A 3D segmentation method of lung parenchyma based on CT image sequences

Three dimensional (3D) pulmonary parenchyma segmentation is an indispensable step in CAD (Computer-Aided Detection) that used for pulmonary nodule detection based on CT images. In this paper, we proposed a simple and effective 3D lung parenchyma segmentation method, which combined the adaptive threshold, connected regional labeling and morphological operations. The method has four main steps. Firstly, the CT sequences images were binarized. Secondly, the lung parenchyma is extracted from the CT images by 3D connected component labeling. And then, the trachea is removed by 3D region growing. Finally, a sequence of morphological operations is used to smooth the boundaries and fill the holes caused by small vessels, nodules and trachea/bronchus. Using our method to segment 20 group lung CT clinical data, the average segmentation accuracy is 91.55%, and the average time-consuming to deal with a single group data is 167.4563s(about 0.6 second for processing a single slice). The experimental results showed that this method can automatically and quickly segments the lung parenchyma, and which formed the basis for the follow-up pulmonary nodules computer-aided detection technology study.