On modelling of anisotropic viscoelasticity for soft tissue simulation: numerical solution and GPU execution

Area of Science:

• Computational mechanics
• Biomedical engineering
• Medical image computing

Background:

• Soft tissue simulation is crucial for medical applications like surgical simulation and image registration.
• Existing models often simplify tissue mechanics, neglecting anisotropy and time-dependence for efficiency.
• There's a need for accurate yet computationally feasible soft tissue deformation simulation techniques.

Purpose of the Study:

• To develop a generalized anisotropic viscoelastic constitutive framework for soft tissues.
• To create an efficient solution procedure for viscoelastic hereditary integrals in explicit dynamic finite element analysis.
• To implement and evaluate a GPU-based finite element scheme for accelerated soft tissue simulation.

Main Methods:

• Developed a generalized anisotropic viscoelastic constitutive framework.
• Created an efficient solution procedure for viscoelastic hereditary integrals.
• Implemented a GPU-based finite element scheme using the CUDA API.

Main Results:

• The proposed method incorporates anisotropy and viscoelasticity with only a 5.1% increase in computational cost compared to isotropic elastic models.
• The GPU implementation achieved a 56.3x speedup over CPU execution.
• Numerical examples validated the accuracy of the viscoelastic solution procedure and the performance of the GPU implementation.

Conclusions:

• The developed anisotropic viscoelastic model and efficient solution procedure enable accurate soft tissue simulation.
• The GPU implementation significantly accelerates the simulation process, making complex models computationally viable.
• This work advances the field of medical image computing by providing efficient and accurate tools for soft tissue analysis.