Lung Area Extraction from X-ray CT Images for Computer-aided Diagnosis of Pulmonary Nodules by using Active Contour Model

In this paper, we develop a lung area extraction technique from X-ray computed tomography (CT) images for computer-aided diagnosis (CAD) systems. In lung cancer cases, pulmonary nodules are typical pathological changes and thus they are the target to be detected by CAD systems. The isolated nodules can be detected more easily by CAD systems developed previously, while previous CAD systems are often hard to detect non-isolated nodules. The extraction technique can then be used for transforming the non-isolated pulmonary nodules connected to the walls of the chest into isolated ones. The technique proposed here is based on an active contour model, but such model is often trapped into a local optimum solution. To avoid the local optimum solutions, an essential core of the proposed technique is to select an appropriate initial contour by using an anatomical feature of the lung shape in X-ray CT slices. Some experimental results demonstrate the usefulness of the proposed technique for assisting the CAD systems to detect non-isolated nodules more accurately. Key–Words: Computer aided diagnosis, Active contour model, Pulmonary nodules, Anatomical feature, and X-ray CT images.