This thesis presents a Colon Modeler that constructs colon models from patient CT data. The constructed colon models are used in the colonoscopy simulator. The Colon Modeler, first, carries out segmentation of the colon boundary from the CT data, extracts the centerline of the colon, and constructs the surface of the colon model. The colon models are generated and managed by sections of the colon
The region growing method is extended to segment the 3D colon boundary, and the distance map is applied to extract the centerline. The centerline-based parametric representation is used to visualize the 3D colon model. The cross sections of the colon surface are constructed by casting rays along the centerline of the colon.
An intersection-elimination method is designed to interpolate the colon in the sharp curvatures of the colon. The parametric representation method presented in the thesis can significantly save the computational time without seriously degrading the graphic fidelity of the model.
The developed Colon Modeler is integral in providing various training scenarios for the colonoscopy simulator.