A network-based response feature matrix as a brain injury metric

A technique for in situ intracranial strain measurement within a helmeted deformable headform

Despite the broad use of helmets, incidence of concussion remains high. Current methods for helmet evaluation focus on the measurement of head kinematics as the primary tool for quantifying risk of brain injury. Though the primary cause of mild Traumatic Brain Injury (mTBI) is thought to be intracranial strain, helmet testing methodologies are not able to directly resolve these parameters. Computational injury models and impact severity measures are currently used to approximate intracranial strains from head kinematics and predict injury outcomes. Advancing new methodologies that enable experimental intracranial strain measurements in a physical model would be useful in the evaluation of helmet performance. This study presents a proof-of-concept head surrogate and novel helmet evaluation platform that allows for the measurement of intracranial strain using high-speed X-ray digital image correlation (XDIC). In the present work, the head surrogate was subjected to a series of bare and helmeted impacts using a pneumatically-driven linear impactor. Impacts were captured at 5,000 fps using a high-speed X-ray cineradiography system, and strain fields were computed using digital image correlation. This test platform, once validated, will open the door to using brain tissue-level measurements to evaluate helmet performance, providing a tool that can be translated to represent mTBI injury mechanisms, benefiting the helmet design processes.

An integrated finite element method and machine learning algorithm for brain morphology prediction

The human brain development experiences a complex evolving cortical folding from a smooth surface to a convoluted ensemble of folds. Computational modeling of brain development has played an essential role in better understanding the process of cortical folding, but still leaves many questions to be answered. A major challenge faced by computational models is how to create massive brain developmental simulations with affordable computational sources to complement neuroimaging data and provide reliable predictions for brain folding. In this study, we leveraged the power of machine learning in data augmentation and prediction to develop a machine-learning-based finite element surrogate model to expedite brain computational simulations, predict brain folding morphology, and explore the underlying folding mechanism. To do so, massive finite element method (FEM) mechanical models were run to simulate brain development using the predefined brain patch growth models with adjustable surface curvature. Then, a GAN-based machine learning model was trained and validated with these produced computational data to predict brain folding morphology given a predefined initial configuration. The results indicate that the machine learning models can predict the complex morphology of folding patterns, including 3-hinge gyral folds. The close agreement between the folding patterns observed in FEM results and those predicted by machine learning models validate the feasibility of the proposed approach, offering a promising avenue to predict the brain development with given fetal brain configurations.

Approximating subject-specific brain injury models via scaling based on head-brain morphological relationships

Most human head/brain models represent a generic adult male head/brain. They may suffer in accuracy when investigating traumatic brain injury (TBI) on a subject-specific basis. Subject-specific models can be developed from neuroimages; however, neuroimages are not typically available in practice. In this study, we establish simple and elegant regression models between brain outer surface morphology and head dimensions measured from neuroimages along with age and sex information (N = 191; 141 males and 50 females with age ranging 14-25 years). The regression models are then used to approximate subject-specific brain models by scaling a generic counterpart, without using neuroimages. Model geometrical accuracy is assessed using adjusted [Formula: see text] and absolute percentage error (e.g., 0.720 and 3.09 ± 2.38%, respectively, for brain volume when incorporating tragion-to-top). For a subset of 11 subjects (from smallest to largest in brain volume), impact-induced brain strains are compared with those from "morphed models" derived from neuroimage-based mesh warping. We find that regional peak strains from the scaled subject-specific models are comparable to those of the morphed counterparts but could be considerably different from those of the generic model (e.g., linear regression slope of 1.01-1.03 for gray and white matter regions versus 1.16-1.19, or up to ~ 20% overestimation for the smallest brain studied). These results highlight the importance of incorporating brain morphological variations in impact simulation and demonstrate the feasibility of approximating subject-specific brain models without neuroimages using age, sex, and easily measurable head dimensions. The scaled models may improve subject specificity for future TBI investigations.

A Review of Head Injury Metrics Used in Automotive Safety and Sports Protective Equipment

Despite advances in the understanding of human tolerances to brain injury, injury metrics used in automotive safety and protective equipment standards have changed little since they were first implemented nearly a half-century ago. Although numerous metrics have been proposed as improvements over the ones currently used, evaluating the predictive capability of these metrics is challenging. The purpose of this review is to summarize existing head injury metrics that have been proposed for both severe head injuries, such as skull fractures and traumatic brain injuries (TBI), and mild traumatic brain injuries (mTBI) including concussions. Metrics have been developed based on head kinematics or intracranial parameters such as brain tissue stress and strain. Kinematic metrics are either based on translational motion, rotational motion, or a combination of the two. Tissue-based metrics are based on finite element model simulations or in vitro experiments. This review concludes with a discussion of the limitations of current metrics and how improvements can be made in the future.

