Online Multiresolution Volumetric Mass Spring Model for Real Time Soft Tissue Deformation Celine Paloc 1 , Fernando Bello 2 , Richard I. Kitney 1 , and Ara Darzi 2 1 Dept. of Bioengineering, Imperial College, London, UK 2 Dept. of Surgical Oncology and Technology, Imperial College, London, UK c.paloc@ic.ac.uk Abstract. Recent years have seen an increase in the acceptance and demand for Virtual Reality surgical simulators. Although significant ad- vances have been made in the area, real-time accurate simulation of soft tissue deformation is still a major obstacle when developing simulators with haptic feedback. On this paper we present a new multi-resolution volumetric mass-spring model that offers high visual and haptic resolu- tion in and around the region of interaction and other critical regions. Visual and haptic resolution decreases in proportion to the distance from such regions making it possible to distribute the computational workload optimally in order to achieve real-time haptic simulation. 1 Introduction Surgical simulation is an extremely challenging area of research combining med- ical imagery, computer graphics and mathematical modelling. Recent advances make it possible to represent complex tissue structures and perform virtual fly- through operations, but a great deal of research in soft tissue modelling is still needed to develop the next generation of surgical simulators. 1.1 Physically-Based Deformable Models Various approaches founded on the laws governing the dynamics of non-rigid bodies have been proposed for simulating deformable soft tissue. The Finite Element Method (FEM) is a common and accurate way to com- pute complex deformations of soft tissue, but conventional FEM has high com- putational cost and large storage requirements. Hybrid models based on global parameterized deformations and local deformations based on FEM have been introduced [1,2,3] to tackle this problem. Large-scale multi-processor computers to obtain soft tissue deformation at interactive rates have also been employed [4], while in [5,6] pre-computed elementary deformations and speed-up algorithms were used. Most of these methods, however, are only applicable to linear de- formations and valid for small displacements. Furthermore, they tend to rely on pre-computing the complete matrix system and are therefore unable to cope with topological changes occurred during cutting or tearing. Mass Spring Systems (MSS) have been widely used in soft tissue simula- tion [7,8,9,10] because of their ability to generate dynamic behaviors that allow real time deformation and topological changes. The main limitation of MSS is T. Dohi and R. Kikinis (Eds.): MICCAI 2002, LNCS 2489, pp. 219–226, 2002. c  Springer-Verlag Berlin Heidelberg 2002