Q16 - Incorporation of Fiber Tractography for Finite Element Modeling of Mouthguard-Derived Concussion Kinematics

Saturday, October 11, 2025

10:00 AM - 11:00 AM PDT

Location: Exhibit Hall F, G & H

Primary Investigator(s)

Theo Chiang

Researcher
Stanford University
Palo Alto, California, United States

Presenting Author(s)

Jessica Towns, MS

PhD Candidate
Stanford University
Stanford, California, United States

Co-Author(s)

Jessica Towns, MS

PhD Candidate
Stanford University
Stanford, California, United States

Last Author(s)

David Camarillo

Associate Professor
Stanford University, United States

Co-Author(s)

MZ

Michael Zeineh

Associate Professor
Stanford University, United States

Introduction: : Mild traumatic brain injury (mTBI), or concussion, is the most common form of traumatic brain injury, accounting for 80-90% of an estimated 69 million annual cases worldwide. [1,2] Head impact biomechanics, specifically linear and rotational kinematics as well as resulting brain strain, are traditionally used to estimate concussion risk. Rotational forces in particular are found to drive shear deformation of axons, which is associated with concussion. [3,4]

Instrumented mouthguards (iMGs), equipped with six-degree-of-freedom kinematic sensors, have become the state-of-the-art technology for facilitating direct measurement of head impact kinematics. IMGs supersede previous helmet systems which directly measure linear accelerations but only estimate rotational accelerations. [4] They also offer higher accuracy due to their rigid coupling to the upper dentition, which minimizes relative motion between the sensors and the skull. [5] High-fidelity kinematics captured by iMGs can be used as input for finite element (FE) head models to compute tissue-scale metrics such as axonal strain, a measure shown to improve concussion prediction. [6]

Although prior work has reported kinematic and strain metrics for concussive impacts, few studies have leveraged directly measured rotational kinematics from iMGs across multiple sports, competitive levels, and both sexes to quantify axonal strain. [6] Thus, the aim of this abstract is to identify the 95th percentile maximum axonal strains for concussive impacts across a broad population of athletes. To achieve this, we embedded white‑matter fiber tractography, representative of individual axons, into an established FE head model and simulated head impacts using iMG‑captured concussive and non-concussive impacts.

Materials and

Methods: : We fitted fibers from the 1 mm HCP-1065 template fiber map to the Global Human Body Models Consortium (GHBMC) 50th percentile male (M50) V6 brain model. [7,8] To register from ICBM152 to GHBMC space, we performed surface-based alignment following methods described in Wu et al., [6] using iterative closest point (ICP) for affine adjustments and thin plate spline (TPS) for nonlinear warping.

The GHBMC pia, which precisely conforms to the cortex, represented the FE model surface, while a wavefront object of the ICBM152 diffusion-weighted image template served as the surface reference for the fiber tractography during registration. Because ICP cannot handle scaling or reflections, we empirically determined an initial transform for rough alignment, including a 0.821 scale factor which minimized subsequent ICP median error (3.875 mm). After applying TPS (target registration error = 0.720 mm), we replicated all transformations on the fiber tractography, yielding 3,450 registered fibers after downsampling (Figure 1).

Using LS-DYNA, we simulated 23 concussive and 40 non-concussive impacts from the Stanford iMG dataset [9] (Table 1), computing global 95th percentile maximum principal and axonal strains (MPS95, MAS95; see Figure 2C). Due to computational cost, we analyzed a subset of cases. Axonal material properties were assigned *MAT_NULL to avoid volume redundancy, [10] while GHBMC properties were unaltered. To calculate axonal strain, centroids of individual beam elements were matched to corresponding solid element centroids using nearest-neighbor. Strain tensors of the solid elements were then projected along directions of corresponding beam elements to determine axonal strain.

Results, Conclusions, and Discussions:: MPS95 and MAS95 averaged 0.117 ± 0.075 and 0.033 ± 0.018 for non-concussive impacts, and 0.319 ± 0.125 and 0.086 ± 0.033 for concussive impacts, respectively (Table 1). A Wilcoxon rank-sum test yielded Bonferroni-corrected p-values of 1.307×10⁻⁷ and 9.492×10⁻⁸ for MPS95 and MAS95 respectively, indicating that MAS95 could exhibit greater sensitivity for distinguishing injury status. Within the subcortical white matter, higher axonal strains appear along the lateral temporal lobes, the superior-posterior areas of the frontal lobes, and the midbrain, mirroring MPS95 patterns (Figure 2A-B).

