Neural networks approach for soft tissue modelling

Fast, realistic physically based deformable models are required for real time Virtual Reality applications such as surgical simulators. The main focus of this paper is to introduce a Neural Network based deformable model called NNDM that is fast enough to allow interactive frame rates and produces a plausible deformation. The main idea is to introduce the neural networks to approximate the deformation and the transmission of the force from the manipulated vertex to its adjacent neighbors and from those neighbors to their neighbors and so on, also the deformation can be simulated locally or globally. The model is tested on virtual objects those modeled as surface mesh and integrated in an interactive Cataract Eye Surgery Simulator to model the deformation of eye tissues when it is manipulated with surgical instrument.