Precise knowledge of bronchovascular anatomy is critical for lung surgery. In this video tutorial, we present a manual, semi-automatic segmentation workflow using open-source image processing software to create detailed three-dimensional models of the lungs from computed tomography scans. Structures segmented include the bronchus, lung lobes, pulmonary arteries and veins, and tumours. In this method, we have used various segmentation techniques including thresholding, painting, logical operations and smoothing to perform anatomical labelling. By following this workflow, segmentation times range from 1.5 to 2 hours per patient, which can be affected by both computed tomography image quality and contrast enhancement. Although manual segmentation is time-consuming, it offers a cost-effective alternative to commercial three-dimensional reconstruction software, particularly for educational and research applications.
Mank et al. (Mon,) studied this question.