Comparison of Head and Neck Kinematics and Electromyography Response for Low-Speed Frontal Impacts with Pediatric and Young Adult Volunteers

No data exist on the electromyography (EMG) responses in children exposed to dynamic impacts and the resulting head and neck kinematics. The objective of the current study was to measure the EMG responses of the neck, torso and lower extremity in children and young adults during a low speed frontal impact using surface electrodes bilaterally and compare these muscle responses to key kinematic events. Low speed frontal impact tests were performed on seated and restrained pediatric (n=11, ages 8-14 years) and adult (n=9, ages 18-30 years) male subjects. Subjects experienced a maximum acceleration pulse of 4.9 g in 55.7 msec. The timing and magnitude of the EMG responses were compared between the pediatric and young adult groups. Since no bilateral differences were observed in muscle response, the left and right muscle responses were combined for the comparison with kinematics. The interquartile range (IQR) of maximum forward head excursion occurred at 183.8-224.6 msec. The IQR of onset times for the Cervical Paraspinous, Sternocleidomastoid and Upper Trapezius were 34.00-52.67 msec, 50.8382.67 msec, and 48.17-98.50 msec, respectively. Also of note, the IQR for time of peak muscle activity for the Cervical Paraspinous, Sternocleidomastoid, and Upper Trapezius were 84.00-154.0 msec, 121.5190.0 msec, and 116.2-169.3 msec. It can be seen that in the low-speed frontal impact tests, all the neck muscles under evaluation were activated (onset) and had reached their peak muscle response prior to the time of maximum forward head excursion. These preliminary results indicate that muscle response may influence head motion in the low-speed environment. Further analysis will examine the relationship between onset times, peak response times and maximum head rotation as well as between magnitude of muscle response and maximum head excursion. INTRODUCTION According to the US Centers for Disease Control, the leading cause of death for individuals 5-34 years is motor vehicle collisions (MVCs) [CDC, 2009]. Computational modeling provides a tool for the safety industry to examine occupant motion during MVCs and determine injury causation in an effort to design safety countermeasures. Accurate determination of injury causation may require an understanding of the role of muscle activation in occupant kinematics during MVCs. Electromyography (EMG) provides a means to measure and evaluate muscle response during a dynamic event. Surface EMG (sEMG) electrodes are placed on the skin over the muscle of interest using a gel-based adhesive to reduce skin resistance and thereby facilitate conduction. These electrodes are able to detect, amplify and record variations in skin voltage due to underlying muscle contractions. Such data have never been collected for children in an automotive environment. Currently, improvements in vehicle safety are achieved through analyses of post-mortem human subjects and anthropomorphic test device (crash test dummy or ATD) kinematics, which do not account for the effect of active musculature on occupant kinematics. Several studies have evaluated muscle activity of adult volunteers in response to dynamic events. Low-speed frontal impact tests on adult volunteers conducted by Ejima et al. [2007] showed a considerable influence of muscle response on the occupant kinematics. Occupants that were tensed prior to the impact demonstrated more controlled motion than when relaxed pre-impact. Choi et al. [2005] quantified the muscle tensing activity of adult occupants during pre-impact bracing and found good correlation with simulated muscle tensing behavior. These results were further used to validate computational human models with simulated muscle activity [Ejima et al., 2009; Choi et al., 2005]. Additionally, Kumar et al [2006] examined the effect of seat belt use on cervical muscle activity in response to whiplash-type events for multiple directions and varying magnitudes of acceleration, and found that with increasing acceleration, the time to onset of EMG decreased, indicating an effect of acceleration on muscle activation. These results were also supported by their study of low-velocity frontal impacts [Kumar et al, 2003]. In a study to evaluate the cervical muscle activity of volunteers in low-speed rear impacts, Magnusson et al [1999] concluded that due to the onset times of the cervical muscles, the muscles could influence injury patterns. They also determined that the location of the muscle relative to the spinal axis influences the muscle’s reaction time. Bose et al. [2008] used musculoskeletal modeling to study the role of bracing on kinematics of restrained occupants. Their study demonstrated the utility of measuring muscle activity during dynamic events to incorporate muscle response in models optimizing adaptive restraint systems. They found that the injury outcomes of the occupant, as well as their interaction with the restraint system, were sensitive to pre-impact bracing by the occupant. Studies on occupant awareness have shown that unaware subjects have increased excursion than those with some level of awareness of the imminent perturbation [Siegmund et al, 2003; Kumar et al, 2000; Kumar et al, 2002]. All of the aforementioned studies were conducted on adult subjects; similar data do