Athletic environments, American football in particular, have long been viewed as a potentially rich test-bed for understanding concussions. Over the last decade, two methods have been employed to obtain head acceleration measures occurring from actual impact events. Laboratory reconstructions utilize precision measurement equipment to recreate on-field impacts but are limited by cost, technical expertise, and the necessary assumptions of the human surrogate models. The HIT System is an on-field measurement system that allows for large-scale data collection by actively measuring head acceleration of athletes during play. Given that the current knowledge base has been developed using both techniques, the current study was required to establish the relationship between these two measurement methodologies. From these laboratory tests, a relationship has been established between the two systems, particularly when considering peak metrics such as linear and rotational acceleration for distributions of impacts; however, it is important for users of HIT System technology to understand the practical limitation of ensuring proper fit of a player’s helmet. This limitation could result in error for single impacts that is similar to those previously reported for laboratory reconstructions. Results from this study indicate that measurements from the two methods of study are correlated and provide a link that can be used to better interpret findings from future study using either technology.

In general, though, the relationship found through regression suggests distributions of measurements obtained on-the field with the HIT System will be nearly identical to those obtained in the laboratory as long as the laboratory tests reflect field conditions. This relationship appears to be strong even for distributions of relatively few data points, as shown by the 54 impacts evaluated in this study.

This relationship, though, should only be assumed for conditions included in the test distribution. For example, considering the A, B, C, and D impact sites and associated test velocities were previously found to be representative of concussion in the NFL, it would be expected from these test results that HIII measures obtained by Pellman et al. would closely match those recorded by the HIT System for concussive injuries when compared as a group. This close relationship has been confirmed in several studies

We chose to use regression analysis as the primary means for correlating output between the HIT System and HIII. One of the primary benefits of the HIT System is its ability to record large amounts of in vivo impact data that are inclusive of all on-field scenarios that can be experienced by different playing positions and skill levels.5,11,12,27,42,44 It can reasonably be assumed that each of the 286,636 impacts reported in the previously described study by Crisco et al. represents a specific combination of input conditions that could influence the player’s head kinematics. Because of this variability in conditions, researchers have primarily chosen to present data from the HIT System in the form of impact distributions inclusive of a variety of impact conditions (e.g., impact location, impact severity, playing position, etc.).

Mihalik et al. have reported statistically lower average peak linear acceleration for impacts to the frontal region of the head than to the top.

The results from this study predict that impacts to the A′ and A′′ tests measured by the instrumented helmet will result in much higher accelerations (e.g., 2–5 times the actual value) for a given impact velocity when compared to other impact locations. It is important to note, however, that, independent of the large differences in acceleration magnitudes recorded between the systems at A′ and A′′, the impact location was closely identified. While impacts will inevitably occur on the field at these sites, we do not believe that the A′ and A′′ sites and, in some instances, the A site, as currently tested using the linear impactor represent realistic cases of helmet impact and helmet motion that occurs on the football field

The unknown parameters of head mass, head moment of inertia, and the point of rotation about the neck are determined by the relationship of peak linear and rotational acceleration of on-field impacts obtained from a similar device paired with a full six degree of freedom processing algorithm.

Researchers have long sought to determine the relationship between head kinematics following impact and the pathophysiology of traumatic brain injury. One approach to understanding this relationship for mTBI has been to treat the football playing field, where athletes may sustain more than 2000 impacts to the head during a season,11, 12, 13 as a living laboratory. Specifically, researchers have monitored the impacts sustained by football players and have attempted to correlate their impact exposure with signs and symptoms of injury. Over the last decade, two primary approaches have been implemented for quantifying head kinematics following impacts experienced during play: laboratory reconstruction of impacts recorded on video and on-field measurement using instrumented helmets. The aim of this study was to correlate HIT System output with Hybrid III ATD output in a laboratory test using a linear impactor system under realistic input conditions.