Use of Brain Biomechanical Models for Monitoring Impact Exposure in Contact Sports

Head acceleration measurement sensors are now widely deployed in the field to monitor head kinematic exposure in contact sports. The wealth of impact kinematics data provides valuable, yet challenging, opportunities to study the biomechanical basis of mild traumatic brain injury (mTBI) and subconcussive kinematic exposure. Head impact kinematics are translated into brain mechanical responses through physics-based computational simulations using validated brain models to study the mechanisms of injury. First, this article reviews representative legacy and contemporary brain biomechanical models primarily used for blunt impact simulation. Then, it summarizes perspectives regarding the development and validation of these models, and discusses how simulation results can be interpreted to facilitate injury risk assessment and head acceleration exposure monitoring in the context of contact sports. Recommendations and consensus statements are presented on the use of validated brain models in conjunction with kinematic sensor data to understand the biomechanics of mTBI and subconcussion. Mainly, there is general consensus that validated brain models have strong potential to improve injury prediction and interpretation of subconcussive kinematic exposure over global head kinematics alone. Nevertheless, a major roadblock to this capability is the lack of sufficient data encompassing different sports, sex, age and other factors. The authors recommend further integration of sensor data and simulations with modern data science techniques to generate large datasets of exposures and predicted brain responses along with associated clinical findings. These efforts are anticipated to help better understand the biomechanical basis of mTBI and improve the effectiveness in monitoring kinematic exposure in contact sports for risk and injury mitigation purposes.

Finding the Spatial Co-Variation of Brain Deformation With Principal Component Analysis

OBJECTIVE

Strain and strain rate are effective traumatic brain injury metrics. In finite element (FE) head model, thousands of elements were used to represent the spatial distribution of these metrics. Owing that these metrics are resulted from brain inertia, their spatial distribution can be represented in more concise pattern. Since head kinematic features and brain deformation vary largely across head impact types (Zhan et al., 2021), we applied principal component analysis (PCA) to find the spatial co-variation of injury metrics (maximum principal strain (MPS), MPS rate (MPSR) and MPS × MPSR) in four impact types: simulation, football, mixed martial arts and car crashes, and used the PCA to find patterns in these metrics and improve the machine learning head model (MLHM).

METHODS

We applied PCA to decompose the injury metrics for all impacts in each impact type, and investigate the spatial co-variation using the first principal component (PC1). Furthermore, we developed a MLHM to predict PC1 and then inverse-transform to predict for all brain elements. The accuracy, the model complexity and the size of training dataset of PCA-MLHM are compared with previous MLHM (Zhan et al., 2021).

RESULTS

PC1 explained variance on the datasets. Based on PC1 coefficients, the corpus callosum and midbrain exhibit high variance on all datasets. Finally, the PCA-MLHM reduced model parameters by 74% with a similar MPS estimation accuracy.

CONCLUSION

The brain injury metric in a dataset can be decomposed into mean components and PC1 with high explained variance.

SIGNIFICANCE

The spatial co-variation analysis enables better interpretation of the patterns in brain injury metrics. It also improves the efficiency of MLHM.

American Football Helmet Effectiveness Against a Strain-Based Concussion Mechanism

