Failed rib region prediction in a human body model during crash events with precrash braking.

OBJECTIVE: The objective of this study is 2-fold. We used a validated human body finite element model to study the predicted chest injury (focusing on rib fracture as a function of element strain) based on varying levels of simulated precrash braking. Furthermore, we compare deterministic and probabilistic methods of rib injury prediction in the computational model. METHODS: The Global Human Body Models Consortium (GHBMC) M50-O model was gravity settled in the driver position of a generic interior equipped with an advanced 3-point belt and airbag. Twelve cases were investigated with permutations for failure, precrash braking system, and crash severity. The severities used were median (17 kph), severe (34 kph), and New Car Assessment Program (NCAP; 56.4 kph). Cases with failure enabled removed rib cortical bone elements once 1.8% effective plastic strain was exceeded. Alternatively, a probabilistic framework found in the literature was used to predict rib failure. Both the probabilistic and deterministic methods take into consideration location (anterior, lateral, and posterior). The deterministic method is based on a rubric that defines failed rib regions dependent on a threshold for contiguous failed elements. The probabilistic method depends on age-based strain and failure functions. RESULTS: Kinematics between both methods were similar (peak max deviation: ΔXhead = 17 mm; ΔZhead = 4 mm; ΔXthorax = 5 mm; ΔZthorax = 1 mm). Seat belt forces at the time of probabilistic failed region initiation were lower than those at deterministic failed region initiation. The probabilistic method for rib fracture predicted more failed regions in the rib (an analog for fracture) than the deterministic method in all but 1 case where they were equal. The failed region patterns between models are similar; however, there are differences that arise due to stress reduced from element elimination that cause probabilistic failed regions to continue to rise after no deterministic failed region would be predicted. CONCLUSIONS: Both the probabilistic and deterministic methods indicate similar trends with regards to the effect of precrash braking; however, there are tradeoffs. The deterministic failed region method is more spatially sensitive to failure and is more sensitive to belt loads. The probabilistic failed region method allows for increased capability in postprocessing with respect to age. The probabilistic failed region method predicted more failed regions than the deterministic failed region method due to force distribution differences.

A preliminary study of human model head and neck response to frontal loading in nontraditional occupant seating configurations.

OBJECTIVE: Computational human body models (HBMs) are nominally omnidirectional surrogates given their structural basis in human anatomy. As a result, such models are well suited for studies related to occupant safety in anticipated highly automated vehicles (HAVs). We utilize a well-validated HBM to study the head and neck kinematics in simulations of nontraditional occupant seating configurations. METHODS: The GHBMC M50-O v. 4.4 HBM was gravity settled into a generic seat buck and situated in a seated posture. The model was simulated in angular increments of 15 degrees clockwise from forward facing to rear facing. A pulse of 17.0 kph (NASS median) was used in each to simulate a frontal impact for each of the 13 seating configurations. Belt anchor points were rotated with the seat; the airbag was appropriately powered based on delta-V, and was not used in rear-facing orientations. Neck forces and moments were calculated. RESULTS: The 30-degree oblique case was found to result in the maximum neck load and sagittal moment, and thus Neck Injury Criteria (NIJ). Neck loads were minimized in the rear facing condition. The moments and loads, however, were greatest in the lateral seating configuration for these frontal crash simulations. CONCLUSIONS: In a recent policy statement on HAVs, the NHTSA indicated that vehicle manufacturers will be expected to provide countermeasures that will fully protect occupants given any planned seating or interior configurations. Furthermore, the agency indicated that virtual tests using human models could be used to demonstrate such efficacy. While the results presented are only appropriate for comparison within this study, they do indicate that human models provide reasonable biomechanical data for nontraditional occupant seating arrangements.

A method to measure predictive ability of an injury risk curve using an observation-adjusted area under the receiver operating characteristic curve.

Area under the receiver operating characteristic curve (AROC) is commonly used to choose a biomechanical metric from which to construct an injury risk curve (IRC). However, AROC may not handle censored datasets adequately. Survival analysis creates robust estimates of IRCs which accommodate censored data. We present an observation-adjusted ROC (oaROC) which uses the survival-based IRC to estimate the AROC. We verified and evaluated this method using simulated datasets of different censoring statuses and sample sizes. For a dataset with 1000 left and right censored observations, the median AROC closely approached the oaROCTrue, or the oaROC calculated using an assumed "true" IRC, differing by a fraction of a percent, 0.1%. Using simulated datasets with various censoring, we found that oaROC converged onto oaROCTrue in all cases. For datasets with right and non-censored observations, AROC did not converge onto oaROCTrue. oaROC for datasets with only non-censored observations converged the fastest, and for a dataset with 10 observations, the median oaROC differed from oaROCTrue by 2.74% while the corresponding median AROC with left and right censored data differed from oaROCTrue by 9.74%. We also calculated the AROC and oaROC for a published side impact dataset, and differences between the two methods ranged between -24.08% and 24.55% depending on metric. Overall, when compared with AROC, we found oaROC performs equivalently for doubly censored data, better for non-censored data, and can accommodate more types of data than AROC. While more validation is needed, the results indicate that oaROC is a viable alternative which can be incorporated into the metric selection process for IRCs.

