Robust Airway Generation Labelling with Airway Segmentation for Reliable Airway Assessment

This study focuses on the accurate classification of airways within the human respiratory system, which is characterized by a complex, tree-like structure with diverse patterns. While recent advancements in deep learning show potential for automatic airway segmentation, further research is necessary to make these methods practical for assessing a patient’s airway status and enabling personalized treatment. A thorough investigation into airway generation is essential for improving diagnostics and therapies, as it aids in understanding specific structures and identifying issues such as airway narrowing and lung compliance. Current techniques, including template matching and machine learning approaches that treat airway classification as a problem, have limitations in capturing the complexities of higher-generation airways. To address these challenges, our study introduces a novel methodology. The Prim algorithm is employed to establish a minimum spanning tree from the centerline, creating an accurate tree structure that reflects airway connections. A new secondary branch is generated when a main branch with an outdegree of 1 is identified, ensuring precise labeling of airway generation .