This paper gives integrated methods that perform virtual reality based planning and training of minimally invasive surgery. The authors propose an advanced framework of volumetric soft tissue cutting, which consistently combines topological changes into deformable models. Collision detection and physical constraint are provided in simulated surgery space. Time series MRI dataset are applied to a visual and haptic system with a force feedback device. The reconstructed environment enables to identify approachable region, and to discuss optimum strategy. Figures and evaluations assure that the developed methods are effective in preoperative planning as well as in learning surgeries.