Auto-segmentation of the tibia and femur from knee MR images via deep learning and its application to cartilage strain and recovery.

The ability to efficiently and reproducibly generate subject-specific 3D models of bone and soft tissue is important to many areas of musculoskeletal research. However, methodologies requiring such models have largely been limited by lengthy manual segmentation times. Recently, machine learning, and more specifically, convolutional neural networks, have shown potential to alleviate this bottleneck in research throughput. Thus, the purpose of this work was to develop a modified version of the convolutional neural network architecture U-Net to automate segmentation of the tibia and femur from double echo steady state knee magnetic resonance (MR) images. Our model was trained on a dataset of over 4,000 MR images from 34 subjects, segmented by three experienced researchers, and reviewed by a musculoskeletal radiologist. For our validation and testing sets, we achieved dice coefficients of 0.985 and 0.984, respectively. As further testing, we applied our trained model to a prior study of tibial cartilage strain and recovery. In this analysis, across all subjects, there were no statistically significant differences in cartilage strain between the machine learning and ground truth bone models, with a mean difference of 0.2 ± 0.7 % (mean ± 95 % confidence interval). This difference is within the measurement resolution of previous cartilage strain studies from our lab using manual segmentation. In summary, we successfully trained, validated, and tested a machine learning model capable of segmenting MR images of the knee, achieving results that are comparable to trained human segmenters.

Use of a Novel Multimodal Imaging Technique to Model In Vivo Quadriceps Force and ACL Strain During Dynamic Activity.

BACKGROUND: Quadriceps loading of the anterior cruciate ligament (ACL) may play a role in the noncontact mechanism of ACL injury. Musculoskeletal modeling techniques are used to estimate the intrinsic force of the quadriceps acting at the knee joint. PURPOSE/HYPOTHESIS: The purpose of this paper was to develop a novel musculoskeletal model of in vivo quadriceps force during dynamic activity. We used the model to estimate quadriceps force in relation to ACL strain during a single-leg jump. We hypothesized that quadriceps loading of the ACL would reach a local maximum before initial ground contact with the knee positioned in extension. STUDY DESIGN: Descriptive laboratory study. METHODS: Six male participants underwent magnetic resonance imaging in addition to high-speed biplanar radiography during a single-leg jump. Three-dimensional models of the knee joint, including the femur, tibia, patellofemoral cartilage surfaces, and attachment-site footprints of the patellar tendon, quadriceps tendon, and ACL, were created from the magnetic resonance imaging scans. The bone models were registered to the biplanar radiographs, thereby reproducing the positions of the knee joint at the time of radiographic imaging. The magnitude of quadriceps force was determined for each knee position based on a 3-dimensional balance of the forces and moments of the patellar tendon and the patellofemoral cartilage contact acting on the patella. Knee kinematics and ACL strain were determined for each knee position. RESULTS: A local maximum in average quadriceps force of approximately 6500 N (8.4× body weight) occurred before initial ground contact. ACL strain increased concurrently with quadriceps force when the knee was positioned in extension. CONCLUSION: This novel participant-specific modeling technique provides estimates of in vivo quadriceps force during physiologic dynamic loading. A local maximum in quadriceps force before initial ground contact may tension the ACL when the knee is positioned in extension. CLINICAL RELEVANCE: These data contribute to understanding noncontact ACL injury mechanisms and the potential role of quadriceps activation in these injuries.

Mechanical metrics may show improved ability to predict osteoarthritis compared to T1rho mapping.

