- Email: [email protected]
Influence of the crash pulse shape on the peak loading and the injury tolerance levels of the neck in in vitro low-speed side-collisions
ARTICLE IN PRESS
Journal of Biomechanics 39 (2006) 323–329 www.elsevier.com/locate/jbiomech www.JBiomech.com
Influence of the crash pulse shape on the peak loading and the injury tolerance levels of the neck in in vitro low-speed side-collisions Annette Kettler, Kai Fruth, Lutz Claes, Hans-Joachim Wilke Institute of Orthopaedic Research and Biomechanics, University of Ulm, Helmholtzstr. 14, D-89081 Ulm, Germany Accepted 14 November 2004
Abstract The aim of the present in vitro study was to investigate the effect of the crash pulse shape on the peak loading and the injury tolerance levels of the human neck. In a custom-made acceleration apparatus 12 human cadaveric cervical spine specimens, equipped with a dummy head, were subjected to a series of incremental side accelerations. While the duration of the acceleration pulse of the sled was kept constant at 120 ms, its shape was varied: Six specimens were loaded with a slowly increasing pulse, i.e. a low loading rate, the other six specimens with a fast increasing pulse, i.e. a high loading rate. The loading of the neck was quantified in terms of the peak linear and angular acceleration of the head, the peak shear force and bending moment of the lower neck and the peak translation between head and sled. The shape of the acceleration curve of the sled only seemed to influence the peak translation between head and sled but none of the other four parameters. The neck injury tolerance level for the angular acceleration of the head and for the bending moment of the lower neck was almost identical for both, the high and the low loading rate. In contrast, the injury tolerance level for the linear acceleration of the head and for the shear force of the lower neck was slightly higher for the low loading rate as compared to the high loading rate. For the translation between head and sled this difference was even statistically significant. Thus, if the shape of the crash pulse is not known, solely the peak bending moment of the lower neck and the peak angular acceleration of the head seem to be suitable predictors for the neck injury risk but not the peak shear force of the lower neck, the peak linear acceleration of the head and the translation between head and thorax. r 2004 Elsevier Ltd. All rights reserved. Keywords: Neck injury; Whiplash trauma; Crash pulse; Injury criterion; Injury threshold
1. Introduction The whiplash trauma of the cervical spine is still one of the most common injuries in traffic accidents. Whiplash injuries therefore continue to represent a substantial societal problem worldwide with associated costs, which are estimated at $4.5–10 billion annually in the US (Tencer et al., 2003; Yoganandan et al., 1999, 2001), at h1 billion annually in Germany (Hell and Langwieder, 1998) and in average at over CAD$3,800 per whiplash subject in Quebec (Spitzer et al., 1995). Corresponding author. Tel.: +49 731 500 23485; fax: +49 731 500 23498. E-mail address: [email protected] (A. Kettler).
0021-9290/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.jbiomech.2004.11.017
In view of these enormous costs the development of new seats, head restraints and vehicles is of high importance. To determine their performance in protecting the occupant against whiplash, the use of injury criteria such as the neck displacement criterion (NDC) (Viano and Davidsson, 2002), the Nkm criterion (Schmitt et al., 2001), or the neck injury criterion (NIC) (Bostro¨m et al., 1996) has been proposed. These parameters are either directly or indirectly related to the loading of the neck: The NIC is calculated from the velocity and acceleration of the head relative to the first thoracic vertebra, the Nkm criterion is based on the upper neck flexion/extension moment and shear force and the NDC is based on the angular and linear displacement response of the occipital condyles with respect to the first thoracic vertebra.
ARTICLE IN PRESS 324
A. Kettler et al. / Journal of Biomechanics 39 (2006) 323–329
To determine human tolerance levels for these injury criteria, however, is difficult since the human tolerance to complex three-dimensional (3D) loading is not yet well characterised (Myers and Winkelstein, 1995) and the rate-dependence of the human tolerance to whiplash loading is still unknown. This rate-dependence is of special interest since the real crash pulse shapes are varying considerably from pulses with high to pulses with low loading rates (Krafft et al., 1998, 2002). The aim of the present in vitro study therefore was to investigate the effect of the crash pulse shape on the peak loading and the injury tolerance levels of the human neck.
suspension cord
dummy head specimen
2. Materials and methods Twelve fresh frozen human cadaveric cervical spine specimens including the occiput (C0) and the first thoracic vertebra (T1) were selected for this study. Exclusion criteria were spinal disorders except for minor degeneration. The age of the donors was 81 years in mean. Before testing, the specimens were thawed at 4 1C and all soft tissue surrounding the discoligamentous spine was carefully removed. C0 and T1 were embedded in polymethylmethacrylate (PMMA) while the foramen magnum was aligned horizontally. Then, the specimens were mounted to a custom-made acceleration apparatus, that is basically composed of a sled, a railtrack and a pneumatic acceleration unit (Kettler et al., 2004). Since the trunk of a vehicle passenger is known to move during a real-life crash, these movements had to be simulated in the present experiment. This was accomplished by the use of a pivot table, which was located between the sled and the lower end of the specimens and which was allowed to rotate passively about an axis perpendicular to the direction of the acceleration. A dummy head made of wood and brass was developed to represent the average human head with a mass of 4.5 kg (Clemens 1972) and a humanlike outer geometry. This dummy head was rigidly fixed to the PMMA block on C0 with its centre of gravity located analogous to that of the human head with respect to the cervical spine (Vital and Senegas, 1986). Before acceleration the dummy head was suspended using a thin cord connected at one side to the upper surface of the dummy head and at the other side to the frame of the sled, approximately 10 cm above the head (Fig. 1). By the use of this cord the specimens could be balanced in the upright, neutral position before acceleration. At the beginning of the acceleration, however, the cord was cut to allow the head to move completely unconstrained. Each specimen was subjected to a series of incremental 901 side collisions from the right. The first impact
pivot table
Fig. 1. In the custom-made pneumatic acceleration apparatus the lower end of the specimens was fixed to a pivot table. On the occipital bone a dummy head was mounted, which had to be balanced with a suspension cord. This cord was cut at the beginning of each impact.
