Efficient Metaballs-Based Collision Detection for VR Neurosurgery Simulation on GPU

This paper presents a novel hybrid model comprising both surface mesh and the metaballs which occupy organs’ interior for the soft tissue modeling. Through the utility of metaballs, we are capable of simplifying the organ interior using a set of overlapping spheres with different radii. We first develop an adaptive approach based on Voronoi Diagram for the initialization of inner metaballs. Then, we resort to global optimization and devise an electrostatic attraction technique to drive the metaballs to best fill the space inside the organ’s boundary. We simplify the surgical instrument as a collection of cylinders with different radii and orientation, and develop an adaptive collision detection method to facilitate the collision between the surgical instrument and metaballs. Our framework is built on the parallel computation architecture of CUDA, and thus can afford interactive performance on a commodity desktop. To illustrate the effectiveness, the above techniques have all been integrated into a VR-based ventriculoscopic surgery simulator.