Changes in cartilage structure and composition are commonly observed during the progression of osteoarthritis (OA). Importantly, quantitative magnetic resonance imaging (MRI) methods, such as T1rho relaxation imaging, can noninvasively provide in vivo metrics that reflect changes in cartilage composition and therefore have the potential for use in early OA detection. Changes in cartilage mechanical properties are also hallmarks of OA cartilage; thus, measurement of cartilage mechanical properties may also be beneficial for earlier OA detection. However, the relative predictive ability of compositional versus mechanical properties in detecting OA has yet to be determined. Therefore, we developed logistic regression models predicting OA status in an ex vivo environment using several mechanical and compositional metrics to assess which metrics most effectively predict OA status. Specifically, in this study the compositional metric analyzed was the T1rho relaxation time, while the mechanical metrics analyzed were the stiffness and recovery (defined as a measure of how quickly cartilage returns to its original shape after loading) of the cartilage. Cartilage recovery had the best predictive ability of OA status both alone and in a multivariate model including the T1rho relaxation time. These findings highlight the potential of cartilage recovery as a non-invasive marker of in vivo cartilage health and motivate future investigation of this metric clinically.

Meniscus cell regional phenotypes: Dedifferentiation and reversal by biomaterial embedding.

Meniscus injuries are common and a major cause of long-term joint degeneration and disability. Current treatment options are limited, so novel regenerative therapies or tissue engineering strategies are urgently needed. The development of new therapies is hindered by a lack of knowledge regarding the cellular biology of the meniscus and a lack of well-established methods for studying meniscus cells in vitro. The goals of this study were to (1) establish baseline expression profiles and dedifferentiation patterns of inner and outer zone primary meniscus cells, and (2) evaluate the utility of poly(ethylene glycol) diacrylate (PEGDA) and gelatin methacrylate (GelMA) polymer hydrogels to reverse dedifferentiation trends for long-term meniscus cell culture. Using reverse transcription-quantitative polymerase chain reaction, we measured expression levels of putative meniscus phenotype marker genes in freshly isolated meniscus tissue, tissue explant culture, and monolayer culture of inner and outer zone meniscus cells from porcine knees to establish baseline dedifferentiation characteristics, and then compared these expression levels to PEGDA/GelMA embedded passaged meniscus cells. COL1A1 showed robust upregulation, while CHAD, CILP, and COMP showed downregulation with monolayer culture. Expression levels of COL2A1, ACAN, and SOX9 were surprisingly similar between inner and outer zone tissue and were found to be less sensitive as markers of dedifferentiation. When embedded in PEGDA/GelMA hydrogels, expression levels of meniscus cell phenotype genes were significantly modulated by varying the ratio of polymer components, allowing these materials to be tuned for phenotype restoration, meniscus cell culture, and tissue engineering applications.

Four-Point Bending Testing for Mechanical Assessment of Mouse Bone Structural Properties.

One of the primary functions of bone is to support the skeleton by withstanding load. In the diseased state, bone's ability to perform this function is altered. Quantification of the features of bone that support its functional behavior, and how they may change with disease, is accomplished through mechanical testing. As such, mechanical testing is a useful tool for scientists studying orthopedic-related diseases. Furthermore, a common animal model used to investigate disease and its treatment is the mouse. Therefore, in this chapter we (1) describe central concepts of mechanical testing, (2) describe factors that influence the mechanical behavior of bone, and (3) describe the application of a widely used mechanical testing technique, four-point bending, to the mouse bone for characterization of its structural properties.

The Characteristic Recovery Time as a Novel, Noninvasive Metric for Assessing In Vivo Cartilage Mechanical Function.