was characterised by a peak acceleration of the sled of approximately 1 g. In each following impact this acceleration was increased by another 1 g. The experiment was stopped as soon as any structural failure became macroscopically visible with the naked eye. After structural injury had occurred, anteroposterior and lateral radiographs were taken for documentation purposes and to assess bony injuries, which were not visible from outside. While the duration of the acceleration pulse of the sled was kept constant at approximately 120 ms for all specimens, its shape was varied: Six specimens were loaded with slowly increasing pulse, i.e. with a low loading rate, the other six specimens with a fast increasing pulse, i.e. with a high loading rate (Fig. 2). The uniaxial acceleration of the sled (EGE-73AE1100D1, measurement range7100 g, Entran Ludwigshafen, Germany), the 3D linear acceleration of the head centre of gravity (EGE-73AE1-100D1, measurement range 7100 g, Entran Ludwigshafen, Germany) and the 3D forces and moments between neck and pivot platform (lower neck load cell 4894J, measurement range 452 Nm, respectively, 13345 N, Denton COE GmbH, Heidelberg, Germany) were recorded simulta-
ARTICLE IN PRESS A. Kettler et al. / Journal of Biomechanics 39 (2006) 323–329
ase
fast inc
asled
asled
w slo
e fast decr
e
as
re
inc
rease
neously during impact. Presample filtering was carried out with an analogue Butterworth low-pass filter at a cut-off frequency of 250 Hz (acceleration of the sled) or 2000 Hz (all other parameters). Then, data were sampled at a rate of 9.6 kHz for each channel and digitally lowpass filtered at a cut-off frequency of 60 Hz (acceleration of the sled), 600 Hz (bending moment lower neck) or 1,000 Hz (accelerations of the head and shear force lower neck) (adapted to SAE J211 and DIN ISO 6487). The velocity and the distance covered by the sled and the head were obtained by integration of the filtered accelerometer data. According to the 6-accelerometer scheme, the angular acceleration of the head was calculated based on the data recorded by the linear accelerometers of the head centre of gravity and the data collected by another three linear accelerometers, which were placed on the surface of the head. Then, the following five loading parameters were determined (Fig. 3):
0
120
time [ms]
slo w
de
cre
0
325
the peak linear acceleration of the head along the local y-axis (ay head): the peak angular acceleration of the head about the local x-axis (ax head); the peak shear force of the lower neck along the local y-axis (Fy lower neck); the peak bending moment of the lower neck about the local x-axis (Mx lower neck); the peak translation between head and sled in direction of the acceleration of the sled (translation head-sled).
Injury tolerance levels were calculated for each of these five parameters. They were defined as the peak value of the respective parameter during the last acceleration, which not yet caused structural injury. Wilcoxon signed rank Tests were used to evaluate the effect of the crash pulse shape on the injury thresholds of the five loading parameters. Statistical significance was assumed at po0:05:
2. Results as
e
120
time [ms]
Fig. 2. Schematic sled acceleration curves. For six specimens the acceleration curve of the sled (asled) was characterised by a slow increase and a fast decrease (left) and for the other six by a fast increase and a slow decrease (right). In all cases the duration of the acceleration pulse was approximately 120 ms.
The shape of the acceleration curve of the sled only seemed to influence the translation between head and sled (Fig. 8, left) but none of the other four parameters (Figs. 4–7, left). In the case of the translation between head and sled, the low loading rate, i.e. the slowly increasing acceleration pulse of the sled, was associated with higher peak translations than the high loading rate (Fig. 8, left).
Fig. 3. The linear acceleration of the head along the local y-axis (ay ), the angular acceleration of the head about the local x-axis (ax ), the shear force of the lower neck along the local y-axis (F y ), the bending moment of the lower neck about the local x-axis (M x ) and the linear acceleration of the sled were measured simultaneously.
ARTICLE IN PRESS 326
A. Kettler et al. / Journal of Biomechanics 39 (2006) 323–329
Fig. 4. Left: Peak linear acceleration (ay) of the head correlated with the mean acceleration (amean) of the sled. Linear regressions across the impacts, which did NOT lead to structural injury (black rhombi for the low loading rate resp. grey quadrates for the high loading rate). Right: Peak linear acceleration of the head during the last impacts before the specimens were structurally injured. Mean value and standard deviation across the six specimens, which were accelerated with the slowly increasing pulse (left bar) and with the fast increasing pulse (right bar). P-value calculated with the Wilcoxon signed rank test.
Fig. 6. Left: Peak shear force (F y ) of the lower neck correlated with the mean acceleration (amean) of the sled. Linear regressions across the impacts, which did NOT lead to structural injury (black rhombi for the low loading rate, respectively, grey quadrates for the high loading rate). Right: Peak shear force of the lower neck during the last impacts before the specimens were structurally injured. Mean value and standard deviation across the six specimens, which were accelerated with the slowly increasing pulse (left bar) and with the fast increasing pulse (right bar). P-value calculated with the Wilcoxon signed rank test.
Fig. 5. Left: Peak angular acceleration (ax ) of the head correlated with the mean acceleration (amean) of the sled. Linear regressions across the impacts, which did NOT lead to structural injury (black rhombi for the low loading rate, respectively, grey quadrates for the high loading rate). Right: Peak angular acceleration of the head during the last impacts before the specimens were structurally injured. Mean value and standard deviation across the six specimens, which were accelerated with the slowly increasing pulse (left bar) and with the fast increasing pulse (right bar). P-value calculated with the Wilcoxon signed rank test.
The neck injury tolerance level for the angular acceleration ax of the head and for the bending moment M x of the lower neck was almost identical for both, the high and the low loading rate (Figs. 5 and 7, right). In contrast, the injury tolerance level for the linear acceleration ay of the head and for the shear force F y of the lower neck was slightly higher for the low loading rate than for the high loading rate (Figs. 4 and 6, right). For the translation between head and sled this difference was even statistically significant (p ¼ 0:0247) (Fig. 8, right). The structural injury was always localised in one of the segments C4–5, C5–6, C6–7, or C7–T1. Eleven out of the 12 specimen sustained a soft tissue rupture of the left facet joint capsule (impact from the right) and the intervertebral disc. In some cases an additional partial rupture of the right facet joint capsule and/or an additional fracture of the upper right articular process of the caudally adjacent vertebra was observed. On the radiographs no injury to the vertebral bodies or endplates were detected. In contrast to these 11 specimens, the remaining one only sustained a partial rupture of the left facet joint capsule without involvement of the disc. This specimen was excluded from evaluation, since otherwise the crash pulse shape would have no more been the only influencing variable and differences between the peak loading and between the injury thresholds would have also been attributable to differences in the extend of the injury.
