Concussion classification via deep learning using whole-brain white matter fiber strains

Deep learning-based overall survival prediction model in patients with rare cancer: a case study for primary central nervous system lymphoma

PURPOSE

Primary central nervous system lymphoma (PCNSL) is a rare, aggressive form of extranodal non-Hodgkin lymphoma. To predict the overall survival (OS) in advance is of utmost importance as it has the potential to aid clinical decision-making. Though radiomics-based machine learning (ML) has demonstrated the promising performance in PCNSL, it demands large amounts of manual feature extraction efforts from magnetic resonance images beforehand. deep learning (DL) overcomes this limitation.

METHODS

In this paper, we tailored the 3D ResNet to predict the OS of patients with PCNSL. To overcome the limitation of data sparsity, we introduced data augmentation and transfer learning, and we evaluated the results using r stratified k-fold cross-validation. To explain the results of our model, gradient-weighted class activation mapping was applied.

RESULTS

We obtained the best performance (the standard error) on post-contrast T1-weighted (T1Gd)-area under curve [Formula: see text], accuracy [Formula: see text], precision [Formula: see text], recall [Formula: see text] and F1-score [Formula: see text], while compared with ML-based models on clinical data and radiomics data, respectively, further confirming the stability of our model. Also, we observed that PCNSL is a whole-brain disease and in the cases where the OS is less than 1 year, it is more difficult to distinguish the tumor boundary from the normal part of the brain, which is consistent with the clinical outcome.

CONCLUSIONS

All these findings indicate that T1Gd can improve prognosis predictions of patients with PCNSL. To the best of our knowledge, this is the first time to use DL to explain model patterns in OS classification of patients with PCNSL. Future work would involve collecting more data of patients with PCNSL, or additional retrospective studies on different patient populations with rare diseases, to further promote the clinical role of our model.

Artificial intelligence CAD tools in trauma imaging: a scoping review from the American Society of Emergency Radiology (ASER) AI/ML Expert Panel

BACKGROUND

AI/ML CAD tools can potentially improve outcomes in the high-stakes, high-volume model of trauma radiology. No prior scoping review has been undertaken to comprehensively assess tools in this subspecialty.

PURPOSE

To map the evolution and current state of trauma radiology CAD tools along key dimensions of technology readiness.

METHODS

Following a search of databases, abstract screening, and full-text document review, CAD tool maturity was charted using elements of data curation, performance validation, outcomes research, explainability, user acceptance, and funding patterns. Descriptive statistics were used to illustrate key trends.

RESULTS

A total of 4052 records were screened, and 233 full-text articles were selected for content analysis. Twenty-one papers described FDA-approved commercial tools, and 212 reported algorithm prototypes. Works ranged from foundational research to multi-reader multi-case trials with heterogeneous external data. Scalable convolutional neural network-based implementations increased steeply after 2016 and were used in all commercial products; however, options for explainability were narrow. Of FDA-approved tools, 9/10 performed detection tasks. Dataset sizes ranged from < 100 to > 500,000 patients, and commercialization coincided with public dataset availability. Cross-sectional torso datasets were uniformly small. Data curation methods with ground truth labeling by independent readers were uncommon. No papers assessed user acceptance, and no method included human-computer interaction. The USA and China had the highest research output and frequency of research funding.

CONCLUSIONS

Trauma imaging CAD tools are likely to improve patient care but are currently in an early stage of maturity, with few FDA-approved products for a limited number of uses. The scarcity of high-quality annotated data remains a major barrier.

An end-end deep learning framework for lesion segmentation on multi-contrast MR images-an exploratory study in a rat model of traumatic brain injury

Traumatic brain injury (TBI) engenders traumatic necrosis and penumbra-areas of secondary neural injury which are crucial targets for therapeutic interventions. Segmenting manually areas of ongoing changes like necrosis, edema, hematoma, and inflammation is tedious, error-prone, and biased. Using the multi-parametric MR data from a rodent model study, we demonstrate the effectiveness of an end-end deep learning global-attention-based UNet (GA-UNet) framework for automatic segmentation and quantification of TBI lesions. Longitudinal MR scans (2 h, 1, 3, 7, 14, 30, and 60 days) were performed on eight Sprague-Dawley rats after controlled cortical injury was performed. TBI lesion and sub-regions segmentation was performed using 3D-UNet and GA-UNet. Dice statistics (DSI) and Hausdorff distance were calculated to assess the performance. MR scan variations-based (bias, noise, blur, ghosting) data augmentation was performed to develop a robust model.Training/validation median DSI for U-Net was 0.9368 with T2w and MPRAGE inputs, whereas GA-UNet had 0.9537 for the same. Testing accuracies were higher for GA-UNet than U-Net with a DSI of 0.8232 for the T2w-MPRAGE inputs.Longitudinally, necrosis remained constant while oligemia and penumbra decreased, and edema appearing around day 3 which increased with time. GA-UNet shows promise for multi-contrast MR image-based segmentation/quantification of TBI in large cohort studies.

