Detection of pulmonary vessels in 3D lung CT using improved Graph Cut

Abstract

The problem of pulmonary vessel detection from 3D pulmonary CT Scan is a very challenging one. The identification of vessels is important for clinical evaluation. In this paper, we proposed a vessel segmentation technique based on improved graph cut algorithm by designing the energy function. First of all, the enhanced image is modeled with adaptive k-means algorithm to give the regional parameter of the energy function. Then the improved energy function is given to graph cut algorithm for vessel segmentation. Graph cut algorithm creates a graph which is cut using minimum cut theory. The segmentation is done with the data provided in VESSEL12 site. This automatic segmentation gives quite satisfactory results.