Brain strain is increasingly being used in helmet design and safety performance evaluation as it is generally considered as the primary mechanism of concussion. In this study, we investigate whether different helmet designs can meaningfully alter brain strains using two commonly used metrics, peak maximum principal strain (MPS) of the whole brain and cumulative strain damage measure (CSDM). A convolutional neural network (CNN) that instantly produces detailed brain strains is first tested for accuracy for helmeted head impacts. Based on N = 144 impacts in 12 impact conditions from three random and representative helmet models, we conclude that the CNN is sufficiently accurate for helmet testing applications, for elementwise MPS (success rate of 98.6%), whole-brain peak MPS and CSDM (coefficient of determination of 0.977 and 0.980, with root mean squared error of 0.015 and 0.029, respectively). We then apply the technique to 23 football helmet models (N = 1104 impacts) to reproduce elementwise MPS. Assuming a concussion would occur when peak MPS or CSDM exceeds a threshold, we sweep their thresholds across the value ranges to evaluate the number of predicted hypothetical concussions that different helmets sustain across the impact conditions. Relative to the 12 impact conditions tested, we find that the "best" and "worst" helmets differ by an average of 22.5% in terms of predicted concussions, ranging from 0 to 42% (the latter achieved at the threshold value of 0.28 for peak MPS and 0.4 for CSDM, respectively). Such a large variation among helmets in strain-based concussion predictions demonstrate that helmet designs can still be optimized in a clinically meaningful way. The robustness and accuracy of the CNN tool also suggest its potential for routine use for helmet design and safety performance evaluation in the future. The CNN is freely available online at https://github.com/Jilab-biomechanics/CNN-brain-strains .

Real-time dynamic simulation for highly accurate spatiotemporal brain deformation from impact

Real-time dynamic simulation remains a significant challenge for spatiotemporal data of high dimension and resolution. In this study, we establish a transformer neural network (TNN) originally developed for natural language processing and a separate convolutional neural network (CNN) to estimate five-dimensional (5D) spatiotemporal brain-skull relative displacement resulting from impact (isotropic spatial resolution of 4 mm with temporal resolution of 1 ms). Sequential training is applied to train (N = 5184 samples) the two neural networks for estimating the complete 5D displacement across a temporal duration of 60 ms. We find that TNN slightly but consistently outperforms CNN in accuracy for both displacement and the resulting voxel-wise four-dimensional (4D) maximum principal strain (e.g., root mean squared error (RMSE) of ~1.0% vs. ~1.6%, with coefficient of determination, R ² >0.99 vs. >0.98, respectively, and normalized RMSE (NRMSE) at peak displacement of 2%-3%, based on an independent testing dataset; N = 314). Their accuracies are similar for a range of real-world impacts drawn from various published sources (dummy, helmet, football, soccer, and car crash; average RMSE/NRMSE of ~0.3 mm/~4%-5% and average R ² of ~0.98 at peak displacement). Sequential training is effective for allowing instantaneous estimation of 5D displacement with high accuracy, although TNN poses a heavier computational burden in training. This work enables efficient characterization of the intrinsically dynamic brain strain in impact critical for downstream multiscale axonal injury model simulation. This is also the first application of TNN in biomechanics, which offers important insight into how real-time dynamic simulations can be achieved across diverse engineering fields.

An interdisciplinary computational model for predicting traumatic brain injury: Linking biomechanics and functional neural networks

The brain is a complex network consisting of neuron cell bodies in the gray matter and their axonal projections, forming the white matter tracts. These neurons are supported by an equally complex vascular network as well as glial cells. Traumatic brain injury (TBI) can lead to the disruption of the structural and functional brain networks due to disruption of both neuronal cell bodies in the gray matter as well as their projections and supporting cells. To explore how an impact can alter the function of brain networks, we integrated a finite element (FE) brain mechanics model with linked models of brain dynamics (Kuramoto oscillator) and vascular perfusion (Balloon-Windkessel) in this study. We used empirical resting-state functional magnetic resonance imaging (MRI) data to optimize the fit of our brain dynamics and perfusion models to clinical data. Results from the FE model were used to mimic injury in these optimized brain dynamics models: injury to the nodes (gray matter) led to a decrease in the nodal oscillation frequency, while damage to the edges (axonal connections/white matter) progressively decreased coupling among connected nodes. A total of 53 cases, including 33 non-injurious and 20 concussive head impacts experienced by professional American football players were simulated using this integrated model. We examined the correlation of injury outcomes with global measures of structural connectivity, neural dynamics, and functional connectivity of the brain networks when using different lesion methods. Results show that injurious head impacts cause significant alterations in global network topology regardless of lesion methods. Changes between the disrupted and healthy functional connectivity (measured by Pearson correlation) consistently correlated well with injury outcomes (AUC≥0.75), although the predictive performance is not significantly different (p>0.05) to that of traditional kinematic measures (angular acceleration). Intriguingly, our lesion model for gray matter damage predicted increases in global efficiency and clustering coefficient with increases in injury risk, while disrupting axonal connections led to lower network efficiency and clustering. When both injury mechanisms were combined into a single injury prediction model, the injury prediction performance depended on the thresholds used to determine neurodegeneration and mechanical tolerance for axonal injury. Together, these results point towards complex effects of mechanical trauma to the brain and provide a new framework for understanding brain injury at a causal mechanistic level and developing more effective diagnostic methods and therapeutic interventions.