ARTICLE IN PRESS A. Kettler et al. / Journal of Biomechanics 39 (2006) 323–329
Fig. 7. Left: Peak bending moment (M x ) of the lower neck correlated with the mean acceleration (amean) of the sled. Linear regressions across the impacts, which did NOT lead to structural injury (black rhombi for the low loading rate, respectively, grey quadrates for the high loading rate). Right: Peak bending moment of the lower neck during the last impacts before the specimens were structurally injured. Mean value and standard deviation across the six specimens, which were accelerated with the slowly increasing pulse (left bar) and with the fast increasing pulse (right bar). P-value calculated with the Wilcoxon signed rank test.
4. Discussion According to the degree to which the crash pulse shape influenced the peak loading and the injury thresholds of the neck, the five loading parameters, which were evaluated in the present study could be assigned to three different groups: 4.1. Group 1: no effect on peak values and injury thresholds The peak values and injury thresholds of the angular acceleration ax of the head and of the bending moment M x of the lower neck were not affected by the shape of the acceleration curve of the sled (Figs. 5 and 7). These two parameters therefore seem to be useful to predict the neck injury risk especially in cases where the shape of the crash pulse is not known. 4.2. Group 2: no effect on peak values, effect on injury thresholds
327
Fig. 8. Left: Peak translation between head and sled correlated with the mean acceleration (amean) of the sled. Linear regressions across the impacts, which did NOT lead to structural injury (black rhombi for the low loading rate , respectively, grey quadrates for the high loading rate). Right: Peak translation between head and sled during the last impacts before the specimens were structurally injured. Mean value and standard deviation across the six specimens, which were accelerated with the slowly increasing pulse (left bar) and with the fast increasing pulse (right bar). P-value calculated with the Wilcoxon signed rank test.
uniaxial tensile failure tests cervical spine ligaments (ligamentum longitudinale anterius and ligamentum flavum) were shown to have the inverse behaviour with higher failure forces under high loading rates than under low loading rates (Yoganandan et al., 1989) and human cadaver head–neck specimens tolerated higher compressive loads the faster the load was applied (Pintar et al., 1998). Using a linear regression model the authors could additionally show that an increasing age reduced the effect of loading rate and at approximately 82 years, this effect did no more exist. Thus, the correlation between the crash pulse shape and the injury thresholds for the linear acceleration ay of the head and for the shear force F y of the lower neck does not match the correlation expected considering the pertinent literature. This contradiction supports the assumption that neither the linear acceleration of the head nor the shear force of the lower neck were directly responsible for the occurrence of the structural neck injury. Thus, these two parameters do not seem to be appropriate to predict the neck injury risk. 4.3. Group 3: effect on peak values and injury thresholds
Similarly, the peak values of the linear acceleration ay of the head and of the shear force F y of the lower neck also did not depend on the crash pulse shape. However, their injury thresholds did (Figs. 4 and 6): slightly higher peak values were tolerated if the crash pulse had a slow increase than if it had a fast increase. In contrast, in
In contrast to the above-mentioned loading parameters, the peak values of the translation between head and sled depended on the shape of the acceleration curve: they were slightly higher for the low loading rate as compared to the high loading rate (Fig. 8). This effect
ARTICLE IN PRESS 328
A. Kettler et al. / Journal of Biomechanics 39 (2006) 323–329
can be explained by the viscous behaviour of human tissue, since viscous material is characterised by a ratedependency of the deformation magnitude. Besides the peak values, the injury tolerance levels also were affected by the crash pulse shape: higher peak translations between head and sled were tolerated for the low loading rate than for the high loading rate. In contrast to these results, the failure deformations of cervical spine ligaments under tension and human cadaver head–neck specimens under compression reported in the literature were similar for all tested loading rates (Yoganandan et al., 1989, 1990). Since the own results do not support these published data, the translation between head and neck is most probably not directly related to the local deformations which finally may cause injury. This assumption is supported by an in vitro study on cadaver cervical spine specimens. It was shown that the maximum head translation during rear end impacts increased with increasing sled acceleration while the maximum posterior head rotation decreased (Cholewicki et al., 1998). I.e. one and the same head translation may be associated with very different local deformations within the single cervical spine segments. The translation between head and sled therefore seems to be unsuitable to predict the local deformations and with it the neck injury risk especially if the shape of the crash pulse is unknown. These correlations between the crash pulse shape and the peak loading and injury tolerance levels of the neck primarily apply to structural injuries but might also be extrapolated to non-structural whiplash-like injuries under the following assumptions: First, whiplash injuries have to be caused mechanically by inertial loads exerted on the cervical spine. Even though some researchers hypothesise that not the spine itself but rather the neck muscles or the nerve roots are the primary site of injury (Bostro¨m et al., 1996; Brault et al., 2000) or that whiplash mainly is a psychological or societal problem (Castro et al., 2001; Obelieniene et al., 1999), most agree with this assumption. Second, structural injuries have to be preceded by subfailure injuries, which on their part have to be responsible for the symptoms the whiplash patient is suffering from. This assumption is supported by in vitro studies, which all showed that structural injuries actually are preceded by functional injuries, i.e. local instabilities without any detectable structural damage (Hartwig et al., 2004; Panjabi et al., 1998). Third, side collisions have to cause injuries which are similar to those caused by rear-end or frontal collisions. This assumption is partially supported by in vitro experiments on rear end collisions (Panjabi et al., 1998). Kind (soft tissue injury) and localisation (lower cervical spine) of the injuries were similar to the present study on side collisions. A limitation of the present study is the fact that the stabilising effect of the neck muscles was missing. During