Coefficients of determination were indeterminate due to poor correlation

Linear acceleration time series data from both the HIT System and HIII were processed in real time to obtain the resultant linear acceleration for each impact. Because a simultaneous trigger was not feasible, the linear acceleration resultants were synchronized post-processing by minimizing the RMS error between resultants.1,41 HIII data were then truncated to a 40-ms time window to allow direct time series comparison with the instrumented helmet. Resultant linear acceleration data from both systems were utilized to calculate two impact metrics used for relating time-weighted head acceleration to risk of injury, GSI and HIC15. GSI was calculated for each impact event19:GSI=∫Ta(t)2.5dtGSI=∫Ta(t)2.5dt {\text{GSI}} = \int\limits_{{}}^{T} {a\left( t \right)^{2.5} dt} (1)where a(t) is the linear resultant acceleration of the head CG, and T is the impact duration. HIC15 was evaluated over an incremental time window38,45 of maximum duration t2 − t1 = 15 ms:HIC15=⎧⎩⎨⎪⎪⎪⎪(t2−t1)⎡⎣⎢∫t1t2a(t)dt/(t2−t1)⎤⎦⎥2.5⎫⎭⎬⎪⎪⎪⎪maxHIC15={(t2−t1)[∫t1t2a(t)dt/(t2−t1)]2.5}max {\text{HIC}}_{15} = \left\{ {\left( {t_{2} - t_{1} } \right)\left[ {\int\limits_{{t_{1} }}^{{t_{2} }} {a\left( t \right)dt/\left( {t_{2} - t_{1} } \right)} } \right]^{2.5} } \right\}_{\max } (2)where t1 and t2 are the initial and final points of the time window and a(t) is the linear resultant acceleration of the CG.To establish a correlation between severity measures from the HIT System and HIII, linear regression analysis (Microsoft Excel 2010) was performed on all acceleration and severity measures—peak linear acceleration, peak rotational acceleration, GSI, and HIC15.1 All regressions were performed both on the entire dataset and by impact site using Eq. (3):y=mx+y0y=mx+y0 y = mx + y_{0} (3)where x is the HIII measure, y is the HIT System measure, and the linear slope, m, is the relationship between the measurements. For all conditions, y0 was constrained to be zero since both systems have a baseline output of zero when not impacted. The coefficient of determination (r2) was also calculated for each regression as a measure of goodness of fit. Finally, the average absolute location difference between the estimated instrumented helmet location and the direction vector (in spherical coordinates) of the peak HIII linear acceleration was calculated for each impact. The overall average difference between the two measures along with the difference by impact site was calculated.

Primary Impact SitesFour primary impact sites, designated as A, B, C, and D (Fig. 2), were identified by Pellman et al.37 as points of contact that most frequently result in concussion for NFL athletes. In that study, the average impact velocity for impacts associated with and without concussion was determined to be 9.3 ± 1.9 m/s and 7.0 ± 2.6 m/s, respectively. For this study, each site was impacted at four target speeds: 4.4, 7.4, 9.3, and 11.2 m/s. The highest three velocities correspond to the average speed, ±1 standard deviation, of video-reconstructed impacts sustained before diagnosis of concussion in NFL players.38 4.4 m/s represents the average velocity, −1 standard deviation, of all sub-concussive impacts evaluated by the NFL. This lower impact velocity was selected to insure inclusion of conditions that may not result in injury, but occur most often in the field.20,24,30 Between three and five trials were conducted for each site and speed combination (Table 1).Open image in new windowFigure 2Four primary impact sites were (A, B, C, and D) were identified as points of contact that most frequently result in mTBI for NFL athletes. Each site was impacted at four target speeds: 4.4, 7.4, 9.3, and 11.2 m/sTable 1Kinematic parameters were recorded from instrumented helmets and a Hybrid III ATD following 54 linear impactor tests to correlate output measures from on-field and laboratory measurement systemsTarget velocity (m/s)Number of trials per primary impact siteABCD4.433347.433349.3333511.23335Total trials12121218Test conditions are representative of sites and velocities previously identified as being most associated with concussion in the NFL. The number of trials conducted by impact site and target impactor velocity is providedSecondary Impact SitesAfter initial testing and analyses were conducted, two additional impact sites, A′ and A′′ (Fig. 2), were added to the test protocol at the request of the then co-chairman of the NFL mTBI Committee to characterize HIT System measurement capability during conditions where extensive helmet movement was anticipated to occur. Sites A′ and A′′ were identified by Craig (2007) who, when reevaluating video from the initial NFL studies, suggested a sub-category of site A impacts existed that resulted in chinstrap loading.9 The two additional sites were variations of the original site A, located directly in line with the mid-sagittal plane of the headform, and tilted away from the impactor at −10° and −20° angles, respectively. Based on the recommendations made by Craig,9 and previous experience which predicted that facemask deformation and HIII neck damage were likely to occur, the 11.2 m/s speed was not used for A′ and A′′ tests. Three trials were conducted at each test condition except for A′′ at 4.4 m/s which had four trials.