Cerebral vascular strains in dynamic head impact using an upgraded model with brain material property heterogeneity

Cerebral vascular injury (CVI) is a frequent consequence of traumatic brain injury but has often been neglected. Substantial experimental work exists on vascular material properties and failure/subfailure thresholds. However, little is known about vascular in vivo loading conditions in dynamic head impact, which is necessary to investigate the risk, severity, and extent of CVI. In this study, we resort to the Worcester Head Injury Model (WHIM) V2.1 for investigation. The model embeds the cerebral vasculature network and is further upgraded to incorporate brain material property heterogeneity based on magnetic resonance elastography. The brain material property is calibrated to match with the previously validated anisotropic V1.0 version in terms of whole-brain strains against six experimental datasets of a wide range of blunt impact conditions. The upgraded WHIM is finally used to simulate five representative real-world head impacts drawn from contact sports and automotive crashes. We find that peak strains in veins are considerably higher than those in arteries and that peak circumferential strains are also higher than peak axial strains. For a typical concussive head impact, cerebral vascular axial strains reach the lowest reported yield strain of ∼7-8%. For severe automotive impacts, axial strains could reach ∼20%, which is on the order of the lowest reported ultimate failure strain of ∼24%. These results suggest in vivo mechanical loading conditions of the cerebral vasculature (excluding bridging veins not assessed here) due to rapid head rotation are at the lower end of failure/subfailure thresholds established from ex vivo experiments. This study provides some first insight into the risk, severity, and extent of CVI in real-world head impacts.

Quantification of Head Acceleration Events in Rugby League: An Instrumented Mouthguard and Video Analysis Pilot Study

Instrumented mouthguards (iMG) were used to collect head acceleration events (HAE) in men's professional rugby league matches. Peak linear acceleration (PLA), peak angular acceleration (PAA) and peak change in angular velocity (ΔPAV) were collected using custom-fit iMG set with a 5 g single iMG-axis recording threshold. iMG were fitted to ten male Super League players for thirty-one player matches. Video analysis was conducted on HAE to identify the contact event; impacted player; tackle stage and head loading type. A total of 1622 video-verified HAE were recorded. Approximately three-quarters of HAE (75.7%) occurred below 10 g. Most (98.2%) HAE occurred during tackles (59.3% to tackler; 40.7% to ball carrier) and the initial collision stage of the tackle (43.9%). The initial collision stage resulted in significantly greater PAA and ΔPAV than secondary contact and play the ball tackle stages (p < 0.001). Indirect HAE accounted for 29.8% of HAE and resulted in significantly greater ΔPAV (p < 0.001) than direct HAE, but significantly lower PLA (p < 0.001). Almost all HAE were sustained in the tackle, with the majority occurring during the initial collision stage, making it an area of focus for the development of player protection strategies for both ball carriers and tacklers. League-wide and community-level implementation of iMG could enable a greater understanding of head acceleration exposure between playing positions, cohorts, and levels of play.

Displacement voxelization to resolve mesh-image mismatch: Application in deriving dense white matter fiber strains

BACKGROUND AND OBJECTIVE

It is common to combine biomechanical modeling and medical images for multimodal analyses. However, mesh-image mismatch may occur that prevents direct information exchange. To eliminate mesh-image mismatch, we develop a simple but elegant displacement voxelization technique based on image voxel corner nodes to achieve voxel-wise strain. We then apply the technique to derive dense white matter fiber strains along whole-brain tractography (∼35 k fiber tracts consisting of ∼3.3 million sampling points) resulting from head impact.

METHODS

Displacements at image voxel corner nodes are first obtained from model simulation via scattered interpolation. Each voxel is then scaled linearly to form a unit hexahedral element. This allows convenient and efficient voxel-wise strain tensor calculation and displacement interpolation at arbitrary fiber sampling points via shape functions. Fiber strains from displacement interpolation are then compared with those from the commonly used strain tensor projection using either voxel- or element-wise strain tensors.