Volumetric MRI Findings in Mild Traumatic Brain Injury (mTBI) and Neuropsychological Outcome

Region of interest (ROI) volumetric assessment has become a standard technique in quantitative neuroimaging. ROI volume is thought to represent a coarse proxy for making inferences about the structural integrity of a brain region when compared to normative values representative of a healthy sample, adjusted for age and various demographic factors. This review focuses on structural volumetric analyses that have been performed in the study of neuropathological effects from mild traumatic brain injury (mTBI) in relation to neuropsychological outcome. From a ROI perspective, the probable candidate structures that are most likely affected in mTBI represent the target regions covered in this review. These include the corpus callosum, cingulate, thalamus, pituitary-hypothalamic area, basal ganglia, amygdala, and hippocampus and associated structures including the fornix and mammillary bodies, as well as whole brain and cerebral cortex along with the cerebellum. Ventricular volumetrics are also reviewed as an indirect assessment of parenchymal change in response to injury. This review demonstrates the potential role and limitations of examining structural changes in the ROIs mentioned above in relation to neuropsychological outcome. There is also discussion and review of the role that post-traumatic stress disorder (PTSD) may play in structural outcome in mTBI. As emphasized in the conclusions, structural volumetric findings in mTBI are likely just a single facet of what should be a multimodality approach to image analysis in mTBI, with an emphasis on how the injury damages or disrupts neural network integrity. The review provides an historical context to quantitative neuroimaging in neuropsychology along with commentary about future directions for volumetric neuroimaging research in mTBI.

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.

Towards defining biomarkers to evaluate concussions using virtual reality and a moving platform (BioVRSea)

Current diagnosis of concussion relies on self-reported symptoms and medical records rather than objective biomarkers. This work uses a novel measurement setup called BioVRSea to quantify concussion status. The paradigm is based on brain and muscle signals (EEG, EMG), heart rate and center of pressure (CoP) measurements during a postural control task triggered by a moving platform and a virtual reality environment. Measurements were performed on 54 professional athletes who self-reported their history of concussion or non-concussion. Both groups completed a concussion symptom scale (SCAT5) before the measurement. We analyzed biosignals and CoP parameters before and after the platform movements, to compare the net response of individual postural control. The results showed that BioVRSea discriminated between the concussion and non-concussion groups. Particularly, EEG power spectral density in delta and theta bands showed significant changes in the concussion group and right soleus median frequency from the EMG signal differentiated concussed individuals with balance problems from the other groups. Anterior–posterior CoP frequency-based parameters discriminated concussed individuals with balance problems. Finally, we used machine learning to classify concussion and non-concussion, demonstrating that combining SCAT5 and BioVRSea parameters gives an accuracy up to 95.5%. This study is a step towards quantitative assessment of concussion.

Safeguarding Athletes Against Head Injuries Through Advances in Technology: A Scoping Review of the Uses of Machine Learning in the Management of Sports-Related Concussion

Sports injury prevention is an important part of the athlete welfare and safeguarding research field. In sports injury prevention, sport-related concussion (SRC) has proved to be one of the most difficult and complex injuries to manage in terms of prevention, diagnosis, classification, treatment and rehabilitation. SRC can cause long-term health issues and is a commonly reported injury in both adult and youth athletes around the world. Despite increased knowledge of the prevalence of SRC, very few tools are available for diagnosing SRC in athletic settings. Recent technological innovations have resulted in different machine learning and deep learning methodologies being tested to improve the management of this complex sports injury. The purpose of this article is to summarize and map the existing research literature on the use of machine learning in the management of SRC, ascertain where there are gaps in the existing research and identify recommendations for future research. This is explored through a scoping review. A systematic search in the three electronic databases SPORTDiscus, PubMed and Scopus identified an initial 522 studies, of which 24 were included in the final review, the majority of which focused on machine learning for the prediction and prevention of SRC (N = 10), or machine learning for the diagnosis and classification of SRC (N = 11). Only 3 studies explored machine learning approaches for the treatment and rehabilitation of SRC. A main finding is that current research highlights promising practical uses (e.g., more accurate and rapid injury assessment or return-to-sport participation criteria) of machine learning in the management of SRC. The review also revealed a narrow research focus in the existing literature. As current research is primarily conducted on male adolescents or adults from team sports in North America there is an urgent need to include wider demographics in more diverse samples and sports contexts in the machine learning algorithms. If research datasets continue to be based on narrow samples of athletes, the development of any new diagnostic and predictive tools for SRC emerging from this research will be at risk. Today, these risks appear to mainly affect the health and safety of female athletes.

