Centerline Extraction Method for Virtual Vascular Model in Virtual Reality Interventional Training Systems

Virtual reality (VR) interventional training systems are commonly used for vascular interventional surgery training. Compared with traditional training method, including using human cadavers, live animals and vascular phantom, VR interventional training has many advantages such as low training cost and variable training model. For virtual interventional radiology, simulating catheter interaction is a challenging work. Centerline of the vasculature is often used to detect the contact between blood vessel wall and surgical tools. In this paper, we proposed an improved centerline extraction method based on generalized rotational symmetry axis. The method discretizes the vasculature by a set of continuous cylindrical shapes. This discretization obtains an effective strategy for vasculature centerline extraction. In order to improve the algorithm efficiency, we use a pre-processing strategy to merge duplicate points and normal vector for vasculature mesh. This strategy turns the vasculature mesh into vasculature point cloud and reduced the number of calculation points. The performance of our method is experimentally validated.