Poster P18 - Simulating the Deformation Response of the Aging Brain Using 3D Subject-Specific Finite Element Models

Introduction: Traumatic brain injury (TBI) is a major health concern with an estimated 1.7 million cases occurring annually in the United States. The risk of TBI and its severity depend on multiple factors, one of which is the patient's age. Epidemiological studies have shown that more than 60% of TBI related hospitalizations and deaths affect adults older than 65 years. The underlying mechanisms that contribute to the age-dependent differences in TBI risk and prognosis are not well understood. Natural changes in the aging brain such as cerebral atrophy and ventricle enlargement may affect the brain’s biomechanical response. Therefore, understanding the biomechanics of the brain under injurious conditions is important for predicting TBI risk. The aim of this study is to investigate the age-dependent biomechanical response of the brain to injurious impacts in older adults using 3D subject-specific finite element (FE) models. Due to increased TBI risk in older adults, we hypothesize that the simulated brain deformation will increase with age. To achieve this aim, we use FE models from a longitudinal magnetic resonance imaging (MRI) dataset of healthy aging in older adults. The aging brain’s deformation (or strain) response to an injurious head impact was modeled for each subject at all longitudinal scan sessions and the magnitude and distribution of brain strain were quantified.

Materials and

Methods: Subjects were selected from the Baltimore Longitudinal Study of Aging (BLSA), a healthy aging dataset of older adults, if they completed at least three MRI scanning sessions (196 males and 238 females) over several years. Selection of the top 10 male and 10 female subjects for simulation was based on the largest change in normalized ventricle volume (ratio of ventricle volume to total brain volume) with respect to age change from initial to final sessions (Fig 1.B). Normalized ventricle volume was chosen as a simple measure of cerebral atrophy, invariant to baseline brain volume. MRI scans were segmented with the deep learning SLANT CRUISE algorithm to label various brain regions. Using a custom MATLAB code, 3D subject-specific FE models were generated with a voxelized mesh of 1.5 mm spatial resolution (Fig 1.A). Previously, our team employed this technique to develop subject-specific models from MRI data, evaluating the biofidelity of strain responses against in vivo brain deformation. Simulations were conducted in LS-Dyna using an axial head rotation loading condition with peak angular velocity of 40 rad/s and peak angular acceleration of 3600 rad/s2. For each subject's longitudinal scans, the 95th percentile maximum principal strain (MPS), representing the maximum deformation each voxel experiences in any direction, was calculated and compared to change in structural atrophy markers, like ventricle volume change, to quantify the strain trends and distribution. The difference in 95th percentile MPS over the entire brain were calculated for each scan relative to the baseline scan.

Results, Conclusions, and Discussions: 20 subjects were simulated across all longitudinal scans, with each subject undergoing between 3 to 6 scanning sessions, for a total of 79 simulations. The strain trends were consistent, with all subjects showing decreases in the 50th and 95th percentile MPS, as well as volume fraction of brain experiencing larger than 0.25 MPS, with respect to change in age. The simulation results indicate that the aging brain experiences 0.8% and 1.7% decrease in 95th percentile MPS per year for male and female subjects, respectively, relative to strain magnitudes at baseline age for each subject (Fig 1.C). The simulated results show consistent trends with R-squared values of 0.32 and 0.76 for male and female subjects, respectively (Fig 1.C). This suggests that the structural changes that the aging brain experiences lower the amount of brain deformation under injurious impacts.

While the correlation coefficient from our results is moderate, there are other subject-specific factors such as baseline age, sex, head size, and amount of atrophy that could contribute and will be investigated. More longitudinal aging subjects, with differing levels of atrophy, will also be simulated to provide a larger sample size. A regional and voxel-wise analysis of the strain trends and distribution will be conducted to better understand spatial differences in brain deformation with age. This study provides valuable insights into the biomechanical response of the aging brain to TBI. It is important to note that while these results suggest a decrease in brain deformation with age-related cerebral atrophy, they are only based on the biomechanical response. Other factors associated with aging, including secondary cascades and recovery time, cannot be captured with the FE models. The findings, however, highlight the importance of considering age and cerebral atrophy as critical factors in the modeling and prediction of TBI risk. The subject-specific FE models developed in this study paired with biomechanical imaging could serve as a useful tool for predicting the individualized response of the patient’s brain to TBI, thereby aiding in the development of personalized predictive, mitigative, and preventive strategies.

Acknowledgements (Optional):