Osteoarthritis (OA) is a disease characterized by the degeneration of cartilage tissue, and is a leading cause of disability in the United States. The clinical diagnosis of OA includes the presence of pain and radiographic imaging findings, which typically do not present until advanced stages of the disease when treatment is difficult. Therefore, identifying new methods of OA detection that are sensitive to earlier pathological changes in cartilage, which may be addressed prior to the development of irreversible OA, is critical for improving OA treatment. A potentially promising avenue for developing early detection methods involves measuring the tissue's in vivo mechanical response to loading, as changes in mechanical function are commonly observed in ex vivo studies of early OA. However, thus far the mechanical function of cartilage has not been widely assessed in vivo. Therefore, the purpose of this study was to develop a novel methodology that can be used to measure an in vivo mechanical property of cartilage: the characteristic recovery time. Specifically, in this study we quantified the characteristic recovery time of cartilage thickness after exercise in relatively young subjects with asymptomatic cartilage. Additionally, we measured baseline cartilage thickness and T1rho and T2 relaxation times (quantitative MRI) prior to exercise in these subjects to assess whether baseline MRI measures are predictive of the characteristic recovery time, to understand whether or not the characteristic recovery time provides independent information about cartilage's mechanical state. Our results show that the mean recovery strain response across subjects was well-characterized by an exponential approach with a characteristic time of 25.2 min, similar to literature values of human characteristic times measured ex vivo. Further, we were unable to detect a statistically significant linear relationship between the characteristic recovery time and the baseline metrics measured here (T1rho relaxation time, T2 relaxation time, and cartilage thickness). This might suggest that the characteristic recovery time has the potential to provide additional information about the mechanical state of cartilage not captured by these baseline MRI metrics. Importantly, this study presents a noninvasive methodology for quantifying the characteristic recovery time, an in vivo mechanical property of cartilage. As mechanical response may be indicative of cartilage health, this study underscores the need for future studies investigating the characteristic recovery time and in vivo cartilage mechanical response at various stages of OA.

Comparison of Cartilage Mechanical Properties Measured During Creep and Recovery.

The diagnosis of osteoarthritis (OA) currently depends on the presence of pain and radiographic imaging findings, which generally do not present until later stages of the disease when the condition is difficult to treat. Therefore, earlier detection of OA pathology is needed for improved disease management. Ex vivo cartilage studies indicate that changes in the mechanical function of cartilage occur as degeneration progresses during OA. Thus, measurement of the in vivo cartilage mechanical response may serve as an earlier indicator of OA pathology. Though mechanical characterization is classically performed during loading, the unloading (recovery) response of cartilage may also enable determination of mechanical response. Therefore, the purpose of this study was to validate the use of the recovery response for mechanical characterization of cartilage in a controlled, ex vivo environment. To do so, confined compression creep and recovery tests were conducted on cartilage explants (N = 10), and the resulting mechanical properties from both the creep and recovery phases were compared. No statistically significant differences were found in the mechanical properties between the two phases, reinforcing the hypothesis that unloading (recovery) may be a good surrogate for loading.

The response of the pediatric head to impacts onto a rigid surface.

The study of pediatric head injury relies heavily on the use of finite element models and child anthropomorphic test devices (ATDs). However, these tools, in the context of pediatric head injury, have yet to be validated due to a paucity of pediatric head response data. The goal of this study is to investigate the response and injury tolerance of the pediatric head to impact. Twelve pediatric heads were impacted in a series of drop tests. The heads were dropped onto five impact locations (forehead, occiput, vertex and right and left parietal) from drop heights of 15 and 30 cm. The head could freely fall without rotation onto a flat 19 mm thick platen. The impact force was measured using a 3-axis piezoelectric load cell attached to the platen. Age and drop height were found to be significant factors in the impact response of the pediatric head. The head acceleration (14%-15 cm; 103-30 cm), Head Injury Criterion (HIC) (253%-15 cm; 154%-30 cm) and impact stiffness (5800%-15 cm; 3755%-30 cm) when averaged across all impact locations increased with age from 33 weeks gestation to 16 years, while the pulse duration (66%-15 cm; 53%-30 cm) decreased with age. Increases in head acceleration, HIC and impact stiffness were also observed with increased drop height, while pulse duration decreased with increased drop height. One important observation was that three of the four cadaveric heads between the ages of 5-months and 22-months sustained fractures from the 15 cm and 30 cm drop heights. The 5-month-old sustained a right parietal linear fracture while the 11- and 22-month-old sustained diastatic linear fractures.

In vivo assessment of the interaction of patellar tendon tibial shaft angle and anterior cruciate ligament elongation during flexion.