Phybrata Sensors and Machine Learning for Enhanced Neurophysiological Diagnosis and Treatment

Concussion injuries remain a significant public health challenge. A significant unmet clinical need remains for tools that allow related physiological impairments and longer-term health risks to be identified earlier, better quantified, and more easily monitored over time. We address this challenge by combining a head-mounted wearable inertial motion unit (IMU)-based physiological vibration acceleration (“phybrata”) sensor and several candidate machine learning (ML) models. The performance of this solution is assessed for both binary classification of concussion patients and multiclass predictions of specific concussion-related neurophysiological impairments. Results are compared with previously reported approaches to ML-based concussion diagnostics. Using phybrata data from a previously reported concussion study population, four different machine learning models (Support Vector Machine, Random Forest Classifier, Extreme Gradient Boost, and Convolutional Neural Network) are first investigated for binary classification of the test population as healthy vs. concussion (Use Case 1). Results are compared for two different data preprocessing pipelines, Time-Series Averaging (TSA) and Non-Time-Series Feature Extraction (NTS). Next, the three best-performing NTS models are compared in terms of their multiclass prediction performance for specific concussion-related impairments: vestibular, neurological, both (Use Case 2). For Use Case 1, the NTS model approach outperformed the TSA approach, with the two best algorithms achieving an F1 score of 0.94. For Use Case 2, the NTS Random Forest model achieved the best performance in the testing set, with an F1 score of 0.90, and identified a wider range of relevant phybrata signal features that contributed to impairment classification compared with manual feature inspection and statistical data analysis. The overall classification performance achieved in the present work exceeds previously reported approaches to ML-based concussion diagnostics using other data sources and ML models. This study also demonstrates the first combination of a wearable IMU-based sensor and ML model that enables both binary classification of concussion patients and multiclass predictions of specific concussion-related neurophysiological impairments.

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.

White matter structural connectivity as a biomarker for detecting juvenile myoclonic epilepsy by transferred deep convolutional neural networks with varying transfer rates

Objective. By detecting abnormal white matter changes, diffusion magnetic resonance imaging (MRI) contributes to the detection of juvenile myoclonic epilepsy (JME). In addition, deep learning has greatly improved the detection performance of various brain disorders. However, there is almost no previous study effectively detecting JME by a deep learning approach with diffusion MRI.Approach. In this study, the white matter structural connectivity was generated by tracking the white matter fibers in detail based on Q-ball imaging and neurite orientation dispersion and density imaging. Four advanced deep convolutional neural networks (CNNs) were deployed by using the transfer learning approach, in which the transfer rate searching strategy was proposed to achieve the best detection performance.Main results. Our results showed: (a) Compared to normal control, the white matter' neurite density of JME was significantly decreased. The most significantly abnormal fiber tracts between the two groups were found to be cortico-cortical connection tracts. (b) The proposed transfer rate searching approach contributed to find each CNN's best performance, in which the best JME detection accuracy of 92.2% was achieved by using the Inception_resnet_v2 network with a 16% transfer rate.Significance. The results revealed: (a) Through detection of the abnormal white matter changes, the white matter structural connectivity can be used as a useful biomarker for detecting JME, which helps to characterize the pathophysiology of epilepsy. (b) The proposed transfer rate, as a new hyperparameter, promotes the CNNs transfer learning performance in detecting JME.

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.

Machine learning and deep learning to predict mortality in patients with spontaneous coronary artery dissection

Machine learning (ML) and deep learning (DL) can successfully predict high prevalence events in very large databases (big data), but the value of this methodology for risk prediction in smaller cohorts with uncommon diseases and infrequent events is uncertain. The clinical course of spontaneous coronary artery dissection (SCAD) is variable, and no reliable methods are available to predict mortality. Based on the hypothesis that machine learning (ML) and deep learning (DL) techniques could enhance the identification of patients at risk, we applied a deep neural network to information available in electronic health records (EHR) to predict in-hospital mortality in patients with SCAD. We extracted patient data from the EHR of an extensive urban health system and applied several ML and DL models using candidate clinical variables potentially associated with mortality. We partitioned the data into training and evaluation sets with cross-validation. We estimated model performance based on the area under the receiver-operator characteristics curve (AUC) and balanced accuracy. As sensitivity analyses, we examined results limited to cases with complete clinical information available. We identified 375 SCAD patients of which mortality during the index hospitalization was 11.5%. The best-performing DL algorithm identified in-hospital mortality with AUC 0.98 (95% CI 0.97-0.99), compared to other ML models (P < 0.0001). For prediction of mortality using ML models in patients with SCAD, the AUC ranged from 0.50 with the random forest method (95% CI 0.41-0.58) to 0.95 with the AdaBoost model (95% CI 0.93-0.96), with intermediate performance using logistic regression, decision tree, support vector machine, K-nearest neighbors, and extreme gradient boosting methods. A deep neural network model was associated with higher predictive accuracy and discriminative power than logistic regression or ML models for identification of patients with ACS due to SCAD prone to early mortality.

