Prediction of Neural Space Narrowing and Soft Tissue Injury of the Cervical Spine Concerning Head Restraint Arrangements in Traffic Collisions

Common quantitative assessments of neck injury criteria do not predict anatomical neck injuries and lack direct relations to design parameters of whiplash-protection systems. This study aims to provide insights into potential soft tissue-level injury sites based on the interactions developed in-between different anatomical structures in case of a rear-end collision. A detailed finite element human model has exhibited an excellent biofidelity when validated against volunteer impacts. Three head restraint arrangements were simulated, predicting both the kinematic response and the anatomical pain source at each arrangement. Head restraint’s contribution has reduced neck shear and head kinematics by at least 70 percent, minimized pressure gradients acting on ganglia and nerve roots less than half. Posterior column ligaments were the most load-bearing components, followed by the lower intervertebral discs and upper capsular ligaments. Sprain of the interspinous ligamentum flavum at early stages has caused instability in the craniovertebral structure causing its discs and facet joints to be elevated compressive loads. Excessive hyperextension motion, which occurred in the absence of the head restraint, has promoted a stable avulsion teardrop fracture of the fourth vertebral body’s anteroinferior aspect and rupture the anterior longitudinal ligament. The observed neck injuries can be mathematically related to head–torso relative kinematics. These relations will lead to the development of a comprehensive neck injury criterion that can predict the injury level. This, in turn, will impose a significant impact on the design processes of vehicle anti-whiplash safety equipment.


Introduction
The cervical spine is among the most vulnerable road injuries with a high risk of morbidity [1]. Whiplash trauma is a neck injury due to forceful, rapid back-and-forth movement of the neck that is often associated with rear-end traffic collisions. The reported annual treatment costs in the United States are as high as $5.2 billion [1]. In 2013, the United Kingdom announced that 76% of the insurance claims were associated with whiplash injuries, compared to an average of 48% throughout the rest of Europe [2]. Around 80.9% of patients had a mild neck injury following a traffic accident, and 19.1% had severe health complications [3]. Out of those, soft tissue injuries were the leading source of pain (16.2%), followed by disabilities due to spinal cord trauma (3.8%) and fracture-dislocations (1.3%).
The key to reducing the whiplash's motion is by minimizing the relative kinematics between the occupant's head and torso. The primary countermeasure is the head restraints equipped in the vehicle to limit the neck's hyperextension motion relative to the torso. However, their effectiveness is highly influenced by their position relative to the occupant's head. Unfortunately, passengers rarely adjust their head restraint to the recommended position due to low public awareness [4]. Trempel and Edward [5] have reported a reduction of 11.2% in neck injuries for head restraints with good rating to poorly rated head restraints according to Insurance Institute for Highway Safety (IIHS). They have analyzed data from 36 insurance companies over 17 US states for 1754 vehicles between 2001-2014 models, and the claims were added up to 603,755 caused by rear-end collisions.
Traffic accident patients often have persistent chronic complaints due to whiplash, as addressed by several clinical reports [6]. Their findings suggested that the central nervous system's hyperexcitability, which might start three to six months after the initial injury, seems to play the primary role in sustaining those pain complaints [6]. Nevertheless, the underlying anatomical pain mechanisms are still vague, even though several stimuli (mechanical, thermal, electrical) were used to assess the injury [6]. This is because whiplash trauma is sensitive to several factors related to the impact scenario and human parameters, such as age and gender [7].
Neck injury criteria have been extensively used as a tool to predict the injury level in the head-neck-torso structure to impact while adjusting the head restraint to various positions. These criteria were developed to exclusively provide injury assessment based on the spine relative kinematic or kinetic responses. The finite element (FE) method is commonly utilized since volunteer tests are limited to low impacts, and cadaver or crash dummy tests do not represent the accurate active responses [8]. However, utilizing FE analysis aims to provide insights into tissue-level injury mechanisms and thresholds rather than general assessment. Although some studies have provided details of failed cervical components at a segmental level, only a few have used an entire detailed human model [9,10]. The head-neck-torso interactions play a significant role in altering tissue injury sites relative to head restraint arrangement [8]. Therefore, it is necessary to utilize a convenient detailed human model to expand the knowledge base and explain pain's physiological sources as the causing scenario conditions change.
This study has performed numerical FE simulations using an entire human model that comprises skeleton, muscular, ligament, and nervous structures to predict the cervical spine kinematic response to rear-end collision. This, in turn, will provide insights into potential soft tissue-level injury sites based on the interactions developed in-between different anatomical structures. To explore the range of injuries that could develop; as a result, a rear-end impact and better understand the interactions between cervical components, three head restraint arrangements were simulated; properly positioned, poorly adjusted, and without a head restraint. Those findings may potentially contribute to anticipating spinal injury's physiological causes and developing a comprehensive neck injury criterion that provides protective measures in automotive crash scenarios.