RESULTS

Based on a synthetic displacement field, fiber strains interpolated from voxelized displacement are considerably more accurate than those from strain tensor projection relative to the prescribed ground-truth (determinant of coefficient (R²) of 1.00 and root mean squared error (RMSE) of 0.01 vs. 0.87 and 0.10, respectively). For a set of real-world reconstructed head impacts (N = 53), the strain tensor projection method performs similarly poorly (R² of 0.80-0.90 and RMSE of 0.03-0.07), with overestimation strongly correlated with strain magnitude (Pearson correlation coefficient >0.9). Up to ∼15% of the fiber strains are overestimated by more than the lower bound of a conservative injury threshold of 0.09. The percentage increases to ∼37% when halving the threshold. Voxel interpolation is also significantly more efficient (15 s vs. 40 s for element strain tensor projection, without parallelization).

CONCLUSIONS

Voxelized displacement interpolation is considerably more accurate and efficient in deriving dense white matter fiber strains than strain tensor projection. The latter generally overestimates with overestimation magnitude strongly correlating with fiber strain magnitude. Displacement voxelization is an effective technique to eliminate mesh-image mismatch and generates a convenient image representation of tissue deformation. This technique can be generalized to broadly facilitate a diverse range of image-related biomechanical problems for multimodal analyses. The convenient image format may also promote and facilitate biomechanical data sharing in the future.

Instantaneous Brain Strain Estimation for Automotive Head Impacts via Deep Learning

Efficient brain strain estimation is critical for routine application of a head injury model. Lately, a convolutional neural network (CNN) has been successfully developed to estimate spatially detailed brain strains instantly and accurately in contact sports. Here, we extend its application to automotive head impacts, where impact profiles are typically more complex with longer durations. Head impact kinematics (N=458) from two public databases were used to generate augmented impacts (N=2694). They were simulated using the anisotropic Worcester Head Injury Model (WHIM) V1.0, which provided baseline elementwise peak maximum principal strain (MPS). For each augmented impact, rotational velocity (v_rot) and the corresponding rotational acceleration (a_rot) profiles were concatenated as static images to serve as CNN input. Three training strategies were evaluated: 1) "baseline", using random initial weights; 2) "transfer learning", using weight transfer from a previous CNN model trained on head impacts drawn from contact sports; and 3) "combined training", combining previous training data from contact sports (N=5661) for training. The combined training achieved the best performances. For peak MPS, the CNN achieved a coefficient of determination (R²) of 0.932 and root mean squared error (RMSE) of 0.031 for the real-world testing dataset. It also achieved a success rate of 60.5% and 94.8% for elementwise MPS, where the linear regression slope, k, and correlation coefficient, r, between estimated and simulated MPS did not deviate from 1.0 (when identical) by more than 0.1 and 0.2, respectively. Cumulative strain damage measure (CSDM) from the CNN estimation was also highly accurate compared to those from direct simulation across a range of thresholds (R² of 0.899-0.943 with RMSE of 0.054-0.069). Finally, the CNN achieved an average k and r of 0.98±0.12 and 0.90±0.07, respectively, for six reconstructed car crash impacts drawn from two other sources independent of the training dataset. Importantly, the CNN is able to efficiently estimate elementwise MPS with sufficient accuracy while conventional kinematic injury metrics cannot. Therefore, the CNN has the potential to supersede current kinematic injury metrics that can only approximate a global peak MPS or CSDM. The CNN technique developed here may offer enhanced utility in the design and development of head protective countermeasures, including in the automotive industry. This is the first study aimed at instantly estimating spatially detailed brain strains for automotive head impacts, which employs >8.8 thousand impact simulations generated from ~1.5 years of nonstop computations on a high-performance computing platform.

A Machine Learning Approach to Investigate the Uncertainty of Tissue-Level Injury Metrics for Cerebral Contusion

