In this paper, a technique is presented for CT image analysis and visualization of the bronchial airways. The technique provides a non-invasive way to examine the interior of the bronchial tubes and to detect various properties of the tubes such as abnormal morphology caused by foreign objects stuck in the airways or some other disease. The input to the procedure are chest images obtained by spiral computed tomography CT. 3-D neural network-based segmentation of CT images is performed to extract the airways. The resulting 3-D binary volume representing the location of the airways is thinned to extract the medial axis of the bronchial tubes. The marching cube algorithm is used to perform a triangulation of the airway surface. The extracted axis and the triangulated surface is converted to Virtual Reality Modeling Language VRML format. The virtual environment in VRML format is then used for inspection and visualization of the bronchial tube interior.
Sven Loncaric, T. Markovinovic, T. Petrovic, D. Ra