A hybrid lung segmentation algorithm based on histogram-based fuzzy C-means clustering

The lung is an essential organ and is dark in Computed Tomography (CT) images because of air. Lung segmentation and correct lung region separation is a prerequisite for the development of computer-aided diagnostic algorithms and disease treatment planning. However, this remains a nontrivial problem because of lung anatomical structures. Here, we addressed this problem and proposed a reliable and robust solution that is based on a histogram-based fuzzy C-means (FCM) algorithm and morphological mathematical algorithms. There were 1632 high resolution CT slices with 1 mm thickness used from asthma patients with low dose; right and left lungs were classified using the proposed algorithm. We extracted right lung regions with 96.05% accuracy and left lung regions at 96.32%. The computation time is 1.3 s per slice.