Model Development
This study utilized nonlinear dynamic analysis explicit solver in LS-DYNA ® FE software to simulate a rear-end vehicle collision with the Total Human Model for Safety (THUMS v.4) (Figure 1). This human model consists of a skeletal structure, muscular structure, organs, tendons, ligaments, and nerve system. The solid elements were used to model the bulky muscles, intervertebral discs (IVDs), and cancellous bones, whereas ligaments, spinal cord, and cortical bones were modeled with shell elements. The majority of the components were meshed with hexahedron elements, except for cortical shells and thin muscles composed of a tetrahedron and one-dimensional elements. Table 1 presents the material properties of the cervical spine components. Johnson and Cook elasto-plastic model was defined for the cortical and cancellous bony structures of the occipital condyle (OC), the entire cervical vertebra (C1-C7), and the first thoracic vertebra (T1) (Figure 1c) [11]. Their intervertebral nucleus pulposus and annulus fibrosus were defined as fluid and Hill hyper-elastic materials, respectively [12,13]. Anterior longitudinal ligament (ALL), posterior longitudinal ligament (PLL), intertransverse ligament (ITL), ligamentum flavum (LF), capsular ligament (CL), and interspinous ligament (ISL) were modeled using generalized Maxwell and Kelvin-Voigt viscoelastic equations [14,15]. Feng hyper-elastic behavior was used to describe the thick bulky neck muscles, whereas thin muscles were modeled using nonlinear curves [16].

Cortical
Johnson and Cook Elasto-plastic ρ = 1.82, σ f = 155, E = 16800, σ y = 110, ν = 0.3, b = 100, n = 0.1, c = 1 [11] Cancellous Johnson and Cook Elasto-plastic ρ = 0.17, σ f = 2.23, E = 100, σ y = 1.92, ν = 0.29, b = 20, n = 1, c = 1 [11] Nucleus pulposus Fluid ρ = 1, K = 1720 [12] Annulus fibrosus Hill hyper-elastic Spinal cord Quasilinear viscoelastic The seat's base and backrest cushioning material were made of SAF 6060 polyurethane foam [18], whereas a softer foam was assigned to the head restraint [19]. Ogden hyper-elastic foam material model was used to describe polyurethane foam behavior. Table 2 summarizes seat material properties and overall seat dimensions. The seat contained wings padding the backrest and base to arrest the human body during the impact properly. A three-point seatbelt and a rigid floorpan were added to support the body. The velocity boundary condition was applied horizontally to the rigid lower seat base, recliner joint, and the backrest and head restraint shell. The seatbelt retractor was initially locked without pre-tensions applied. The seat's lateral and vertical motions were constrained, whereas no constraints nor forces were applied to the human model.

Parametric Study
Three arrangements were developed (i.e., A (good), B (poor), and C (without head restraint)) to investigate the effect of head restraint position on cervical spine injuries. The head restraint level was initially 10 mm above the head. Arrangement A had a properly adjusted head restraint at 20 mm away from the head, whereas Arrangement B was poorly positioned at 100 mm. Contrarily, Arrangement C did not feature a head restraint. A 10 g pulse velocity testing protocol was prescribed to all arrangements adopted from international insurance whiplash prevention working group center under research council for automobile repairs (RCAR-IIWPG) [20]. Besides, gravitational acceleration (9.81 m/s 2 ) was included to obtain realistic results. The backrest and base were both inclined by 25 • from the vertical and horizontal axes, respectively. This is the proper angle to reduce the submarining risk below the seatbelt in the event of a frontal impact, resulting in internal and spinal injuries [21]. The results were recorded until the head rebounds from the head restraint or after a period of 300 ms from the beginning of the acceleration pulse, whichever comes first.

Model Validation
Validation of the human model responses has been extensively investigated in the literature. Its biomechanics were compared against several cadaver and volunteer human responses reported in the literature [9]. Besides those reported, an additional validation was achieved in this study by comparing its global relative kinematics against the volunteer rear-end impact test conducted by Linder et al. [22]. In their study, eight male volunteers were subjected to a 7 km/h velocity rear-impact profile and a mean acceleration of~2 g. The test was performed using an adaptive seat with a head restraint backset gap of 150 mm. It is worth noting that a restraining harness was used instead of a 3-point seatbelt to ensure the volunteers' safety at a high head restraint backset gap. Figure 2 presents the head retraction (x-displacement) movement concerning the torso and the head's peak angular displacement. The resulted kinematic responses were found to be within the average motion corridors of the human subjects. Head retraction reached a maximum of 49.5 mm after 224 ms, whereas peak head extension was 33.1 degrees at 179 ms.