The effect of precrash velocity reduction on occupant response using a human body finite element model.

OBJECTIVE: The objective of this study is to use a validated finite element model of the human body and a certified model of an anthropomorphic test dummy (ATD) to evaluate the effect of simulated precrash braking on driver kinematics, restraint loads, body loads, and computed injury criteria in 4 commonly injured body regions. METHODS: The Global Human Body Models Consortium (GHBMC) 50th percentile male occupant (M50-O) and the Humanetics Hybrid III 50th percentile models were gravity settled in the driver position of a generic interior equipped with an advanced 3-point belt and driver airbag. Fifteen simulations per model (30 total) were conducted, including 4 scenarios at 3 severity levels: median, severe, and the U.S. New Car Assessment Program (U.S.-NCAP) and 3 extra per model with high-intensity braking. The 4 scenarios were no precollision system (no PCS), forward collision warning (FCW), FCW with prebraking assist (FCW+PBA), and FCW and PBA with autonomous precrash braking (FCW + PBA + PB). The baseline ΔV was 17, 34, and 56.4 kph for median, severe, and U.S.-NCAP scenarios, respectively, and were based on crash reconstructions from NASS/CDS. Pulses were then developed based on the assumed precrash systems equipped. Restraint properties and the generic pulse used were based on literature. RESULTS: In median crash severity cases, little to no risk (<10% risk for Abbreviated injury Scale [AIS] 3+) was found for all injury measures for both models. In the severe set of cases, little to no risk for AIS 3+ injury was also found for all injury measures. In NCAP cases, highest risk was typically found with No PCS and lowest with FCW + PBA + PB. In the higher intensity braking cases (1.0-1.4 g), head injury criterion (HIC), brain injury criterion (BrIC), and chest deflection injury measures increased with increased braking intensity. All other measures for these cases tended to decrease. The ATD also predicted and trended similar to the human body models predictions for both the median, severe, and NCAP cases. Forward excursion for both models decreased across median, severe, and NCAP cases and diverged from each other in cases above 1.0 g of braking intensity. CONCLUSIONS: The addition of precrash systems simulated through reduced precrash speeds caused reductions in some injury criteria, whereas others (chest deflection, HIC, and BrIC) increased due to a modified occupant position. The human model and ATD models trended similarly in nearly all cases with greater risk indicated in the human model. These results suggest the need for integrated safety systems that have restraints that optimize the occupant's position during precrash braking and prior to impact.

A method for developing biomechanical response corridors based on principal component analysis.

The standard method for specifying target responses for human surrogates, such as crash test dummies and human computational models, involves developing a corridor based on the distribution of a set of empirical mechanical responses. These responses are commonly normalized to account for the effects of subject body shape, size, and mass on impact response. Limitations of this method arise from the normalization techniques, which are based on the assumptions that human geometry linearly scales with size and in some cases, on simple mechanical models. To address these limitations, a new method was developed for corridor generation that applies principal component (PC) analysis to align response histories. Rather than use normalization techniques to account for the effects of subject size on impact response, linear regression models are used to model the relationship between PC features and subject characteristics. Corridors are generated using Monte Carlo simulation based on estimated distributions of PC features for each PC. This method is applied to pelvis impact force data from a recent series of lateral impact tests to develop corridor bounds for a group of signals associated with a particular subject size. Comparing to the two most common methods for response normalization, the corridors generated by the new method are narrower and better retain the features in signals that are related to subject size and body shape.

A point-wise normalization method for development of biofidelity response corridors.

An updated technique to develop biofidelity response corridors (BRCs) is presented. BRCs provide a representative range of time-dependent responses from multiple experimental tests of a parameter from multiple biological surrogates (often cadaveric). The study describes an approach for BRC development based on previous research, but that includes two key modifications for application to impact and accelerative loading. First, signal alignment conducted prior to calculation of the BRC considers only the loading portion of the signal, as opposed to the full time history. Second, a point-wise normalization (PWN) technique is introduced to calculate correlation coefficients between signals. The PWN equally weighs all time points within the loading portion of the signals and as such, bypasses aspects of the response that are not controlled by the experimentalist such as internal dynamics of the specimen, and interaction with surrounding structures. An application of the method is presented using previously-published thoracic loading data from 8 lateral sled PMHS tests conducted at 8.9m/s. Using this method, the mean signals showed a peak lateral load of 8.48kN and peak chest acceleration of 86.0g which were similar to previously-published research (8.93kN and 100.0g respectively). The peaks occurred at similar times in the current and previous studies, but were delayed an average of 2.1ms in the updated method. The mean time shifts calculated with the method ranged from 7.5% to 9.5% of the event. The method may be of use in traditional injury biomechanics studies and emerging work on non-horizontal accelerative loading.