whiplash, this stabilising effect is mainly composed of a passive and an active component. While the passive component permanently acts even if the impact is not expected, the effect of the active component depends on a certain reaction time. In low-speed rear-end collisions with volunteers the average reaction time of superficial and deep neck muscles was found to range between 73 and 175 ms referred to the beginning of the movement of the sled (Magnusson et al., 1999). Thus, even if another 100 ms were needed for electromechanical coupling, single muscle fibres would already actively stabilise the cervical spine during the phase of maximum stress and strain, which, in the present study, occurred approximately 150 ms after the onset of the acceleration pulse. Provided that this actively and the passively stabilising effect of the neck muscles only affects the magnitude but not the quality of the loading of the neck, the correlations between the crash pulse shape and the peak loading and injury tolerance levels made in the present study should also apply to real-life crashes, whereas the absolute values of failure do definitely not. Thus, the data of the present study should not be used to define any real-life injury thresholds. In conclusion, the crash pulse shape seems to influence the peak value of the translation between head and thorax (a high loading rate is associated with a small translation and vice versa) but not the peak values of the linear and angular acceleration of the head and the shear force and bending moment of the lower neck. The crash pulse shape also seems to take influence on the injury thresholds for the peak translation between head and thorax. Higher peak translations are tolerated if the loading rate is low than if it is high. The injury thresholds for the peak shear force of the lower neck and the peak linear acceleration of the head also seem to be slightly influenced by the crash pulse shape, the injury threshold for the peak bending moment of the lower neck and the peak angular acceleration of the head however, do not. Therefore, if the shape of the crash pulse is not known, only the peak bending moment of the lower neck and the peak angular acceleration of the head seem to be suitable predictors for the neck injury risk but not the peak shear force of the lower neck, the peak linear acceleration of the head and the translation between head and neck. Acknowledgement The authors gratefully acknowledge the German Research Council (Deutsche Forschungsgemeinschaft, DFG, HA 3276/1-1) for financial support. References Bostro¨m, O., Svensson, M., Aldman, B., Hansson, H., Haland, Y., Lo¨vsund, P., Seeman, T., Suneson, A., Sa¨ljo¨, A., O¨rtengren, T., 1996. A new neck injury criterion candidate based on injury
ARTICLE IN PRESS A. Kettler et al. / Journal of Biomechanics 39 (2006) 323–329 findings in the cervical ganglia after experimental neck extension trauma. Proceedings of the IRCOBI Conference, 123–136. Brault, J.R., Siegmund, G.P., Wheeler, J.B., 2000. Cervical muscle response during whiplash: evidence of a lengthening muscle contraction. Clinical Biomechanics (Bristol, Avon) 15 (6), 426–435. Castro, W.H., Meyer, S.J., Becke, M.E., Nentwig, C.G., Hein, M.F., Ercan, B.I., Thomann, S., Wessels, U., Du Chesne, A.E., 2001. No stress—no whiplash? Prevalence of ‘‘whiplash’’ symptoms following exposure to a placebo rear-end collision. International Journal of Legal Medicine 114 (6), 316–322. Cholewicki, J., Panjabi, M.M., Nibu, K., Babat, L.B., Grauer, J.N., Dvorak, J., 1998. Head kinematics during in vitro whiplash simulation. Accident Analysis and Prevention 30 (4), 469–479. Clemens, H.J., 1972. Weight of the head in man—a biomechanical problem. Archiv fur Orthopadische und Unfallchirurgic 73 (3), 220–228. Hartwig, E., Kettler, A., Schultheiss, M., Kinzl, L., Claes, L., Wilke, H.J., 2004. In vitro low-speed side collisions cause injury to the lower cervical spine but do not damage alar ligaments. European Spine Journal (online first). Hell, W., Langwieder, K., 1998. Epidemiologische Daten zur HWS— Beschleunigungsverletzung. Die Notwendigkeit eines verbesserten Diagnosestandards und bessere Pra¨vention bei Auffahrunfa¨llen. Hefte zu Der Unfallchirurg 272, 80–82. Kettler, A., Schmitt, H., Simon, U., Hartwig, E., Kinzl, L., Claes, L., Wilke, H.J., 2004. A new acceleration apparatus for the study of whiplash with human cadaveric cervical spine specimens. Journal of Biomechanics 37 (10), 1607–1613. Krafft, M., Kullgren, A., Tingvall, C., 1998. Crash pulse recorders in rear impacts—real life data. ESV Proceedings. Krafft, M., Kullgren, A., Ydenius, A., Tingvall, C., 2002. Influence of crash pulse characteristics on whiplash associated disorders in rear impacts—crash recording in real life crashes. Traffic Injury Prevention 3, 141–149. Magnusson, M.L., Pope, M.H., Hasselquist, L., Bolte, K.M., Ross, M., Goel, V.K., Lee, J.S., Spratt, K., Clark, C.R., Wilder, D.G., 1999. Cervical electromyographic activity during low-speed rear impact. European Spine Journal 8 (2), 118–125. Myers, B.S., Winkelstein, B.A., 1995. Epidemiology, classification, mechanism, and tolerance of human cervical spine injuries. Critical Reviews Biomedical Engineering 23 (5–6), 307–409.
329
Obelieniene, D., Schrader, H., Bovim, G., Miseviciene, I., Sand, T., 1999. Pain after whiplash: a prospective controlled inception cohort study [see comments]. Journal of Neurology Neurosurgery and Psychiatry 66 (3), 279–283. Panjabi, M.M., Nibu, K., Cholewicki, J., 1998. Whiplash injuries and the potential for mechanical instability. European Spine Journal 7 (6), 484–492. Pintar, F.A., Yoganandan, N., Voo, L., 1998. Effect of age and loading rate on human cervical spine injury threshold. Spine 23 (18), 1957–1962. Schmitt, K. U., Muser, M.H., Niederer, P., 2001. A new neck injury criterion candidate for rear-end collisions taking into account shear forces and bending moments. Proceedings of the ESV Conference, Paper No. 124. Spitzer, W. O., Skovron, M. L., Salmi, L. R., Cassidy, J. D., Duranceau, J., Suissa, S., Zeiss, E., 1995. Scientific monograph of the Quebec Task Force on Whiplash-Associated Disorders: redefining ‘‘whiplash’’ and its management [see comments] [published erratum appears in Spine 1995 Nov 1; 20(21):2372]. Spine 20 (Suppl. 8), 1S–73S. Tencer, A.F., Mirza, S., Huber, P., 2003. A comparison of injury criteria used in evaluating seats for whiplash protection. ESV Proceedings. Viano, D.C., Davidsson, J., 2002. Neck displacements of volunteers, BioRID P3 and Hybrid III in rear impacts: Implications to whiplash assessments by a neck displacement criterion (NDC). Traffic Injury Prevention 3, 105–116. Vital, J.M., Senegas, J., 1986. Anatomical bases of the study of the constraints to which the cervical spine is subject in the sagittal plane. A study of the center of gravity of the head. Surgical and Radiologic Anatomy 8 (3), 169–173. Yoganandan, N., Pintar, F., Butler, J., Reinartz, J., Sances Jr., A., Larson, S.J., 1989. Dynamic response of human cervical spine ligaments. Spine 14 (10), 1102–1110. Yoganandan, N., Sances Jr, A., Pintar, F., Maiman, D.J., Reinartz, J., Cusick, J.F., Larson, S.J., 1990. Injury biomechanics of the human cervical column. Spine 15 (10), 1031–1039. Yoganandan, N., Pintar, F.A., Kleinberger, M., 1999. Whiplash injury. Biomechanical experimentation [editorial]. Spine 24 (1), 83–85. Yoganandan, N., Cusick, J.F., Pintar, F.A., Rao, R.D., 2001. Whiplash injury determination with conventional spine imaging and cryomicrotomy. Spine 26 (22), 2443–2448.