Controlled cortical impact (CCI) on porcine brain is often utilized to investigate the pathophysiology and functional outcome of focal traumatic brain injury (TBI), such as cerebral contusion (CC). Using a finite element (FE) model of the porcine brain, the localized brain strain and strain rate resulting from CCI can be computed and compared to the experimentally assessed cortical lesion. This way, tissue-level injury metrics and corresponding thresholds specific for CC can be established. However, the variability and uncertainty associated with the CCI experimental parameters contribute to the uncertainty of the provoked cortical lesion and, in turn, of the predicted injury metrics. Uncertainty quantification via probabilistic methods (Monte Carlo simulation, MCS) requires a large number of FE simulations, which results in a time-consuming process. Following the recent success of machine learning (ML) in TBI biomechanical modeling, we developed an artificial neural network as surrogate of the FE porcine brain model to predict the brain strain and the strain rate in a computationally efficient way. We assessed the effect of several experimental and modeling parameters on four FE-derived CC injury metrics (maximum principal strain, maximum principal strain rate, product of maximum principal strain and strain rate, and maximum shear strain). Next, we compared the in silico brain mechanical response with cortical damage data from in vivo CCI experiments on pig brains to evaluate the predictive performance of the CC injury metrics. Our ML surrogate was capable of rapidly predicting the outcome of the FE porcine brain undergoing CCI. The now computationally efficient MCS showed that depth and velocity of indentation were the most influential parameters for the strain and the strain rate-based injury metrics, respectively. The sensitivity analysis and comparison with the cortical damage experimental data indicate a better performance of maximum principal strain and maximum shear strain as tissue-level injury metrics for CC. These results provide guidelines to optimize the design of CCI tests and bring new insights to the understanding of the mechanical response of brain tissue to focal traumatic brain injury. Our findings also highlight the potential of using ML for computationally efficient TBI biomechanics investigations.

Toward a Comprehensive Delineation of White Matter Tract-Related Deformation

Finite element (FE) models of the human head are valuable instruments to explore the mechanobiological pathway from external loading, localized brain response, and resultant injury risks. The injury predictability of these models depends on the use of effective criteria as injury predictors. The FE-derived normal deformation along white matter (WM) fiber tracts (i.e., tract-oriented strain) recently has been suggested as an appropriate predictor for axonal injury. However, the tract-oriented strain only represents a partial depiction of the WM fiber tract deformation. A comprehensive delineation of tract-related deformation may improve the injury predictability of the FE head model by delivering new tract-related criteria as injury predictors. Thus, the present study performed a theoretical strain analysis to comprehensively characterize the WM fiber tract deformation by relating the strain tensor of the WM element to its embedded fiber tract. Three new tract-related strains with exact analytical solutions were proposed, measuring the normal deformation perpendicular to the fiber tracts (i.e., tract-perpendicular strain), and shear deformation along and perpendicular to the fiber tracts (i.e., axial-shear strain and lateral-shear strain, respectively). The injury predictability of these three newly proposed strain peaks along with the previously used tract-oriented strain peak and maximum principal strain (MPS) were evaluated by simulating 151 impacts with known outcome (concussion or non-concussion). The results preliminarily showed that four tract-related strain peaks exhibited superior performance than MPS in discriminating concussion and non-concussion cases. This study presents a comprehensive quantification of WM tract-related deformation and advocates the use of orientation-dependent strains as criteria for injury prediction, which may ultimately contribute to an advanced mechanobiological understanding and enhanced computational predictability of brain injury.

Effective Head Impact Kinematics to Preserve Brain Strain

Conventional kinematics-based brain injury metrics often approximate peak maximum principal strain (MPS) of the whole brain but ignore the anatomical location of occurrence. In this study, we develop effective impact kinematics consisting of peak rotational velocity and the associated rotational axis to preserve not only peak MPS but also spatially detailed MPS. A pre-computed brain response atlas (pcBRA) serves as a common reference. A training dataset (N = 3069) is used to develop a convolutional neural network (CNN) to automate impact simplification. When preserving peak MPS alone, the CNN-estimated effective peak rotational velocity achieves a coefficient of determination ([Formula: see text]) of ~ 0.96 relative to the directly identified counterpart, far outperforming nominal peak velocity from the resultant profiles ([Formula: see text] of ~ 0.34). Impacts from a subset of data (N = 1900) are also successfully matched with pcBRA idealized impacts based on elementwise MPS, where their regression slope and Pearson correlation coefficient do not deviate from 1.0 (when identical) by more than 0.1. The CNN-estimated effective peak rotation velocity and rotational axis are sufficiently accurate for ~ 73.5% of the impacts. This is not possible for the nominal peak velocity or any other conventional injury metric. The performance may be further improved by expanding the pcBRA to include deceleration and focusing on region-wise strains. This study establishes a new avenue to reduce an arbitrary head impact into an idealized but actual "impact mode" characterized by triplets of basic kinematic variables. They retain specific physical interpretations of head impact and may be an advancement over state-of-the-art kinematics-based scalar metrics for more effective impact comparison in the future.

