Parameter Optimisation of a Linear Tetrahedral Mass Tensor Model for a Maxillofacial Soft Tissue Simulator

Abstract

We present an extensive validation of the prediction accuracy of our soft tissue simulator for maxillofacial surgery planning using a linear Tetrahedral Mass Tensor Model (MTM). Prediction accuracy is quantified by measuring distances between the predicted data and the actual post-operative CT data for a database containing 10 patients who underwent maxillofacial surgery. Two different setups are considered. First two important parameters of a homogeneous MTM are optimised, namely the material’s Poisson Ratio ν and the number of tetrahedra contained by the mesh. Optimal results were achieved with ν≈0.46 and N _tetra ≥30.000. Moreover the average simulation time could be reduced to less than 2.5 seconds. In the second setup an inhomogeneous MTM that differentiate between biomechanical properties for fat and muscle tissue is introduced. Simulation results show to be independent of Young’s Moduli and optimal results were achieved for ν _fat = 0.485 and ν _muscle = 0.43. Moreover it turned out that using such an inhomogeneous model doesn’t improve simulation accuracy significantly when compared to the homogeneous model.