Journal of Biomechanics 39 (2006) 323–329 www.elsevier.com/locate/jbiomech www.JBiomech.com
Influence of the crash pulse shape on the peak loading and the injury tolerance levels of the neck in in vitro low-speed side-collisions Annette Kettler, Kai Fruth, Lutz Claes, Hans-Joachim Wilke Institute of Orthopaedic Research and Biomechanics, University of Ulm, Helmholtzstr. 14, D-89081 Ulm, Germany Accepted 14 November 2004
Abstract The aim of the present in vitro study was to investigate the effect of the crash pulse shape on the peak loading and the injury tolerance levels of the human neck. In a custom-made acceleration apparatus 12 human cadaveric cervical spine specimens, equipped with a dummy head, were subjected to a series of incremental side accelerations. While the duration of the acceleration pulse of the sled was kept constant at 120 ms, its shape was varied: Six specimens were loaded with a slowly increasing pulse, i.e. a low loading rate, the other six specimens with a fast increasing pulse, i.e. a high loading rate. The loading of the neck was quantified in terms of the peak linear and angular acceleration of the head, the peak shear force and bending moment of the lower neck and the peak translation between head and sled. The shape of the acceleration curve of the sled only seemed to influence the peak translation between head and sled but none of the other four parameters. The neck injury tolerance level for the angular acceleration of the head and for the bending moment of the lower neck was almost identical for both, the high and the low loading rate. In contrast, the injury tolerance level for the linear acceleration of the head and for the shear force of the lower neck was slightly higher for the low loading rate as compared to the high loading rate. For the translation between head and sled this difference was even statistically significant. Thus, if the shape of the crash pulse is not known, solely the peak bending moment of the lower neck and the peak angular acceleration of the head seem to be suitable predictors for the neck injury risk but not the peak shear force of the lower neck, the peak linear acceleration of the head and the translation between head and thorax. r 2004 Elsevier Ltd. All rights reserved. Keywords: Neck injury; Whiplash trauma; Crash pulse; Injury criterion; Injury threshold
1. Introduction The whiplash trauma of the cervical spine is still one of the most common injuries in traffic accidents. Whiplash injuries therefore continue to represent a substantial societal problem worldwide with associated costs, which are estimated at $4.5–10 billion annually in the US (Tencer et al., 2003; Yoganandan et al., 1999, 2001), at h1 billion annually in Germany (Hell and Langwieder, 1998) and in average at over CAD$3,800 per whiplash subject in Quebec (Spitzer et al., 1995). Corresponding author. Tel.: +49 731 500 23485; fax: +49 731 500 23498. E-mail address: [email protected] (A. Kettler).
0021-9290/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.jbiomech.2004.11.017
In view of these enormous costs the development of new seats, head restraints and vehicles is of high importance. To determine their performance in protecting the occupant against whiplash, the use of injury criteria such as the neck displacement criterion (NDC) (Viano and Davidsson, 2002), the Nkm criterion (Schmitt et al., 2001), or the neck injury criterion (NIC) (Bostro¨m et al., 1996) has been proposed. These parameters are either directly or indirectly related to the loading of the neck: The NIC is calculated from the velocity and acceleration of the head relative to the first thoracic vertebra, the Nkm criterion is based on the upper neck flexion/extension moment and shear force and the NDC is based on the angular and linear displacement response of the occipital condyles with respect to the first thoracic vertebra.
ARTICLE IN PRESS 324
A. Kettler et al. / Journal of Biomechanics 39 (2006) 323–329
To determine human tolerance levels for these injury criteria, however, is difficult since the human tolerance to complex three-dimensional (3D) loading is not yet well characterised (Myers and Winkelstein, 1995) and the rate-dependence of the human tolerance to whiplash loading is still unknown. This rate-dependence is of special interest since the real crash pulse shapes are varying considerably from pulses with high to pulses with low loading rates (Krafft et al., 1998, 2002). The aim of the present in vitro study therefore was to investigate the effect of the crash pulse shape on the peak loading and the injury tolerance levels of the human neck.
suspension cord
dummy head specimen
2. Materials and methods Twelve fresh frozen human cadaveric cervical spine specimens including the occiput (C0) and the first thoracic vertebra (T1) were selected for this study. Exclusion criteria were spinal disorders except for minor degeneration. The age of the donors was 81 years in mean. Before testing, the specimens were thawed at 4 1C and all soft tissue surrounding the discoligamentous spine was carefully removed. C0 and T1 were embedded in polymethylmethacrylate (PMMA) while the foramen magnum was aligned horizontally. Then, the specimens were mounted to a custom-made acceleration apparatus, that is basically composed of a sled, a railtrack and a pneumatic acceleration unit (Kettler et al., 2004). Since the trunk of a vehicle passenger is known to move during a real-life crash, these movements had to be simulated in the present experiment. This was accomplished by the use of a pivot table, which was located between the sled and the lower end of the specimens and which was allowed to rotate passively about an axis perpendicular to the direction of the acceleration. A dummy head made of wood and brass was developed to represent the average human head with a mass of 4.5 kg (Clemens 1972) and a humanlike outer geometry. This dummy head was rigidly fixed to the PMMA block on C0 with its centre of gravity located analogous to that of the human head with respect to the cervical spine (Vital and Senegas, 1986). Before acceleration the dummy head was suspended using a thin cord connected at one side to the upper surface of the dummy head and at the other side to the frame of the sled, approximately 10 cm above the head (Fig. 1). By the use of this cord the specimens could be balanced in the upright, neutral position before acceleration. At the beginning of the acceleration, however, the cord was cut to allow the head to move completely unconstrained. Each specimen was subjected to a series of incremental 901 side collisions from the right. The first impact
pivot table
Fig. 1. In the custom-made pneumatic acceleration apparatus the lower end of the specimens was fixed to a pivot table. On the occipital bone a dummy head was mounted, which had to be balanced with a suspension cord. This cord was cut at the beginning of each impact.