The relationship between brain injury criteria and brain strain across different types of head impacts can be different

Multiple brain injury criteria (BIC) are developed to quickly quantify brain injury risks after head impacts. These BIC originated from different head impact types (e.g. sports and car crashes) are widely used in risk evaluation. However, the accuracy of using the BIC on brain injury risk estimation across head impact types has not been evaluated. Physiologically, brain strain is often considered the key parameter of brain injury. To evaluate the BIC's risk estimation accuracy across five datasets comprising different head impact types, linear regression was used to model 95% maximum principal strain, 95% maximum principal strain at the corpus callosum and cumulative strain damage (15%) on 18 BIC. The results show significantly different relationships between BIC and brain strain across datasets, indicating the same BIC value may suggest different brain strain across head impact types. The accuracy of brain strain regression is generally decreasing if the BIC regression models are fitted on a dataset with a different type of head impact rather than on the dataset with the same type. Given this finding, this study raises concerns for applying BIC to estimate the brain injury risks for head impacts different from the head impacts on which the BIC was developed.

Instantaneous Whole-Brain Strain Estimation in Dynamic Head Impact

Head injury models are notoriously time consuming and resource demanding in simulations, which prevents routine application. Here, we extend a convolutional neural network (CNN) to instantly estimate element-wise distribution of peak maximum principal strain (MPS) of the entire brain (>36 k speedup accomplished on a low-end computing platform). To achieve this, head impact rotational velocity and acceleration temporal profiles are combined into two-dimensional images to serve as CNN input for training and prediction of MPS. Compared with the directly simulated counterparts, the CNN-estimated responses (magnitude and distribution) are sufficiently accurate for 92.1% of the cases via 10-fold cross-validation using impacts drawn from the real world (n = 5661; range of peak rotational velocity in augmented data extended to 2-40 rad/sec). The success rate further improves to 97.1% for "in-range" impacts (n = 4298). When using the same CNN architecture to train (n = 3064) and test on an independent, reconstructed National Football League (NFL) impact dataset (n = 53; 20 concussions and 33 non-injuries), 51 out of 53, or 96.2% of the cases, are sufficiently accurate. The estimated responses also achieve virtually identical concussion prediction performances relative to the directly simulated counterparts, and they often outperform peak MPS of the whole brain (e.g., accuracy of 0.83 vs. 0.77 via leave-one-out cross-validation). These findings support the use of CNN for accurate and efficient estimation of spatially detailed brain strains across the vast majority of head impacts in contact sports. Our technique may hold the potential to transform traumatic brain injury (TBI) research and the design and testing standards of head protective gears by facilitating the transition from acceleration-based approximation to strain-based design and analysis. This would have broad implications in the TBI biomechanics field to accelerate new scientific discoveries. The pre-trained CNN is freely available online at https://github.com/Jilab-biomechanics/CNN-brain-strains.

Rapid Estimation of Entire Brain Strain Using Deep Learning Models

OBJECTIVE

Many recent studies have suggested that brain deformation resulting from a head impact is linked to the corresponding clinical outcome, such as mild traumatic brain injury (mTBI). Even though several finite element (FE) head models have been developed and validated to calculate brain deformation based on impact kinematics, the clinical application of these FE head models is limited due to the time-consuming nature of FE simulations. This work aims to accelerate the process of brain deformation calculation and thus improve the potential for clinical applications.

METHODS

We propose a deep learning head model with a five-layer deep neural network and feature engineering, and trained and tested the model on 2511 total head impacts from a combination of head model simulations and on-field college football and mixed martial arts impacts.

RESULTS

The proposed deep learning head model can calculate the maximum principal strain (Green Lagrange) for every element in the entire brain in less than 0.001s with an average root mean squared error of 0.022, and with a standard deviation of 0.001 over twenty repeats with random data partition and model initialization.

CONCLUSION

Trained and tested using the dataset of 2511 head impacts, this model can be applied to various sports in the calculation of brain strain with accuracy, and its applicability can even further be extended by incorporating data from other types of head impacts.

SIGNIFICANCE

In addition to the potential clinical application in real-time brain deformation monitoring, this model will help researchers estimate the brain strain from a large number of head impacts more efficiently than using FE models.