Figure 2.
Comparison between the developed model results and Linder et al. [22] reported data of eight male volunteers in terms of (a) head retraction in horizontal axis relative to the first thoracic vertebra and (b) head angular displacement.

Head Retraction Phase
The overall cervical spine kinematics were divided into three distinct phases; the S-shape, hyperextension, and rebound phases [8]. The human body interactions with the accelerating seat, restraining seatbelt, and head restraint support are illustrated in Figure 3a. At the beginning of the impulse, the torso gradually absorbs kinetic energy while indenting into the seatback. Consequently, the torso begins to accelerate upward, depending on the passenger's seatback angle and material properties [9]. Meanwhile, the unsupported head does not move accordingly and lags behind the torso due to its inertia. This, in turn, causes the neck to endure combined flexion and extension moments at the upper and bottom levels, respectively, to form what is known as an S-shape posture. In comparison between the head restraint arrangements, the head retraction increased linearly with the head restraint-to-head gap distance (Figure 3b). Besides, the S-shape curvature becomes more severe when the head retracts further behind the torso. Arrangement A was more responsive in providing head support at an early stage before excessive head retraction (<12 mm). Therefore, the cervical structure sustained without failure against the stresses, as shown in Figure 4, A-ii. On the contrary, the poorly adjusted head restraint in Arrangements B and C has permitted the head to retract beyond 52 mm and 76 mm, respectively.
The cervical spine elements started to collapse and tear due to the vertebra's excessive relative rotational movements. Firstly, the upper flexion moment forced the posterior ligaments (LF and ISL) to stretch beyond their physiological limit and fail in the craniovertebral region (Figure 4, B&C-ii). Accordingly, higher stresses were shifted to the CL and ITL ligaments due to increased S-shaped curvature. The ITL ligament was able to withstand those stresses due to its elasticity and strength [23]. Unlike the ITL, however, the CL ligaments were disrupted at C1-C2 facet joints. Secondly, the ISL ligament undertook significant shear stresses at the bottom spinous process (C6-C7-T1) owing to high neck contraction displacement due to extension motion. It is a very delicate ligament and will require long-term treatments to recover [14]. Lastly, shear stresses induced on the bottom IVDs have reached a significant level of more than 2.4 MPa (Figure 4, C-ii). Those will most probably produce damage to the outer annulus' neural arch and delamination [12].

