Optimizing time step size in modeling liver deformation

摘要

Mass Spring Model (MSM) is often used to model human liver due to its easy implementation and low computational complexity. This paper focuses on development of a real-time human liver simulation, which enabled an efficient implementation of liver mechanical response incorporating nonlinearity and viscoelasticity properties. Optimization of the computational problems is necessary to permit real-time liver simulation. Adam Variable Step-Size Predictor-Corrector (AVSPC) method is preferred to solve the governing differential equations and considerably more accurate than Fourth-order Runge-Kutta (RK4) method. AVSPC method with higher accuracy leads to higher quality of liver simulation. Reduction of local truncation error is needed to maintain accuracy as well as preventing model state from rapid change. Optimized time step size 0.0063 was implemented in CHAI 3D to simulate real-time liver deformation caused by surgical indenter. Deformation of liver with higher deformation rate appears to have higher stiffness and higher stress relaxation rate. In conclusion, this model is a plausibly significant liver tissue model which is more suitable for real-time interaction with lower computational cost, more accurate, realistic and acceptable to be used in the near future.