was characterised by a peak acceleration of the sled of approximately 1 g. In each following impact this acceleration was increased by another 1 g. The experiment was stopped as soon as any structural failure became macroscopically visible with the naked eye. After structural injury had occurred, anteroposterior and lateral radiographs were taken for documentation purposes and to assess bony injuries, which were not visible from outside. While the duration of the acceleration pulse of the sled was kept constant at approximately 120 ms for all specimens, its shape was varied: Six specimens were loaded with slowly increasing pulse, i.e. with a low loading rate, the other six specimens with a fast increasing pulse, i.e. with a high loading rate (Fig. 2). The uniaxial acceleration of the sled (EGE-73AE1100D1, measurement range7100 g, Entran Ludwigshafen, Germany), the 3D linear acceleration of the head centre of gravity (EGE-73AE1-100D1, measurement range 7100 g, Entran Ludwigshafen, Germany) and the 3D forces and moments between neck and pivot platform (lower neck load cell 4894J, measurement range 452 Nm, respectively, 13345 N, Denton COE GmbH, Heidelberg, Germany) were recorded simulta-
ARTICLE IN PRESS A. Kettler et al. / Journal of Biomechanics 39 (2006) 323–329
ase
fast inc
asled
asled
w slo
e fast decr
e
as
re
inc
rease
neously during impact. Presample filtering was carried out with an analogue Butterworth low-pass filter at a cut-off frequency of 250 Hz (acceleration of the sled) or 2000 Hz (all other parameters). Then, data were sampled at a rate of 9.6 kHz for each channel and digitally lowpass filtered at a cut-off frequency of 60 Hz (acceleration of the sled), 600 Hz (bending moment lower neck) or 1,000 Hz (accelerations of the head and shear force lower neck) (adapted to SAE J211 and DIN ISO 6487). The velocity and the distance covered by the sled and the head were obtained by integration of the filtered accelerometer data. According to the 6-accelerometer scheme, the angular acceleration of the head was calculated based on the data recorded by the linear accelerometers of the head centre of gravity and the data collected by another three linear accelerometers, which were placed on the surface of the head. Then, the following five loading parameters were determined (Fig. 3):
0
120
time [ms]
slo w
de
cre
0
325
the peak linear acceleration of the head along the local y-axis (ay head): the peak angular acceleration of the head about the local x-axis (ax head); the peak shear force of the lower neck along the local y-axis (Fy lower neck); the peak bending moment of the lower neck about the local x-axis (Mx lower neck); the peak translation between head and sled in direction of the acceleration of the sled (translation head-sled).
Injury tolerance levels were calculated for each of these five parameters. They were defined as the peak value of the respective parameter during the last acceleration, which not yet caused structural injury. Wilcoxon signed rank Tests were used to evaluate the effect of the crash pulse shape on the injury thresholds of the five loading parameters. Statistical significance was assumed at po0:05:
2. Results as
e
120
time [ms]
Fig. 2. Schematic sled acceleration curves. For six specimens the acceleration curve of the sled (asled) was characterised by a slow increase and a fast decrease (left) and for the other six by a fast increase and a slow decrease (right). In all cases the duration of the acceleration pulse was approximately 120 ms.
The shape of the acceleration curve of the sled only seemed to influence the translation between head and sled (Fig. 8, left) but none of the other four parameters (Figs. 4–7, left). In the case of the translation between head and sled, the low loading rate, i.e. the slowly increasing acceleration pulse of the sled, was associated with higher peak translations than the high loading rate (Fig. 8, left).
Fig. 3. The linear acceleration of the head along the local y-axis (ay ), the angular acceleration of the head about the local x-axis (ax ), the shear force of the lower neck along the local y-axis (F y ), the bending moment of the lower neck about the local x-axis (M x ) and the linear acceleration of the sled were measured simultaneously.
ARTICLE IN PRESS 326
A. Kettler et al. / Journal of Biomechanics 39 (2006) 323–329
Fig. 4. Left: Peak linear acceleration (ay) of the head correlated with the mean acceleration (amean) of the sled. Linear regressions across the impacts, which did NOT lead to structural injury (black rhombi for the low loading rate resp. grey quadrates for the high loading rate). Right: Peak linear acceleration of the head during the last impacts before the specimens were structurally injured. Mean value and standard deviation across the six specimens, which were accelerated with the slowly increasing pulse (left bar) and with the fast increasing pulse (right bar). P-value calculated with the Wilcoxon signed rank test.
Fig. 6. Left: Peak shear force (F y ) of the lower neck correlated with the mean acceleration (amean) of the sled. Linear regressions across the impacts, which did NOT lead to structural injury (black rhombi for the low loading rate, respectively, grey quadrates for the high loading rate). Right: Peak shear force of the lower neck during the last impacts before the specimens were structurally injured. Mean value and standard deviation across the six specimens, which were accelerated with the slowly increasing pulse (left bar) and with the fast increasing pulse (right bar). P-value calculated with the Wilcoxon signed rank test.
Fig. 5. Left: Peak angular acceleration (ax ) of the head correlated with the mean acceleration (amean) of the sled. Linear regressions across the impacts, which did NOT lead to structural injury (black rhombi for the low loading rate, respectively, grey quadrates for the high loading rate). Right: Peak angular acceleration of the head during the last impacts before the specimens were structurally injured. Mean value and standard deviation across the six specimens, which were accelerated with the slowly increasing pulse (left bar) and with the fast increasing pulse (right bar). P-value calculated with the Wilcoxon signed rank test.
The neck injury tolerance level for the angular acceleration ax of the head and for the bending moment M x of the lower neck was almost identical for both, the high and the low loading rate (Figs. 5 and 7, right). In contrast, the injury tolerance level for the linear acceleration ay of the head and for the shear force F y of the lower neck was slightly higher for the low loading rate than for the high loading rate (Figs. 4 and 6, right). For the translation between head and sled this difference was even statistically significant (p ¼ 0:0247) (Fig. 8, right). The structural injury was always localised in one of the segments C4–5, C5–6, C6–7, or C7–T1. Eleven out of the 12 specimen sustained a soft tissue rupture of the left facet joint capsule (impact from the right) and the intervertebral disc. In some cases an additional partial rupture of the right facet joint capsule and/or an additional fracture of the upper right articular process of the caudally adjacent vertebra was observed. On the radiographs no injury to the vertebral bodies or endplates were detected. In contrast to these 11 specimens, the remaining one only sustained a partial rupture of the left facet joint capsule without involvement of the disc. This specimen was excluded from evaluation, since otherwise the crash pulse shape would have no more been the only influencing variable and differences between the peak loading and between the injury thresholds would have also been attributable to differences in the extend of the injury.
ARTICLE IN PRESS A. Kettler et al. / Journal of Biomechanics 39 (2006) 323–329
Fig. 7. Left: Peak bending moment (M x ) of the lower neck correlated with the mean acceleration (amean) of the sled. Linear regressions across the impacts, which did NOT lead to structural injury (black rhombi for the low loading rate, respectively, grey quadrates for the high loading rate). Right: Peak bending moment of the lower neck during the last impacts before the specimens were structurally injured. Mean value and standard deviation across the six specimens, which were accelerated with the slowly increasing pulse (left bar) and with the fast increasing pulse (right bar). P-value calculated with the Wilcoxon signed rank test.