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.

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.

A Multibody Model for Predicting Spatial Distribution of Human Brain Deformation Following Impact Loading

With an increasing focus on long-term consequences of concussive brain injuries, there is a new emphasis on developing tools that can accurately predict the mechanical response of the brain to impact loading. Although finite element models (FEM) estimate the brain response under dynamic loading, these models are not capable of delivering rapid (∼seconds) estimates of the brain's mechanical response. In this study, we develop a multibody spring-mass-damper model that estimates the regional motion of the brain to rotational accelerations delivered either about one anatomic axis or across three orthogonal axes simultaneously. In total, we estimated the deformation across 120 locations within a 50th percentile human brain. We found the multibody model (MBM) correlated, but did not precisely predict, the computed finite element response (average relative error: 18.4 ± 13.1%). We used machine learning (ML) to combine the prediction from the MBM and the loading kinematics (peak rotational acceleration, peak rotational velocity) and significantly reduced the discrepancy between the MBM and FEM (average relative error: 9.8 ± 7.7%). Using an independent sports injury testing set, we found the hybrid ML model also correlated well with predictions from a FEM (average relative error: 16.4 ± 10.2%). Finally, we used this hybrid MBM-ML approach to predict strains appearing in different locations throughout the brain, with average relative error estimates ranging from 8.6% to 25.2% for complex, multi-axial acceleration loading. Together, these results show a rapid and reasonably accurate method for predicting the mechanical response of the brain for single and multiplanar inputs, and provide a new tool for quickly assessing the consequences of impact loading throughout the brain.

Predicting Concussion Outcome by Integrating Finite Element Modeling and Network Analysis

Concussion is a significant public health problem affecting 1.6-2.4 million Americans annually. An alternative to reducing the burden of concussion is to reduce its incidence with improved protective equipment and injury mitigation systems. Finite element (FE) models of the brain response to blunt trauma are often used to estimate injury potential and can lead to improved helmet designs. However, these models have yet to incorporate how the patterns of brain connectivity disruption after impact affects the relay of information in the injured brain. Furthermore, FE brain models typically do not consider the differences in individual brain structural connectivities and their purported role in concussion risk. Here, we use graph theory techniques to integrate brain deformations predicted from FE modeling with measurements of network efficiency to identify brain regions whose connectivity characteristics may influence concussion risk. We computed maximum principal strain in 129 brain regions using head kinematics measured from 53 professional football impact reconstructions that included concussive and non-concussive cases. In parallel, using diffusion spectrum imaging data from 30 healthy subjects, we simulated structural lesioning of each of the same 129 brain regions. We simulated lesioning by removing each region one at a time along with all its connections. In turn, we computed the resultant change in global efficiency to identify regions important for network communication. We found that brain regions that deformed the most during an impact did not overlap with regions most important for network communication (Pearson's correlation, ρ = 0.07; p = 0.45). Despite this dissimilarity, we found that predicting concussion incidence was equally accurate when considering either areas of high strain or of high importance to global efficiency. Interestingly, accuracy for concussion prediction varied considerably across the 30 healthy connectomes. These results suggest that individual network structure is an important confounding variable in concussion prediction and that further investigation of its role may improve concussion prediction and lead to the development of more effective protective equipment.

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.