Hyperextension Phase
Once the kinetic energy is transferred to the head through the neck during the S-shape phase, it extends behind the torso. This extension moment results from the lever arm between the head's center of mass and the OC's pulling point. This, in turn, stretches the anterior ligaments and the annulus of the IVDs. Therefore, injuries are mostly limited to or originate from the spine's anterior column [8]. Patients that have experienced rear-end collision often sustain anterior ligament ruptures and disk separation, which lends weight to the hyperextension mechanism [12]. The degree of injury and symptom chronicity depends on the amount of extension that occurs. Figure 3b indicates that the neck experiences complex motion, in Arrangement C, comprised of hyperextension (>56 • ) combined with elevated compressive displacement (>120 mm). The extent of those relative motions is well beyond the neck's threshold tolerance levels [14]. Thus, the presence of head restraint in the vehicle is essential for the safety of the occupant. Although the head restraint in Arrangement B was misplaced, it had a significant impact in lowering the peak neck compression and extension by 72% and 91%, respectively (Figure 3b). On the other hand, Arrangement A has eliminated the relative kinematics since it supported the head in the former phase.
Without head restraint support, the posterior and anterior spinal columns were subjected to sufficient compression and tension loads, respectively, to cause sprains and fractures in its osteoligamentous structure (Figure 4, C-iii). The former stresses were higher than 120 MPa owing to the interface between the C3-C6 spinous process (Figure 5c). Accordingly, microcracks initiation within their cancellous bones were likely to occur [24]. Besides, since the ISL, LF, and CL ligaments have failed in the craniovertebral region, the relative motions between the atlas and axis vertebras were no longer fully constrained. Hence, contact stresses between them were generated, as seen in Figure 4, C-iii. These stresses have up to a 20% probability of causing an unstable type II odontoid fracture in the C2 dens or initiating cancellous bone cracks in C1 and facet joints [25]. In this case, the immediate intervention of anterior screw fixation surgical treatment is required to provide rigid support for the fracture and unify with the preservation of atlantoaxial rotation [25]. Disruption of the ALL ligament occurred at C3-C5 levels due to exposure to tensile stresses beyond 30 MPa, which exceeds its material capabilities (Figure 5a). Furthermore, stable avulsion fracture from the attachments to the inferior corner of the C4 vertebral body (Figure 4, C-iii). As stated by Yoganandan et al. [26], an added range of movement due to neck hyperextension can eventually cause long-term segmental spinal instability and spine degeneration. The PLL and CL ligaments did not endure harmful stresses since they are located on the middle neutral column, and thus their failure is not foreseen in this stage.
Regarding the IVDs, they undertook a complex combination of stresses beyond their physiological limit (greater than 3 MPa), especially at the C3-C4 interface (Figure 5b) [12]. Therefore, the probability of promoting radial fissures and disc prolapse (slipped disc) is very high [12]. It is worth mentioning that those injuries are hazardous and will require years to heal. All the previously mentioned neck injuries can be avoided by adjusting the head restraint properly, similar to Arrangement A. This allows the head restraint to absorb the head's kinetic energy and lower its relative rotational motion with the torso, which neglected neck extension.
Peak levels of neural space narrowing were observed at the upper C2-C4 intervertebral segments during the hyperextension phase. Lateral structural displacements have narrowed the canal diameter by a peak of 1.8 mm at the C3-C4 level in Arrangement C (Figure 6a). This, in turn, has built hydrodynamic pressure gradients in the spinal cord [15]. Pressure levels reached a maximum of 114 kPa were observed in the OC-C1 segment (Figure 7c). This level exceeds the tolerance threshold and will most likely contribute to chronic dysfunction of the nerve roots involved [27]. The foramen width, measured at 45 • to the midsagittal plane, was reduced by 0.87 mm at the C2-C3 level (Figure 6b). Consequently, the ganglia and nerve roots were subjected to potential injury. Since the spinal fluid is incompressible, this causes an outward displacement of the spinal canal's contents [27]. Blood within the anterior internal venous plexus moves through the foramen to the anterior external venous plexus [27]. At the same time, cerebrospinal fluid enters through the nerve root sleeves.
The increased amount of blood and cerebrospinal fluid leads to additional compressive loads being applied to nerve roots. Further, the craniocervical junction ligaments, such as the alar and transverse ligaments, were protruded and buckled inwardly, which most likely applies additional pressure on those roots. The head restraint in the other arrangements with head restraint involvement has minimized these pressure gradients by more than 50 percent (Figure 7). The fact that occupants of vehicles involved in rear-end collisions often experience neck and shoulder pain adds more credit to the previous causes.