The sliding table allows for both more realistic kinematics and improved equipment durability while not affecting the head response, since the head acceleration impulse occurs before significant neck bending.39Open image in new windowFigure 1A pneumatic linear impactor (Biokinetics, Inc) was employed to replicate on-field reconstruction of head impacts. The ram mass, impactor surface, and target velocities were selected to best simulate on-field head impacts occurring in the National Football LeagueImpacts were delivered to the HIII using a pneumatically activated linear impactor

Head Impact Telemetry (HIT) System technology (Simbex, Lebanon, NH; Sideline Response System, Riddell, Chicago, IL) records the frequency, location, and magnitude of impacts sustained by football players during live play,4,11,16,20,27,44 and was specifically developed to enable research to better understand the relationship between measured parameters of head kinematics and concussion. This technology incorporates an array of non-orthogonal accelerometers (Analog Devices, Inc., Cambridge, MA), data acquisition, and RF telemetry hardware into self-contained inserts placed in commercially available helmets. While physically connected to the helmet, the instrumented insert acts as an effective spring to maintain contact with the head during impact and to decouple head acceleration from helmet acceleration.26 Because traditional mathematical approaches to calculate linear and rotational acceleration at the head center of gravity, such as those employed by the HIII, require precise mounting of accelerometers that is not practical for field implementation, the HIT System uses proprietary, simulated-annealing optimization algorithms to estimate linear and rotational acceleration

In an additional study to quantify the effect of head, neck, and torso coupling, Beusenberg et al. found that if a player’s neck coupling with the head (e.g., stiffness, strength, etc.) deviates from that of the HIII head and neck, peak linear acceleration can be altered by more than 15% and peak rotational acceleration will be drastically different.3 Due to the complexity of these reconstructions and the beneficial insight they provide, these errors were deemed to be within an acceptable level, and output measures from the reconstructions were considered reasonable for use in estimating the actual head kinematics for this group of impacts.31

To assess the accuracy of acceleration measures derived from the NFL studies, Newman et al. quantified the error associated with reconstruction variables, including effects of noise in the data acquisition system, processing inaccuracy, and determining the impact velocity from multiple camera angles.

By matching the obtained impact velocity and observed direction, these 31 cases were re-enacted in a laboratory using Hybrid III (HIII) anthropomorphic test devices (ATDs). Head linear and rotational accelerations were measured from accelerometers in the ATD’s head, permitting calculation of injury metrics such as Gadd Severity Index (GSI) and Head Injury Criterion (HIC)

To better understand the etiology of concussion and the relationship between head impacts and injury, however, a large number of measured impacts from multiple athletes, both concussed and non-concussed, are required.

Recent studies suggest a link exists between concussion history and developing mild cognitive impairment, clinical depression, and early onset of Alzheimer’s disease,21,22 creating a public and scientific debate about effective prevention strategies for concussions.

Over the last decade, advances in technology have enabled researchers to evaluate concussion biomechanics through measurement of head impacts sustained during play using two primary methods: (1) laboratory reconstruction of open-field head contact, and (2) instrumented helmets. The purpose of this study was to correlate measures of head kinematics recorded by the Head Impact Telemetry (HIT) System (Simbex, NH) with those obtained from a Hybrid III (HIII) anthropometric headform under conditions that mimicked impacts occurring in the NFL. Linear regression analysis was performed to correlate peak linear acceleration, peak rotational acceleration, Gadd Severity Index (GSI), and Head Injury Criterion (HIC15) obtained from the instrumented helmet and HIII. The average absolute location error between instrumented helmet impact location and the direction of HIII head linear acceleration were also calculated.