Average MPS95 values reported here closely align with previous studies, while average MAS95 values vary across both our results and prior work. Wu et al. reported 50th percentile injury probability thresholds of 0.271 for MPS95 and 0.127 for MAS95 in humans. [11] Moreover, Zhou et al. reported axonal strains averaging ~0.076 for non-concussive, and ~0.185 for concussive impacts, differing from Wu et al. and our findings. [10,11] These discrepancies may reflect unique modeling approaches and datasets; Wu et al. incorporated explicit axon material properties, while Zhou et al. dynamically updated fiber orientations along projected axons within the ADAPT head model. Also, while we used high school and collegiate impact data, Wu et al. evaluated concussive data from the National Football League, and Zhou et al. was limited to two concussive cases, one involving loss of consciousness.

Our prior work has shown MPS95 to be a limited predictor of mTBI, [9] warranting investigation of axonal strain—a potentially more anatomically relevant metric. While existing axon-based FEMs are promising, further work is needed to refine MAS95 mTBI thresholds across a broader dataset. In future iterations, we plan to apply the axon-adapted FEM model to our full dataset (n = 3,852), the largest, most diverse iMG dataset to our knowledge, [9] to assess the predictive ability of MAS95. We will also utilize the Holzapfel-Gasser-Ogden model to assign anisotropic material properties to fibers based on local fractional anisotropy, allowing them to contribute stiffness to the system. Such advancements may enable future axon-based FEMs to elucidate tissue-based mechanisms of mTBI, guiding future concussion risk modeling and headgear design.

Acknowledgements and/or References (Optional): : Acknowledgements:
This work was supported by the Taube Stanford Children’s Concussion Initiative, Pac-12 Conference’s Student-Athlete Health and Well-Being Initiative, the National Institutes of Health (R24NS098518), the Maternal and Child Health Research Institute, the Office of Naval Research, and the National Science Foundation Graduate Research Fellowship Program. This work was also supported by Stanford’s Department of Bioengineering.

References:
1. Blennow, K. et al. Traumatic brain injuries. Nat. Rev. Dis. Primer 2, (2016).
2. Dewan, M. C. et al. Estimating the global incidence of traumatic brain injury. J. Neurosurg. 130, 1080–1097 (2019).
3. Liu, Y. et al. Theoretical and numerical analysis for angular acceleration being determinant of brain strain in mTBI. Preprint at https://doi.org/10.48550/ARXIV.2012.13507 (2020).
4. Beckwith, J. G. et al. Estimated Brain Tissue Response Following Impacts Associated With and Without Diagnosed Concussion. Ann. Biomed. Eng. 46, 819–830 (2018).
5. Liu, Y. et al. Validation and Comparison of Instrumented Mouthguards for Measuring Head Kinematics and Assessing Brain Deformation in Football Impacts. Ann. Biomed. Eng. 48, 2580–2598 (2020).
6. Wu, T., Alshareef, A., Giudice, J. S. & Panzer, M. B. Explicit Modeling of White Matter Axonal Fiber Tracts in a Finite Element Brain Model. Ann. Biomed. Eng. 47, 1908–1922 (2019).
7. Yeh, F.-C. Population-based tract-to-region connectome of the human brain and its hierarchical topology. Nat. Commun. 13, (2022).
8. Mao, H. et al. Development of a Finite Element Human Head Model Partially Validated With Thirty Five Experimental Cases. J. Biomech. Eng. 135, (2013).
9. Towns, J. A. et al. Linear Acceleration Is a Primary Risk Factor for Concussion. Preprint at https://doi.org/10.48550/ARXIV.2507.09098 (2025).
10. Zhou, Z. et al. White Matter Tract-Oriented Deformation Is Dependent on Real-Time Axonal Fiber Orientation. J. Neurotrauma 38, 1730–1745 (2021).
11. Wu, T. et al. Evaluation of Tissue-Level Brain Injury Metrics Using Species-Specific Simulations. J. Neurotrauma 38, 1879–1888 (2021).