A potential cause of non-contact anterior cruciate ligament (ACL) injury is landing on an extended knee. In line with this hypothesis, studies have shown that the ACL is elongated with decreasing knee flexion angle. Furthermore, at low flexion angles the patellar tendon is oriented to increase the anterior shear component of force acting on the tibia. This indicates that knee extension represents a position in which the ACL is taut, and thus may have an increased propensity for injury, particularly in the presence of excessive force acting via the patellar tendon. However, there is very little in vivo data to describe how patellar tendon orientation and ACL elongation interact during flexion. Therefore, this study measured the patellar tendon tibial shaft angle (indicative of the relative magnitude of the shear component of force acting via the patellar tendon) and ACL length in vivo as subjects performed a quasi-static lunge at varying knee flexion angles. Spearman rho rank correlations within each individual revealed that flexion angles were inversely correlated to both ACL length (rho = -0.94 ±â€¯0.07, mean ±â€¯standard deviation, p < 0.05) and patellar tendon tibial shaft angle (rho = -0.99 ±â€¯0.01, p < 0.05). These findings indicate that when the knee is extended, the ACL is both elongated and the patellar tendon tibial shaft angle is increased, resulting in a relative increase in anterior shear force on the tibia acting via the patellar tendon. Therefore, these data support the hypothesis that landing with the knee in extension is a high risk scenario for ACL injury.

A Comparison of Knee Abduction Angles Measured by a 3D Anatomic Coordinate System Versus Videographic Analysis: Implications for Anterior Cruciate Ligament Injury.

BACKGROUND: Knee positions involved in noncontact anterior cruciate ligament (ACL) injury have been studied via analysis of injury videos. Positions of high ACL strain have been identified in vivo. These methods have supported different hypotheses regarding the role of knee abduction in ACL injury. PURPOSE/HYPOTHESIS: The purpose of this study was to compare knee abduction angles measured by 2 methods: using a 3-dimensional (3D) coordinate system based on anatomic features of the bones versus simulated 2-dimensional (2D) videographic analysis. We hypothesized that knee abduction angles measured in a 2D videographic analysis would differ from those measured from 3D bone anatomic features and that videographic knee abduction angles would depend on flexion angle and on the position of the camera relative to the patient. STUDY DESIGN: Descriptive laboratory study. METHODS: Models of the femur and tibia were created from magnetic resonance images of 8 healthy male participants. The models were positioned to match biplanar fluoroscopic images obtained as participants posed in lunges of varying flexion angles (FLAs). Knee abduction angle was calculated from the positioned models in 2 ways: (1) varus-valgus angle (VVA), defined as the angle between the long axis of the tibia and the femoral transepicondylar axis by use of a 3D anatomic coordinate system; and (2) coronal plane angle (CPA), defined as the angle between the long axis of the tibia and the long axis of the femur projected onto the tibial coronal plane to simulate a 2D videographic analysis. We then simulated how changing the position of the camera relative to the participant would affect knee abduction angles. RESULTS: During flexion, when CPA was calculated from a purely anterior or posterior view of the joint-an ideal scenario for measuring knee abduction from 2D videographic analysis-CPA was significantly different from VVA (P < .0001). CPA also varied substantially with the position of the camera relative to the participant. CONCLUSION: How closely CPA (derived from 2D videographic analysis) relates to VVA (derived from a 3D anatomic coordinate system) depends on FLA and camera orientation. CLINICAL RELEVANCE: This study provides a novel comparison of knee abduction angles measured from 2D videographic analysis and those measured within a 3D anatomic coordinate system. Consideration of these findings is important when interpreting 2D videographic data regarding knee abduction angle in ACL injury.

A New Stress Test for Knee Joint Cartilage.

Cartilage metabolism-both the synthesis and breakdown of cartilage constituents and architecture-is influenced by its mechanical loading. Therefore, physical activity is often recommended to maintain cartilage health and to treat or slow the progression of osteoarthritis, a debilitating joint disease causing cartilage degeneration. However, the appropriate exercise frequency, intensity, and duration cannot be prescribed because direct in vivo evaluation of cartilage following exercise has not yet been performed. To address this gap in knowledge, we developed a cartilage stress test to measure the in vivo strain response of healthy human subjects' tibial cartilage to walking exercise. We varied both walk duration and speed in a dose-dependent manner to quantify how these variables affect cartilage strain. We found a nonlinear relationship between walk duration and in vivo compressive strain, with compressive strain initially increasing with increasing duration, then leveling off with longer durations. This work provides innovative measurements of cartilage creep behavior (which has been well-documented in vitro but not in vivo) during walking. This study showed that compressive strain increased with increasing walking speed for the speeds tested in this study (0.9-2.0 m/s). Furthermore, our data provide novel measurements of the in vivo strain response of tibial cartilage to various doses of walking as a mechanical stimulus, with maximal strains of 5.0% observed after 60 minutes of walking. These data describe physiological benchmarks for healthy articular cartilage behavior during walking and provide a much-needed baseline for studies investigating the effect of exercise on cartilage health.