4. Discussion According to the degree to which the crash pulse shape influenced the peak loading and the injury thresholds of the neck, the five loading parameters, which were evaluated in the present study could be assigned to three different groups: 4.1. Group 1: no effect on peak values and injury thresholds The peak values and injury thresholds of the angular acceleration ax of the head and of the bending moment M x of the lower neck were not affected by the shape of the acceleration curve of the sled (Figs. 5 and 7). These two parameters therefore seem to be useful to predict the neck injury risk especially in cases where the shape of the crash pulse is not known. 4.2. Group 2: no effect on peak values, effect on injury thresholds
327
Fig. 8. Left: Peak translation between head and sled correlated with the mean acceleration (amean) of the sled. Linear regressions across the impacts, which did NOT lead to structural injury (black rhombi for the low loading rate , respectively, grey quadrates for the high loading rate). Right: Peak translation between head and sled during the last impacts before the specimens were structurally injured. Mean value and standard deviation across the six specimens, which were accelerated with the slowly increasing pulse (left bar) and with the fast increasing pulse (right bar). P-value calculated with the Wilcoxon signed rank test.
uniaxial tensile failure tests cervical spine ligaments (ligamentum longitudinale anterius and ligamentum flavum) were shown to have the inverse behaviour with higher failure forces under high loading rates than under low loading rates (Yoganandan et al., 1989) and human cadaver head–neck specimens tolerated higher compressive loads the faster the load was applied (Pintar et al., 1998). Using a linear regression model the authors could additionally show that an increasing age reduced the effect of loading rate and at approximately 82 years, this effect did no more exist. Thus, the correlation between the crash pulse shape and the injury thresholds for the linear acceleration ay of the head and for the shear force F y of the lower neck does not match the correlation expected considering the pertinent literature. This contradiction supports the assumption that neither the linear acceleration of the head nor the shear force of the lower neck were directly responsible for the occurrence of the structural neck injury. Thus, these two parameters do not seem to be appropriate to predict the neck injury risk. 4.3. Group 3: effect on peak values and injury thresholds
Similarly, the peak values of the linear acceleration ay of the head and of the shear force F y of the lower neck also did not depend on the crash pulse shape. However, their injury thresholds did (Figs. 4 and 6): slightly higher peak values were tolerated if the crash pulse had a slow increase than if it had a fast increase. In contrast, in
In contrast to the above-mentioned loading parameters, the peak values of the translation between head and sled depended on the shape of the acceleration curve: they were slightly higher for the low loading rate as compared to the high loading rate (Fig. 8). This effect
ARTICLE IN PRESS 328
A. Kettler et al. / Journal of Biomechanics 39 (2006) 323–329
can be explained by the viscous behaviour of human tissue, since viscous material is characterised by a ratedependency of the deformation magnitude. Besides the peak values, the injury tolerance levels also were affected by the crash pulse shape: higher peak translations between head and sled were tolerated for the low loading rate than for the high loading rate. In contrast to these results, the failure deformations of cervical spine ligaments under tension and human cadaver head–neck specimens under compression reported in the literature were similar for all tested loading rates (Yoganandan et al., 1989, 1990). Since the own results do not support these published data, the translation between head and neck is most probably not directly related to the local deformations which finally may cause injury. This assumption is supported by an in vitro study on cadaver cervical spine specimens. It was shown that the maximum head translation during rear end impacts increased with increasing sled acceleration while the maximum posterior head rotation decreased (Cholewicki et al., 1998). I.e. one and the same head translation may be associated with very different local deformations within the single cervical spine segments. The translation between head and sled therefore seems to be unsuitable to predict the local deformations and with it the neck injury risk especially if the shape of the crash pulse is unknown. These correlations between the crash pulse shape and the peak loading and injury tolerance levels of the neck primarily apply to structural injuries but might also be extrapolated to non-structural whiplash-like injuries under the following assumptions: First, whiplash injuries have to be caused mechanically by inertial loads exerted on the cervical spine. Even though some researchers hypothesise that not the spine itself but rather the neck muscles or the nerve roots are the primary site of injury (Bostro¨m et al., 1996; Brault et al., 2000) or that whiplash mainly is a psychological or societal problem (Castro et al., 2001; Obelieniene et al., 1999), most agree with this assumption. Second, structural injuries have to be preceded by subfailure injuries, which on their part have to be responsible for the symptoms the whiplash patient is suffering from. This assumption is supported by in vitro studies, which all showed that structural injuries actually are preceded by functional injuries, i.e. local instabilities without any detectable structural damage (Hartwig et al., 2004; Panjabi et al., 1998). Third, side collisions have to cause injuries which are similar to those caused by rear-end or frontal collisions. This assumption is partially supported by in vitro experiments on rear end collisions (Panjabi et al., 1998). Kind (soft tissue injury) and localisation (lower cervical spine) of the injuries were similar to the present study on side collisions. A limitation of the present study is the fact that the stabilising effect of the neck muscles was missing. During
