An Analytical Review of the Numerical Methods used for Finite Element Modeling of Traumatic Brain Injury.

Dozens of finite element models of the human brain have been developed for providing insight into the mechanical response of the brain during impact. Many models used in traumatic brain injury research are based on different computational techniques and approaches. In this study, a comprehensive review of the numerical methods implemented in 16 brain models was performed. Differences in element type, mesh size, element formulation, hourglass control, and solver were found. A parametric study using the SIMon FE brain model was performed to quantify the sensitivity of model outputs to differences in numerical implementation. Model outputs investigated in this study included nodal displacement (commonly used for validation) and maximum principal strain (commonly used for injury assessment), and these results were demonstrated using the loading characteristics of a reconstructed football concussion event. Order-of-magnitude differences in brain response were found when only changing the characteristics of the numerical method. Mesh type and mesh size had the largest effect on model response. These differences have important implications on the interpretation of results among different models simulating the same impacts, and of the results between model and in vitro experiments. Additionally, future studies need to better report the numerical methods used in the models.