Effects of Anterior Cruciate Ligament Deficiency on Tibiofemoral Cartilage Thickness and Strains in Response to Hopping.

BACKGROUND: Changes in knee kinematics after anterior cruciate ligament (ACL) injury may alter loading of the cartilage and thus affect its homeostasis, potentially leading to the development of posttraumatic osteoarthritis. However, there are limited in vivo data to characterize local changes in cartilage thickness and strain in response to dynamic activity among patients with ACL deficiency. PURPOSE/HYPOTHESIS: The purpose was to compare in vivo tibiofemoral cartilage thickness and cartilage strain resulting from dynamic activity between ACL-deficient and intact contralateral knees. It was hypothesized that ACL-deficient knees would show localized reductions in cartilage thickness and elevated cartilage strains. STUDY DESIGN: Controlled laboratory study. METHODS: Magnetic resonance images were obtained before and after single-legged hopping on injured and uninjured knees among 8 patients with unilateral ACL rupture. Three-dimensional models of the bones and articular surfaces were created from the pre- and postactivity scans. The pre- and postactivity models were registered to each other, and cartilage strain (defined as the normalized difference in cartilage thickness pre- and postactivity) was calculated in regions across the tibial plateau, femoral condyles, and femoral cartilage adjacent to the medial intercondylar notch. These measurements were compared between ACL-deficient and intact knees. Differences in cartilage thickness and strain between knees were tested with multiple analysis of variance models with alpha set at P < .05. RESULTS: Compressive strain in the intercondylar notch was elevated in the ACL-deficient knee relative to the uninjured knee. Furthermore, cartilage in the intercondylar notch and adjacent medial tibia was significantly thinner before activity in the ACL-deficient knee versus the intact knee. In these 2 regions, thinning was significantly influenced by time since injury, with patients with more chronic ACL deficiency (>1 year since injury) experiencing greater thinning. CONCLUSION: Among patients with ACL deficiency, the medial femoral condyle adjacent to the intercondylar notch in the ACL-deficient knee exhibited elevated cartilage strain and loss of cartilage thickness, particularly with longer time from injury. It is hypothesized that these changes may be related to posttraumatic osteoarthritis development. CLINICAL RELEVANCE: This study suggests that altered mechanical loading is related to localized cartilage thinning after ACL injury.

Obesity alters the in vivo mechanical response and biochemical properties of cartilage as measured by MRI.

