Currently the medical training mainly based on observation rather than practice because learning done directly from the patient poses ethical problems or medical-legal. Or dexterity required in the use of surgical tools requires practice. The use of training systems based on the simulation can then be an answer to this training problem. The idea is then to provide through a numerical simulation information (such as the position of the organs in which they interact, or the forces exerted upon them) to physical devices on which the apprentice found sensations and situations perceived in reality and a visual rendering of the physiological mechanisms generated during the gesture to improve their understanding by the trainees.

The challenge is to provide realistic simulations of organs in interactive time. That is to say, we do not seek absolute accuracy of the biomechanical behavior of soft tissue, but from a theoretical knowledge we try to simplify the simulations to reduce their computation time while maintaining a relevant global behavior integration into a learning simulator of medical procedures. 

To reduce the computation time of simulations, we tackled over the years the problem from different aspects. We have worked on the proposal of new physical models to manage mixed objects, that is to say broken down into elements geometries and different behavior laws; the contribution of a complex topological model to optimize the topological changes undergone by the subject during his simulation; on defining interactions held within the object that model the physical behavior including the setting up of a surface model to gain complexity over a solid model; and the parallelization of processing performed during the simulation to still gain a little more time calculations. The underlying idea is then to play on these aspects during the simulation by using criteria established according to the intended application.