Segmentation of pathological and diseased lung tissue in CT images using a graph-search algorithm

Abstract

Lung segmentation is an important first step for quantitative lung CT image analysis and computer aided diagnosis. However, accurate and automated lung CT image segmentation may be made difficult by the presence of the abnormalities. Since many lung diseases change tissue density, resulting in intensity changes in the CT image data, intensity-only segmentation algorithms will not work for most pathological lung cases. This paper presents an automatic algorithm for pathological lung CT image segmentation that uses a graph search driven by a cost function combining the intensity, gradient, boundary smoothness, and the rib information. This method was trained by four pathological lung CT images and tested on fifteen 3-D thorax CT data sets with lung diseases. We validate our method by comparing our automatic segmentation result with manually traced segmentation result. Sensitivity, specificity, and Hausdorff distance were calculated to evaluate the method.