BACKGROUND: Obesity is a primary risk factor for the development of knee osteoarthritis (OA). However, there remains a lack of in vivo data on the influence of obesity on knee cartilage mechanics and composition. The purpose of this study was to determine the relationship between obesity and tibiofemoral cartilage properties. METHODS: Magnetic resonance images (3T) of cartilage geometry (double-echo steady-state) and T1rho relaxation of the knee were obtained in healthy subjects with a normal (n = 8) or high (n = 7) body mass index (BMI) before and immediately after treadmill walking. Subjects had no history of lower limb injury or surgery. Bone and cartilage surfaces were segmented and three-dimensional models were created to measure cartilage thickness and strain. T1rho relaxation times were measured before exercise in both the tibial and femoral cartilage in order to characterize biochemical composition. Body fat composition was also measured. RESULTS: Subjects with a high BMI exhibited significantly increased tibiofemoral cartilage strain and T1rho relaxation times (P <0.05). Tibial pre-exercise cartilage thickness was also affected by BMI (P <0.05). Correlational analyses revealed that pre-exercise tibial cartilage thickness decreased with increasing BMI (R2 = 0.43, P <0.01) and body fat percentage (R2 = 0.58, P <0.01). Tibial and femoral cartilage strain increased with increasing BMI (R2 = 0.45, P <0.01; R2 = 0.51, P <0.01, respectively) and increasing body fat percentage (R2 = 0.40, P <0.05; R2 = 0.38, P <0.05, respectively). Additionally, tibial T1rho was positively correlated with BMI (R2 = 0.39, P <0.05) and body fat percentage (R2 = 0.47, P <0.01). CONCLUSIONS: Strains and T1rho relaxation times in the tibiofemoral cartilage were increased in high BMI subjects compared with normal BMI subjects. Additionally, pre-exercise tibial cartilage thickness decreased with obesity. Reduced proteoglycan content may be indicative of pre-symptomatic osteoarthritic degeneration, resulting in reduced cartilage thickness and increased deformation of cartilage in response to loading.

Foul tip impact attenuation of baseball catcher masks using head impact metrics.

Currently, no scientific consensus exists on the relative safety of catcher mask styles and materials. Due to differences in mass and material properties, the style and material of a catcher mask influences the impact metrics observed during simulated foul ball impacts. The catcher surrogate was a Hybrid III head and neck equipped with a six degree of freedom sensor package to obtain linear accelerations and angular rates. Four mask styles were impacted using an air cannon for six 30 m/s and six 35 m/s impacts to the nasion. To quantify impact severity, the metrics peak linear acceleration, peak angular acceleration, Head Injury Criterion, Head Impact Power, and Gadd Severity Index were used. An Analysis of Covariance and a Tukey's HSD Test were conducted to compare the least squares mean between masks for each head injury metric. For each injury metric a P-Value less than 0.05 was found indicating a significant difference in mask performance. Tukey's HSD test found for each metric, the traditional style titanium mask fell in the lowest performance category while the hockey style mask was in the highest performance category. Limitations of this study prevented a direct correlation from mask testing performance to mild traumatic brain injury.

A comparison of patellofemoral cartilage morphology and deformation in anterior cruciate ligament deficient versus uninjured knees.

Anterior cruciate ligament (ACL) deficient patients have an increased rate of patellofemoral joint (PFJ) osteoarthritis (OA) as compared to the general population. Although the cause of post-injury OA is multi-factorial, alterations in joint biomechanics may predispose patients to cartilage degeneration. This study aimed to compare in vivo PFJ morphology and mechanics between ACL deficient and intact knees in subjects with unilateral ACL ruptures. Eight male subjects underwent baseline MRI scans of both knees. They then performed a series of 60 single-legged hops, followed by a post-exercise MRI scan. This process was repeated for the contralateral knee. The MR images were converted into three-dimensional surface models of cartilage and bone in order to assess cartilage thickness distributions and strain following exercise. Prior to exercise, patellar cartilage was significantly thicker in intact knees as compared to ACL deficient knees by 1.8%. In response to exercise, we observed average patellar cartilage strains of 5.4 ±â€¯1.1% and 2.5 ±â€¯1.4% in the ACL deficient and intact knees, respectively. Importantly, the magnitude of patellar cartilage strain in the ACL deficient knees was significantly higher than in the intact knees. However, while trochlear cartilage experienced a mean strain of 2.4 ±â€¯1.6%, there was no difference in trochlear cartilage strain between the ACL deficient and uninjured knees. In summary, we found that ACL deficiency was associated with decreased patellar cartilage thickness and increased exercise-induced patellar cartilage strain when compared to the uninjured contralateral knees.

Effect of the mandible on mouthguard measurements of head kinematics.