whiplash, this stabilising effect is mainly composed of a passive and an active component. While the passive component permanently acts even if the impact is not expected, the effect of the active component depends on a certain reaction time. In low-speed rear-end collisions with volunteers the average reaction time of superficial and deep neck muscles was found to range between 73 and 175 ms referred to the beginning of the movement of the sled (Magnusson et al., 1999). Thus, even if another 100 ms were needed for electromechanical coupling, single muscle fibres would already actively stabilise the cervical spine during the phase of maximum stress and strain, which, in the present study, occurred approximately 150 ms after the onset of the acceleration pulse. Provided that this actively and the passively stabilising effect of the neck muscles only affects the magnitude but not the quality of the loading of the neck, the correlations between the crash pulse shape and the peak loading and injury tolerance levels made in the present study should also apply to real-life crashes, whereas the absolute values of failure do definitely not. Thus, the data of the present study should not be used to define any real-life injury thresholds. In conclusion, the crash pulse shape seems to influence the peak value of the translation between head and thorax (a high loading rate is associated with a small translation and vice versa) but not the peak values of the linear and angular acceleration of the head and the shear force and bending moment of the lower neck. The crash pulse shape also seems to take influence on the injury thresholds for the peak translation between head and thorax. Higher peak translations are tolerated if the loading rate is low than if it is high. The injury thresholds for the peak shear force of the lower neck and the peak linear acceleration of the head also seem to be slightly influenced by the crash pulse shape, the injury threshold for the peak bending moment of the lower neck and the peak angular acceleration of the head however, do not. Therefore, if the shape of the crash pulse is not known, only the peak bending moment of the lower neck and the peak angular acceleration of the head seem to be suitable predictors for the neck injury risk but not the peak shear force of the lower neck, the peak linear acceleration of the head and the translation between head and neck. Acknowledgement The authors gratefully acknowledge the German Research Council (Deutsche Forschungsgemeinschaft, DFG, HA 3276/1-1) for financial support. References Bostro¨m, O., Svensson, M., Aldman, B., Hansson, H., Haland, Y., Lo¨vsund, P., Seeman, T., Suneson, A., Sa¨ljo¨, A., O¨rtengren, T., 1996. A new neck injury criterion candidate based on injury
ARTICLE IN PRESS A. Kettler et al. / Journal of Biomechanics 39 (2006) 323–329 findings in the cervical ganglia after experimental neck extension trauma. Proceedings of the IRCOBI Conference, 123–136. Brault, J.R., Siegmund, G.P., Wheeler, J.B., 2000. Cervical muscle response during whiplash: evidence of a lengthening muscle contraction. Clinical Biomechanics (Bristol, Avon) 15 (6), 426–435. Castro, W.H., Meyer, S.J., Becke, M.E., Nentwig, C.G., Hein, M.F., Ercan, B.I., Thomann, S., Wessels, U., Du Chesne, A.E., 2001. No stress—no whiplash? Prevalence of ‘‘whiplash’’ symptoms following exposure to a placebo rear-end collision. International Journal of Legal Medicine 114 (6), 316–322. Cholewicki, J., Panjabi, M.M., Nibu, K., Babat, L.B., Grauer, J.N., Dvorak, J., 1998. Head kinematics during in vitro whiplash simulation. Accident Analysis and Prevention 30 (4), 469–479. Clemens, H.J., 1972. Weight of the head in man—a biomechanical problem. Archiv fur Orthopadische und Unfallchirurgic 73 (3), 220–228. Hartwig, E., Kettler, A., Schultheiss, M., Kinzl, L., Claes, L., Wilke, H.J., 2004. In vitro low-speed side collisions cause injury to the lower cervical spine but do not damage alar ligaments. European Spine Journal (online first). Hell, W., Langwieder, K., 1998. Epidemiologische Daten zur HWS— Beschleunigungsverletzung. Die Notwendigkeit eines verbesserten Diagnosestandards und bessere Pra¨vention bei Auffahrunfa¨llen. Hefte zu Der Unfallchirurg 272, 80–82. Kettler, A., Schmitt, H., Simon, U., Hartwig, E., Kinzl, L., Claes, L., Wilke, H.J., 2004. A new acceleration apparatus for the study of whiplash with human cadaveric cervical spine specimens. Journal of Biomechanics 37 (10), 1607–1613. Krafft, M., Kullgren, A., Tingvall, C., 1998. Crash pulse recorders in rear impacts—real life data. ESV Proceedings. Krafft, M., Kullgren, A., Ydenius, A., Tingvall, C., 2002. Influence of crash pulse characteristics on whiplash associated disorders in rear impacts—crash recording in real life crashes. Traffic Injury Prevention 3, 141–149. Magnusson, M.L., Pope, M.H., Hasselquist, L., Bolte, K.M., Ross, M., Goel, V.K., Lee, J.S., Spratt, K., Clark, C.R., Wilder, D.G., 1999. Cervical electromyographic activity during low-speed rear impact. European Spine Journal 8 (2), 118–125. Myers, B.S., Winkelstein, B.A., 1995. Epidemiology, classification, mechanism, and tolerance of human cervical spine injuries. Critical Reviews Biomedical Engineering 23 (5–6), 307–409.
329
Obelieniene, D., Schrader, H., Bovim, G., Miseviciene, I., Sand, T., 1999. Pain after whiplash: a prospective controlled inception cohort study [see comments]. Journal of Neurology Neurosurgery and Psychiatry 66 (3), 279–283. Panjabi, M.M., Nibu, K., Cholewicki, J., 1998. Whiplash injuries and the potential for mechanical instability. European Spine Journal 7 (6), 484–492. Pintar, F.A., Yoganandan, N., Voo, L., 1998. Effect of age and loading rate on human cervical spine injury threshold. Spine 23 (18), 1957–1962. Schmitt, K. U., Muser, M.H., Niederer, P., 2001. A new neck injury criterion candidate for rear-end collisions taking into account shear forces and bending moments. Proceedings of the ESV Conference, Paper No. 124. Spitzer, W. O., Skovron, M. L., Salmi, L. R., Cassidy, J. D., Duranceau, J., Suissa, S., Zeiss, E., 1995. Scientific monograph of the Quebec Task Force on Whiplash-Associated Disorders: redefining ‘‘whiplash’’ and its management [see comments] [published erratum appears in Spine 1995 Nov 1; 20(21):2372]. Spine 20 (Suppl. 8), 1S–73S. Tencer, A.F., Mirza, S., Huber, P., 2003. A comparison of injury criteria used in evaluating seats for whiplash protection. ESV Proceedings. Viano, D.C., Davidsson, J., 2002. Neck displacements of volunteers, BioRID P3 and Hybrid III in rear impacts: Implications to whiplash assessments by a neck displacement criterion (NDC). Traffic Injury Prevention 3, 105–116. Vital, J.M., Senegas, J., 1986. Anatomical bases of the study of the constraints to which the cervical spine is subject in the sagittal plane. A study of the center of gravity of the head. Surgical and Radiologic Anatomy 8 (3), 169–173. Yoganandan, N., Pintar, F., Butler, J., Reinartz, J., Sances Jr., A., Larson, S.J., 1989. Dynamic response of human cervical spine ligaments. Spine 14 (10), 1102–1110. Yoganandan, N., Sances Jr, A., Pintar, F., Maiman, D.J., Reinartz, J., Cusick, J.F., Larson, S.J., 1990. Injury biomechanics of the human cervical column. Spine 15 (10), 1031–1039. Yoganandan, N., Pintar, F.A., Kleinberger, M., 1999. Whiplash injury. Biomechanical experimentation [editorial]. Spine 24 (1), 83–85. Yoganandan, N., Cusick, J.F., Pintar, F.A., Rao, R.D., 2001. Whiplash injury determination with conventional spine imaging and cryomicrotomy. Spine 26 (22), 2443–2448.