Neck Flexion Phase
The head rebound phase begins when the head restraint releases its stored energy after the maximum indentation. This energy causes the head to rebound at a specific rate, and at the same time, the cervical spine straightens [26]. This rate is influenced by the seat design and its material properties. In Arrangement C, the rebound interval started when the seatbelt restrained the torso rebounding motion after the neck extension at 136 ms (Figure 3a). The head's relative velocity with the torso was lower than the other arrangement since the head restraint was not present to provide the head with additional energy. Simulated results were terminated at the end of the rebound phase for A (good), B (poor), and C (without head restraint) Arrangements at 120, 180, and 250 ms, respectively.
Variations in the neck tension between the three head restraint arrangements were insignificant due to the absence of airbag deployment (Figure 3b). Overall, a maximum elongation of less than 19.6 mm was recorded within acceptable neck displacement boundaries. However, the peak head flexion angle of 32 degrees relative to the T1 was beyond the safe range of motion [11]. Consequently, the posterior cervical ligaments were the most load-bearing components, followed by the cancellous bone, CL, and IVDs. The results suggest that ISL and LF ligaments were strained at a high rate, especially in Arrangements A and B, and thus failed at multiple bottom levels (Figure 4, A&B-iv). Similar observations were reported from experimentation on intact biofidelic fresh-frozen cadaveric cervical spines [28].
The middle and lower cervical spine (C3-C4 through C6-C7) were the most vulnerable segments for ISL and LF failure at low impacts and throughout the entire spine at 10 g impact. In general, the ISL was at higher elongations than LF, possibly because they are located at a greater distance from the center of rotation during hyperflexion. There was no sign of a teardrop fracture of the anteroinferior aspect of the C2-C4 vertebral bodies since the hyperflexion motion was combined with tension rather than compression displacements. However, their corresponding IVD was exposed to pinching stresses from the anterior annular fiber layer [12] (Figure 4, C-iv). Those stresses were higher than the injury tolerance of 2.7 MPa [12].
Failure of the posterior ligaments has added additional structural instability resulting in increased compressive loadings on the anterior IVDs and leading to disc and facet degeneration. As a result, chronic pain will appear from facet joint osteoarthritis, as demonstrated by radiographic studies [28]. Fortunately, those injuries were not accommodated by a displacement of the posterior portion of its body into the spinal canal. Therefore, injury to the spinal cord was not predicted in this stage.
Stresses on the CL were the greatest at C4-C6 levels and infrequently exceeded the sub failure injury threshold at 3.5 MPa. Therefore, they are less likely to be injured. Those injuries do not lead to death, but the joints cartilage may bleed and get damaged, which will promote chronic long-term neck pain [12]. They are more vulnerable when the flexion motion is combined with an axial rotational moment, as shown at the end of Arrangement B in Figure 4, B-iv. These rotational moments most likely appear from the three-point restraining seatbelt, where one shoulder of the torso is fully restrained, unlike the other. Studies predicted that they would carry up to 70 percent of the total load, which leads to failure in its structures [11]. Since this stage was comprised of chronic injuries at multiple levels, an airbag deployment system is mandatorily equipped in vehicles.
It is worth mentioning that ligaments and bony injuries predicted by the FE model provided insights into the neck response to rear-end collision and highlighted the most vulnerable cervical components to fail. However, these results should not be considered a final statement before further investigation utilizing extensive clinical trials. Several limitations were associated that may or may not affect the biofidelity of the study. For instance, muscle pre-tension was not defined in the human model to establish equilibrium spinal posture and active structural responses. Besides, the availability of cervical components' material properties and injury thresholds were limited with some diversity in their values, which may have caused prediction variations in the prolonged, painful response.

Conclusions
Overall, head restraints have exhibited adequate neck protection towards limiting peak shear and extension displacements during the retraction stage. Even the poorly adjusted one has reduced the relative kinematics by 72 percent. Tissue-level injuries were mainly caused by hyperflexion motion during the rebound phase regardless of the head restraint position. That is true, especially when this motion was combined with axial rotation, due to the three-point restraining seatbelt, where further loads were subjected to the intertransverse and capsular ligaments. The interspinous and ligamentum flavum ligaments were the most vulnerable cervical components in the entire spine. During the s-shape stage, their sprain occurred when the head restraint was poorly positioned and thus caused relative motion instabilities between the atlas and axis vertebra. This, in turn, caused the intervertebral discs, capsular ligaments, and facet joints at the craniovertebral level to undergo high compressive stresses beyond their physiological injury threshold. A stable extension-teardrop avulsion fracture of the anteroinferior aspect of the fourth vertebral body occurred when the head restraint was absent. Accordingly, the anterior longitudinal ligament was ruptured, whereas the posterior longitudinal ligament remained intact. Lastly, the spinal cord at the craniocervical level underwent high hydrodynamic pressure gradients due to the structure's unstable relative motions. Moreover, the ganglia and nerve roots were subjected to sufficient compressive loads to cause injury due to narrowing in their neural space. Consequently, chronic shoulder pain and numbness are common symptoms of post whiplash trauma due to the dysfunction of those roots.
These findings illustrate the complexity of the injury that extends to not only damage to cervical muscles and ligaments but also its nervous system and vertebral body fractures. These injuries are related to head-torso relative kinematics that is highly dependent on the head restraint position, as demonstrated in this study. A more representative neck injury criterion can be developed as per these relations that can predict the level of the injury. Such criterion will provide significant insights into the effect of various safety design parameters in limiting whiplash injuries.
A properly adjusted head restraint was very effective in limiting potential whiplash injuries. However, having the need to manually adjust the head restraint every time an occupant sits in the vehicle or change his posture while driving is not practical. Unfortunately, active head restraint systems are being featured only on luxury cars as an advanced safety pack with other features such as road-assist lane monitoring. There is a need to make it mandatory to equip every new vehicle with an active head restraint system and develop a retrofit system for the older models.

Conflicts of Interest:
The authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest or non-financial interest in the subject matter or materials discussed in this manuscript. Ethical approval was not required since this study did not involve human nor animal subjects. The data collected in this study is not publicly available due to research work copyrights.