Wearable sensors are becoming increasingly popular for measuring head motions and detecting head impacts. Many sensors are worn on the skin or in headgear and can suffer from motion artifacts introduced by the compliance of soft tissue or decoupling of headgear from the skull. The instrumented mouthguard is designed to couple directly to the upper dentition, which is made of hard enamel and anchored in a bony socket by stiff ligaments. This gives the mouthguard superior coupling to the skull compared with other systems. However, multiple validation studies have yielded conflicting results with respect to the mouthguard׳s head kinematics measurement accuracy. Here, we demonstrate that imposing different constraints on the mandible (lower jaw) can alter mouthguard kinematic accuracy in dummy headform testing. In addition, post mortem human surrogate tests utilizing the worst-case unconstrained mandible condition yield 40% and 80% normalized root mean square error in angular velocity and angular acceleration respectively. These errors can be modeled using a simple spring-mass system in which the soft mouthguard material near the sensors acts as a spring and the mandible as a mass. However, the mouthguard can be designed to mitigate these disturbances by isolating sensors from mandible loads, improving accuracy to below 15% normalized root mean square error in all kinematic measures. Thus, while current mouthguards would suffer from measurement errors in the worst-case unconstrained mandible condition, future mouthguards should be designed to account for these disturbances and future validation testing should include unconstrained mandibles to ensure proper accuracy.

Injuries of the head from backface deformation of ballistic protective helmets under ballistic impact.

Modern ballistic helmets defeat penetrating bullets by energy transfer from the projectile to the helmet, producing helmet deformation. This deformation may cause severe injuries without completely perforating the helmet, termed "behind armor blunt trauma" (BABT). As helmets become lighter, the likelihood of larger helmet backface deformation under ballistic impact increases. To characterize the potential for BABT, seven postmortem human head/neck specimens wearing a ballistic protective helmet were exposed to nonperforating impact, using a 9 mm, full metal jacket, 124 grain bullet with velocities of 400-460 m/s. An increasing trend of injury severity was observed, ranging from simple linear fractures to combinations of linear and depressed fractures. Overall, the ability to identify skull fractures resulting from BABT can be used in forensic investigations. Our results demonstrate a high risk of skull fracture due to BABT and necessitate the prevention of BABT as a design factor in future generations of protective gear.

Pediatric head and neck dynamics in frontal impact: analysis of important mechanical factors and proposed neck performance corridors for 6- and 10-year-old ATDs.

OBJECTIVE: Traumatic injuries are the leading cause of death of children aged 1-19 in the United States and are principally caused by motor vehicle collisions, with the head being the primary region injured. The neck, though not commonly injured, governs head kinematics and thus influences head injury. Vehicle improvements necessary to reduce these injuries are evaluated using anthropomorphic testing devices (ATDs). Current pediatric ATD head and neck properties were established by scaling adult properties using the size differences between adults and children. Due to the limitations of pediatric biomechanical research, computational models are the only available methods that combine all existing data to produce injury-relevant biofidelity specifications for ATDs. The purpose of this study is to provide the first frontal impact biofidelity corridors for neck flexion response of 6- and 10-year-olds using validated computational models, which are compared to the Hybrid III (HIII) ATD neck responses and the Mertz flexion corridors. METHODS: Our virtual 6- and 10-year-old head and neck multibody models incorporate pediatric biomechanical properties obtained from pediatric cadaveric and radiological studies, include the effect of passive and active musculature, and are validated with data including pediatric volunteer 3 g dynamic frontal impact responses. We simulate ATD pendulum tests-used to calibrate HIII neck bending stiffness-to compare the pediatric model and HIII ATD neck bending stiffness and to compare the model flexion bending responses with the Mertz scaled neck flexion corridors. Additionally, pediatric response corridors for pendulum calibration tests and high-speed (15 g) frontal impacts are estimated through uncertainty analyses on primary model variables, with response corridors calculated from the average ± SD response over 650 simulations. RESULTS AND CONCLUSIONS: The models are less stiff in dynamic anterioposterior bending than the ATDs; the secant stiffness of the 6- and 10-year-old models is 53 and 67 percent less than that of the HIII ATDs. The ATDs exhibit nonlinear stiffening and the models demonstrate nonlinear softening. Consequently, the models do not remain within the Mertz scaled flexion bending corridors. The more compliant model necks suggest an increased potential for head impact via larger head excursions. The pediatric anterioposterior bending corridors developed in this study are extensible to any frontal loading condition through calculation and sensitivity analysis. The corridors presented in this study are the first based on pediatric cadaveric data and provide the basis for future, more biofidelic, designs of 6- and 10-year-old ATD necks.

