns cross-sectional frames based on lumen contour similarity. In vivo imaging of rabbit airways demonstrates that the proposed method successfully reconstructs respiration-resolved 4D airway structures from endoscopic OCT, underscoring its potential for dynamic airway imaging.