Lung segmentation for CT images based on mean shift and region growing

Abstract

Segmentation of the lungs in chest-computed tomography (CT) is a precursor to most pulmonary image analysis applications. A new lung segmentation based on the 3D CT image series is proposed integrating mean shift smoothing and region growing algorithms together. As medical images are mostly fuzzy, Mean Shift cluster algorithm is used to smooth the CT images. Then some seed points for left and right lung separately are selected by the user, and the growing criterion is calculated automatically by the analyzing the neighboring sub-blocks. Then region growing method is applied to get the final segmentation. Experiments results show the proposed method can efficiently segment the lung region from serial abdominal CT images with little user interaction. © Springer Science+Business Media Dordrecht 2014.