Importance of muscle activations for biofidelic pediatric neck response in computational models.

OBJECTIVE: During dynamic injury scenarios, such as motor vehicle crashes, neck biomechanics contribute to head excursion and acceleration, influencing head injuries. One important tool in understanding head and neck dynamics is computational modeling. However, realistic and stable muscle activations for major muscles are required to realize meaningful kinematic responses. The objective was to determine cervical muscle activation states for 6-year-old, 10-year-old, and adult 50th percentile male computational head and neck models. Currently, pediatric models including muscle activations are unable to maintain the head in an equilibrium position, forcing models to begin from nonphysiologic conditions. Recent work has realized a stationary initial geometry and cervical muscle activations by first optimizing responses against gravity. Accordingly, our goal was to apply these methods to Duke University's head-neck model validated using living muscle response and pediatric cadaveric data. METHODS: Activation schemes maintaining an upright, stable head for 22 muscle pairs were found using LS-OPT. Two optimization problems were investigated: a relaxed state, which minimized muscle fatigue, and a tensed activation state, which maximized total muscle force. The model's biofidelity was evaluated by the kinematic response to gravitational and frontal impact loading conditions. Model sensitivity and uncertainty analyses were performed to assess important parameters for pediatric muscle response. Sensitivity analysis was conducted using multiple activation time histories. These included constant activations and an optimal muscle activation time history, which varied the activation level of flexor and extensor groups, and activation initiation and termination times. RESULTS: Relaxed muscle activations decreased with increasing age, maintaining upright posture primarily through extensor activation. Tensed musculature maintained upright posture through coactivation of flexors and extensors, producing up to 32 times the force of the relaxed state. Without muscle activation, the models fell into flexion due to gravitational loading. Relaxed musculature produced 28.6-35.8 N of force to the head, whereas tensed musculature produced 450-1023 N. Pediatric model stiffnesses were most sensitive to muscle physiological cross-sectional area. CONCLUSIONS: Though muscular loads were not large enough to cause vertebral compressive failure, they would provide a prestressed state that could protect the vertebrae during tensile loading but might exacerbate risk during compressive loading. For example, in the 10-year-old, a load of 602 N was produced, though estimated compressive failure tolerance is only 2.8 kN. Including muscles and time-variant activation schemes is vital for producing biofidelic models because both vary by age. The pediatric activations developed represent physiologically appropriate sets of initial conditions and are based on validated adult cadaveric data.

How few? Bayesian statistics in injury biomechanics.

In injury biomechanics, there are currently no general a priori estimates of how few specimens are necessary to obtain sufficiently accurate injury risk curves for a given underlying distribution. Further, several methods are available for constructing these curves, and recent methods include Bayesian survival analysis. This study used statistical simulations to evaluate the fidelity of different injury risk methods using limited sample sizes across four different underlying distributions. Five risk curve techniques were evaluated, including Bayesian techniques. For the Bayesian analyses, various prior distributions were assessed, each incorporating more accurate information. Simulated subject injury and biomechanical input values were randomly sampled from each underlying distribution, and injury status was determined by comparing these values. Injury risk curves were developed for this data using each technique for various small sample sizes; for each, analyses on 2000 simulated data sets were performed. Resulting median predicted risk values and confidence intervals were compared with the underlying distributions. Across conditions, the standard and Bayesian survival analyses better represented the underlying distributions included in this study, especially for extreme (1, 10, and 90%) risk. This study demonstrates that the value of the Bayesian analysis is the use of informed priors. As the mean of the prior approaches the actual value, the sample size necessary for good reproduction of the underlying distribution with small confidence intervals can be as small as 2. This study provides estimates of confidence intervals and number of samples to allow the selection of the most appropriate sample sizes given known information.