Computational brain models are a useful tool to assess the dynamic strain response of the brain during head impact. Population-average models are typically created, but their predictions depend on many factors (e.g. geometry, material properties) that differ between the model and an individual. As much experimental data as possible should be included when evaluating a model to minimize uncertainties in calibration and validation. The objective of this study is to create individualized computational brain models using magnetic resonance imaging (MRI) techniques with subject-specific anatomy, material properties, and evaluation by comparison to subject-specific brain deformation.
A subject-specific 3D brain model was created from segmented anatomical scans of a 31y, male volunteer. Voxel-level storage and loss moduli were estimated from magnetic resonance elastography (MRE) scans of the same subject at frequencies of 30, 50, and 70 Hz. Tagged MRI experiments were also conducted in the volunteer under mild rotation at 2-3 rad/s, and used to compute 3D displacement and strain fields. All data were co-registered and interpolated to the same resolution (1.5 mm) as the model. Brain materials were modeled as linear viscoelastic (LVE) with two time constants. Material parameters were estimated using a global optimization scheme that utilized the average of MRE moduli within a segmented label, as well as literature material testing data outside of the MRE frequency range. Two types of heterogeneity were investigated by 1) separating the brain into 10 regions (anatomy heterogeneity), and 2) by separating the brain into 10 regions by stratifying MRE stiffness (mechanical heterogeneity). All models were simulated for 100 ms using the material point method in UINTAH with the tagged MRI loading conditions. Initial results show improved agreement between the model response and tagged MRI experiments by incorporating subject-specific LVE properties and mechanical heterogeneity.