not exist in the pediatric population. There is also a dearth of information on the effect of age on the temporal nature of muscle activation during a dynamic event. Therefore, the objective of the current study was to measure the EMG responses of the neck, torso and lower extremity in children during a low speed frontal impact and compare those to similar data from young adults. We hypothesize that under low-speed frontal loading conditions, timing and magnitude of neck muscle activity can influence the occupant’s head and neck kinematics. METHODS This study protocol was reviewed and approved by the Institutional Review Boards at The Children's Hospital of Philadelphia, Philadelphia, PA and Rowan University, Glassboro, NJ. Informed consent / assent was obtained from all the participants of this study. Human Volunteer Instrumentation, Testing, and Data Processing A comprehensive description of the testing method can be found in Arbogast et al [2009]. Briefly, low-speed frontal sled tests were conducted using 20 male human volunteers (11 pediatric subjects: 8-14 years old, 9 young-adult subjects: 18-30 years old). All subjects were between 5 and 95 percentile for their age-appropriate height, weight and body mass index [Centers for Disease Control, 2000; NHANES, 1994]. Subjects with existing neurologic, orthopedic, genetic, or neuromuscular conditions, any previous injury or abnormal pathology related to the head, neck or spine were excluded from the study. A pneumatically actuated, hydraulically controlled low-speed acceleration volunteer sled (Figure 1) consisting of a moving platform with a low back padded seat, lap-shoulder belt and an adjustable foot rest was used to subject restrained human volunteers to a sub-injurious, lowspeed frontal crash pulse. The maximum linear acceleration was 25% below the maximum acceleration measured during an amusement park bumper car ride (4.3 g in 61.9 msec) [Arbogast et al, 2009]. Figure 1: Schematic of low-speed acceleration sled Several anthropometric measurements were obtained from the subjects prior to testing. Photoreflective targets were placed on anatomical landmarks including the head, spine, shoulders, sternum, and legs and were tracked using a 3D motion analysis system at 100 Hz (Model Eagle 4, Motion Analysis Corporation, Santa Rosa, CA). EMG measurements were obtained for all trials. Prior to sEMG electrode placement, the subject’s skin was cleaned by applying Skin Prepping Gel (NUPREP, Weaver and Co., Aurora, CO). Disposable, self-adhesive dual surface EMG electrodes (20 mm inter-electrode distance) (Noraxon, Inc., Scottsdale, AZ) were placed bilaterally (left-L, right-R) on key muscle groups of the neck (Sternocleidomastoid, Cervical Paraspinous and Upper Trapezius), lower torso (Erector Spinae), and lower extremities (Rectus Femoris) to measure the muscle response of the subjects (Figure 2). A grounding electrode was centered over the right mastoidale. Signals from the muscle leads were passed to two battery-operated eight-channel FM transmitters (TeleMyo 2400T V2, Noraxon, Scottsdale, AZ) and recorded throughout each trial at 1,500 Hz per channel. Within the EMG acquisition system, the signals were amplified (gain 1000) with a single-ended amplifier (impedance >10 Mohm) and filtered with a fourth-order Butterworth filter (10–500 Hz) and common mode rejection ratio of 130 dB at direct current (minimum 85 dB across entire frequency of 10–500 Hz). Prior to sled testing, for each subject, the maximum voluntary isometric contraction (MVIC) for these muscles was measured during 10 seconds of attempted neck flexion, neck extension, torso extension and leg extension. Mean MVIC for each of the bilateral muscles was calculated by averaging with a 25 ms window over the middle six seconds of the entire duration of the isometric contraction trial. Each subject performed a total of six dynamic trials on the low speed sled with an interval of 10 minutes between trials. Subjects were encouraged to fully relax prior to each trial and were provided with a countdown leading up to trigger press. Signals from the accelerometers, angular rate sensor (ARS) and load cells were sampled at 10,000 Hz using a T-DAS data acquisition system (Diversified Technical Systems Inc., Seal Beach, CA) with a built-in anti-aliasing filter (4,300 Hz). The time series motion analysis and TDAS data were imported into MATLAB (Mathworks, Inc., Natick, MA) for kinematic calculations using a custom written program. Figure 2: Surface EMG electrode locations on a child subject. Using a customized MATLAB algorithm, the raw EMG signals were filtered using a Band-pass (20–510 Hz) Finite Impulse Response (using a Kaiser Window method) filter [Winter et al, 1980; Merletti et al, 1999; DeLuca et al, 2010]. A root-mean-squared (RMS) method with a 25 ms moving average smoothing window was applied, as it had the least effect on the EMG onset times (Figure 3). This process also served to rectify the signal. The EMG signals were analyzed with respect to several key event time points such as event onset (time zero, beginning of acceleration pulse) and time of maximum head excursion. The mean of the prestimulus noise (i.e. EMG signal from event onset to one second prior) for each muscle was subtracted from its respective signal. For each subject, the processed EMG signals were normalized by their respective MVIC. Sternocleidomastoid Paraspinous