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

Conventional brain injury metrics are scalars that treat the whole head/brain as a single unit but do not characterize the distribution of brain responses. Here, we establish a network-based "response feature matrix" to characterize the magnitude and distribution of impact-induced brain strains. The network nodes and edges encode injury risks to the gray matter regions and their white matter interconnections, respectively. The utility of the metric is illustrated in injury prediction using three independent, real-world datasets: two reconstructed impact datasets from the National Football League (NFL) and Virginia Tech, respectively, and measured concussive and non-injury impacts from Stanford University. Injury predictions with leave-one-out cross-validation are conducted using the two reconstructed datasets separately, and then by combining all datasets into one. Using support vector machine, the network-based injury predictor consistently outperforms four baseline scalar metrics including peak maximum principal strain of the whole brain (MPS), peak linear/rotational acceleration, and peak rotational velocity across all five selected performance measures (e.g., maximized accuracy of 0.887 vs. 0.774 and 0.849 for MPS and rotational acceleration with corresponding positive predictive values of 0.938, 0.772, and 0.800, respectively, using the reconstructed NFL dataset). With sufficient training data, real-world injury prediction is similar to leave-one-out in-sample evaluation, suggesting the potential advantage of the network-based injury metric over conventional scalar metrics. The network-based response feature matrix significantly extends scalar metrics by sampling the brain strains more completely, which may serve as a useful framework potentially allowing for other applications such as characterizing injury patterns or facilitating targeted multi-scale modeling in the future.

Current trends in biomarker discovery and analysis tools for traumatic brain injury

Traumatic brain injury (TBI) affects 1.7 million people in the United States each year, causing lifelong functional deficits in cognition and behavior. The complex pathophysiology of neural injury is a primary barrier to developing sensitive and specific diagnostic tools, which consequentially has a detrimental effect on treatment regimens. Biomarkers of other diseases (e.g. cancer) have provided critical insight into disease emergence and progression that lend to developing powerful clinical tools for intervention. Therefore, the biomarker discovery field has recently focused on TBI and made substantial advancements to characterize markers with promise of transforming TBI patient diagnostics and care. This review focuses on these key advances in neural injury biomarkers discovery, including novel approaches spanning from omics-based approaches to imaging and machine learning as well as the evolution of established techniques.

Mesh Convergence Behavior and the Effect of Element Integration of a Human Head Injury Model

Numerous head injury models exist that vary in mesh density by orders of magnitude. A careful study of the mesh convergence behavior is necessary, especially in terms of strain most relevant to brain injury. To this end, as well as to investigate the effect of element integration scheme on simulated strains, we re-meshed the Worcester Head Injury Model at five mesh densities (~ 7.2-1000 k high-quality hexahedral elements of the brain). Results from explicit dynamic simulations of three cadaveric impacts and an in vivo head rotation were compared. First, scalar metrics of the whole brain only considering magnitude were used, including peak maximum principal strain and population-based median strain. They were further extended to deep white matter regions and the entire brain elements, respectively, to form two "response vectors" to account for both magnitude and distribution. Using benchmark enhanced full-integration elements (C3D8I), a minimum of 202.8 k brain elements were necessary to converge for response vectors of the deep white matter regions. This model was further used to simulate with reduced integration (C3D8R). We found that hourglass energy higher than the common rule of thumb (e.g., up to 44.38% vs. < 10% of internal energy) was necessary to maintain comparable strain relative to C3D8I. Based on these results, it is recommended that a human head injury model should have a minimum number of 202.8 k elements, or an average element size of no larger than 1.8 mm, for the brain. C3D8R formulation with relax stiffness hourglass control using a high scaling factor is also recommended to achieve sufficient accuracy without substantial computational cost.

Performance Evaluation of a Pre-computed Brain Response Atlas in Dummy Head Impacts

A pre-computed brain response atlas (pcBRA) may have the potential to accelerate the investigation of the biomechanical mechanisms of traumatic brain injury on a large-scale. In this study, we further enhance the technique and evaluate its performance using six degree-of-freedom angular velocity profiles from dummy head impacts. Using the pcBRA to simplify profiles into acceleration-only shapes, sufficiently accurate strain estimates were obtained for impacts with a major dominating velocity peak. However, they were largely under-estimated when substantial deceleration occurred that reversed the direction of the angular velocity. For these impacts, estimation accuracy was substantially improved with a biphasic profile simplification (average correlation coefficient and linear regression slope of 0.92 ± 0.03 and 0.95 ± 0.07 for biphasic shapes, respectively, vs. 0.80 ± 0.06 and 0.80 ± 0.08 for acceleration-only shapes). Peak maximum principal strain (ɛ p) and cumulative strain damage measure (CSDM) from the estimated strains consistently correlated stronger than kinematic metrics with respect to the baseline ɛ p and CSDM from the directly simulated responses, regardless of the brain region, and by a large margin (e.g., correlation of 0.93 vs. 0.75 compared to Brain Injury Criterion (BrIC) for ɛ p in the whole-brain, and 0.91 vs. 0.47 compared to BrIC for CSDM in the corpus callosum). These findings further support the pre-computation technique for accurate, real-time strain estimation, which could be important to accelerate model-based brain injury studies in the future.