A segmentation-based automatic navigation technique for virtual endoscopy

Virtual endoscopy (VE) mimics the examination consisting of introducing an endoscope into a tubular structure for diagnostic purposes. One of the main problems in VE is the computation of the navigation path that the virtual camera will travel while producing the animation that will be viewed by the specialist. We propose a technique that allows the automatic computation of navigation paths connecting a starting and one or more end points inside the structure to be visualized. The computation of the path uses the connectedness map produced by the segmentation algorithm that has been used to identify the structure to be visualized in the image. A simple smoothing operation is then applied to this path to produce the final navigation path.