Automatic Extraction of 3D Airway Tree from Multislice Computed Tomography Images

Due to the low contrast value between airway walls and lung tissue on CT image, providing an accurate and in-vivo segmentation algorithm of airway tree from MSCT chest scan is challenging in medical imaging. In this study, an automated approach to segment airway tree was proposed. First, the trachea and major parts of bronchus tree were segmented by Fast Marching Methods. The surface diffusion algorithm was then used to segment the ending of the bronchus tree using the individually seed surfaces. This diffusion algorithm was continued as a feedback loop until no new meaningful seed. Finally, the extraction airway data was refined by the morphology filter. The proposed method was testified to provide much better and higher order airway tree than the growing method.