Extraction and classification of pulmonary organs based on thoracic 3D CT images.

Recent progress in helical CT technology has given rise to considerable interest and expectations regarding diagnosis based on 3D CT images. In this context, computer-aided diagnosis techniques for 3D CT images are required. This study introduces a method for analysis of thoracic 3D CT images, which contain important information for lung cancer diagnosis. The lung field includes the bronchi and pulmonary arteries and veins, as well as other organs; lung cancer influences these organs in a complicated and intermixed way. The proposed method supports extraction and classification of bronchi, arteries, and veins from thoracic 3D CT images using anatomical knowledge and 3D image processing technologies, and analysis of the relations between these pulmonary organs and suspected lung cancer. The extraction of pulmonary organs includes removal of bias components, extraction of bronchi and blood vessels, discrimination between bronchi and blood vessels, and identification of arteries and veins among blood vessels. Quantitative relations between suspected lung cancer and adjacent pulmonary organs are found. The effectiveness of the proposed method is shown using 3D CT images of healthy and cancerous lungs. © 2001 Scripta Technica, Syst Comp Jpn, 32(9): 42–53, 2001 [ABSTRACT FROM AUTHOR]