A paradigm for human body finite element model integration from a set of regional models.

Computational modeling offers versatility, scalability, and cost advantages to researchers in the trauma and injury biomechanics communities. The Global Human Body Models Consortium (GHBMC) is a group of government, industry, and academic researchers developing human body models (HBMs) that aim to become the standard tool to meet this growing research need. The objective of this study is to present the methods used to develop the average seated male occupant model (M50, weight = 78 kg, height = 175 cm) from five separately validated body region models (BRMs). BRMs include the head, neck, thorax, abdomen, and a combined pelvis and lower extremity model. Modeling domains were split at the atlanto-occipital joint, C7-T1 boundary, diaphragm, abdominal cavity (peritoneum/retroperitoneum), and the acetabulum respectively. BRM meshes are based on a custom CAD model of the seated male built from a multi-modality imaging protocol of a volunteer subject found in literature.[1] Various meshing techniques were used to integrate the full body model (FBM) including 1-D beam and discrete element connections (e.g. ligamentous structures), 2D shell nodal connections (e.g. inferior vena cava to right atrium), 3D hexahedral nodal connections (e.g. soft tissue envelope connections between regions), and contact definitions varying from tied (muscle insertions) to sliding (liver and diaphragm contact). The model was developed in a general-purpose finite element code, LS-Dyna (LTSC, Livermore, CA) R4.2.1., and consists of 1.95 million elements and 1.3 million nodes. The element breakdown by type is 41% hexahedral, 33.7% tetrahedral, 19.5% quad shells and 5% tria shell. The integration methodology presented highlights the viability of using a collaborative development paradigm for the construction of HBMs, and will be used as template for expanding the suite of GHBMC models.

External landmark, body surface, and volume data of a mid-sized male in seated and standing postures.

The purpose of this study was to acquire external landmark, undeformed surface, and volume data from a pre-screened individual representing a mid-sized male (height 174.9 cm, weight 78.6 ± 0.77 kg) in the seated and standing postures. The individual matched the 50th percentile value of 15 measures of external anthropometry from previous anthropometric studies with an average deviation of 3%. As part of a related study, a comprehensive full body medical image data set was acquired from the same individual on whom landmark data were collected. Three dimensional bone renderings from this data were used to visually verify the landmark and surface results. A total of 54 landmarks and external surface data were collected using a 7-axis digitizer. A seat buck designed in-house with removable back and seat pan panels enabled collection of undeformed surface contours of the back, buttocks, and posterior thigh. Eight metrics describing the buck positioning are provided. A repeatability study was conducted with three trials to assess intra-observer variability. Total volume and surface area of the seated model were found to be 75.8 × 10(3) cm(3) and 18.6 × 10(3) cm(2) and match the volume and surface area of the standing posture within 1%. Root mean squared error values from the repeatability study were on average 5.9 and 6.6 mm for the seated and standing postures respectively. The peak RMS error as a percentage of the centroid size of the landmark data sets were 3% for both the seated and standing trials. The data were collected as part of a global program on the development of an advanced human body model for blunt injury simulation. In addition, the reported data can be used for many diverse applications of biomechanical research such as ergonomics and morphometrics studies.

Development of a full body CAD dataset for computational modeling: a multi-modality approach.

The objective of this study was to develop full body CAD geometry of a seated 50th percentile male. Model development was based on medical image data acquired for this study, in conjunction with extensive data from the open literature. An individual (height, 174.9 cm, weight, 78.6 ± 0.77 kg, and age 26 years) was enrolled in the study for a period of 4 months. 72 scans across three imaging modalities (CT, MRI, and upright MRI) were collected. The whole-body dataset contains 15,622 images. Over 300 individual components representing human anatomy were generated through segmentation. While the enrolled individual served as a template, segmented data were verified against, or augmented with, data from over 75 literature sources on the average morphology of the human body. Non-Uniform Rational B-Spline (NURBS) surfaces with tangential (G1) continuity were constructed over all the segmented data. The sagittally symmetric model consists of 418 individual components representing bones, muscles, organs, blood vessels, ligaments, tendons, cartilaginous structures, and skin. Length, surface area, and volumes of components germane to crash injury prediction are presented. The total volume (75.7 × 103 cm(3)) and surface area (1.86 × 102 cm(2)) of the model closely agree with the literature data. The geometry is intended for subsequent use in nonlinear dynamics solvers, and serves as the foundation of a global effort to develop the next-generation computational human body model for injury prediction and prevention.