Ranking and Rating Bicycle Helmet Safety Performance in Oblique Impacts Using Eight Different Brain Injury Models

Bicycle helmets are shown to offer protection against head injuries. Rating methods and test standards are used to evaluate different helmet designs and safety performance. Both strain-based injury criteria obtained from finite element brain injury models and metrics derived from global kinematic responses can be used to evaluate helmet safety performance. Little is known about how different injury models or injury metrics would rank and rate different helmets. The objective of this study was to determine how eight brain models and eight metrics based on global kinematics rank and rate a large number of bicycle helmets (n=17) subjected to oblique impacts. The results showed that the ranking and rating are influenced by the choice of model and metric. Kendall’s tau varied between 0.50 and 0.95 when the ranking was based on maximum principal strain from brain models. One specific helmet was rated as 2-star when using one brain model but as 4-star by another model. This could cause confusion for consumers rather than inform them of the relative safety performance of a helmet. Therefore, we suggest that the biomechanics community should create a norm or recommendation for future ranking and rating methods.

Low-Rank Representation of Head Impact Kinematics: A Data-Driven Emulator

Head motion induced by impacts has been deemed as one of the most important measures in brain injury prediction, given that the vast majority of brain injury metrics use head kinematics as input. Recently, researchers have focused on using fast approaches, such as machine learning, to approximate brain deformation in real time for early brain injury diagnosis. However, training such models requires large number of kinematic measurements, and therefore data augmentation is required given the limited on-field measured data available. In this study we present a principal component analysis-based method that emulates an empirical low-rank substitution for head impact kinematics, while requiring low computational cost. In characterizing our existing data set of 537 head impacts, each consisting of 6 degrees of freedom measurements, we found that only a few modes, e.g., 15 in the case of angular velocity, is sufficient for accurate reconstruction of the entire data set. Furthermore, these modes are predominantly low frequency since over 70% of the angular velocity response can be captured by modes that have frequencies under 40 Hz. We compared our proposed method against existing impact parametrization methods and showed significantly better performance in injury prediction using a range of kinematic-based metrics—such as head injury criterion (HIC), rotational injury criterion (RIC), and brain injury metric (BrIC)—and brain tissue deformation-based metrics—such as brain angle metric (BAM), maximum principal strain (MPS), and axonal fiber strains (FS). In all cases, our approach reproduced injury metrics similar to the ground truth measurements with no significant difference, whereas the existing methods obtained significantly different (p < 0.01) values as well as substantial differences in injury classification sensitivity and specificity. This emulator will enable us to provide the necessary data augmentation to build a head impact kinematic data set of any size. The emulator and corresponding examples are available on our website¹.

Displacement- and Strain-Based Discrimination of Head Injury Models across a Wide Range of Blunt Conditions

Successful validation of a head injury model is critical to ensure its biofidelity. However, there is an ongoing debate on what experimental data are suitable for model validation. Here, we report that CORrelation and Analysis (CORA) scores based on the commonly adopted relative brain-skull displacements or recent marker-based strains from cadaveric head impacts may not be effective in discriminating model-simulated whole-brain strains across a wide range of blunt conditions. We used three versions of the Worcester Head Injury Model (WHIM; isotropic and anisotropic WHIM V1.0, and anisotropic WHIM V1.5) to simulate 19 experiments, including eight high-rate cadaveric impacts, seven mid-rate cadaveric pure rotations simulating impacts in contact sports, and four in vivo head rotation/extension tests. All WHIMs achieved similar average CORA scores based on cadaveric displacement (~ 0.70; 0.47-0.88) and strain (V1.0: 0.86; 0.73-0.97 vs. V1.5: 0.78; 0.62-0.96), using the recommended settings. However, WHIM V1.5 produced ~ 1.17-2.69 times strain of the two V1.0 variants with substantial differences in strain distribution as well (Pearson correlation of ~ 0.57-0.92) when comparing their whole-brain strains across the range of blunt conditions. Importantly, their strain magnitude differences were similar to that in cadaveric marker-based strain (~ 1.32-3.79 times). This suggests that cadaveric strains are capable of discriminating head injury models for their simulated whole-brain strains (e.g., by using CORA magnitude sub-rating alone or peak strain magnitude ratio), although the aggregated CORA may not. This study may provide fresh insight into head injury model validation and the harmonization of simulation results from diverse head injury models. It may also facilitate future experimental designs to improve model validation.