Subject-adjustable computational modelling for the analysis of human head–neck motions in rear-end impacts

Whiplash Associated Disorder (WAD) is a general term used to describe minor injuries to the neck, mostly as a result of a rear-end motor vehicle collision. Although the injury is defined as minor, the long-term symptoms such as neck pain, stiffness, headache, or concentration difficulties, result in high costs to the economy, healthcare services and individuals. Consequently, there has been significant amount of research undertaken to understand and prevent WAD, covering experimental and computational studies. However, whiplash injuries are difficult to detect since diagnostic tools such as X-rays, CT (Computed Tomography) scans or MRI (Magnetic Resonance Imaging) are not suitable to identify the location or the extent of the injury. Also, the injury mechanisms are not fully understood; hence mathematical criteria are used as surrogates to estimate the likelihood of injury.

In the present research, a biofidelic, subject-adjustable head-and-neck model (i.e. the model is adjustable for individual subject characteristics) has been developed for rear-end impact whiplash analysis. Existing literature is used to develop the overall research framework (methodology), which has three main objectives: first to explain the importance of personalised protection investigations, second to evaluate the suitability of existing data for a subject-adjustable model, and third to define the required steps in the design of such a model.

To generate the geometry of the model, previously published cascading equations capable of predicting the main vertebrae dimensions based on the subject characteristics age, gender and height are used. Also, in line with previous work, seven cervical neck segments represent the seven cervical vertebrae and all surrounding cervical tissues properties. The mass and moment of inertia properties of each segment are lumped into each respective segment. The intervertebral behaviour for two adjacent segments is described by non-linear spring-damper functions, which change according to subject specific characteristics. The model is driven by specification of the first thoracic vertebra (T1) motion.

The model combines existing data and methods from different sources, utilising available data in the public domain. New procedures and techniques are incorporated to create a homogeneous model, which is adaptable to a wide range of subjects. The developed computational model is not simply a linear scaling of a master-model to other dimensions, but rather uses prediction equations to create the desired anthropometric model. The anthropometric model predictions for body part dimensions and inertia properties are successfully verified using anthropometric surveys available in the literature.

Using lumped and non-linear stiffness and damping equations for the intervertebral joints, and without modelling separate muscles, the model is dynamically calibrated for different experimental sled test data available in the open domain. The joint equations and their coefficients are derived based on published joint data measurements on Post Mortem Human Subjects (PMHS); a scaling of these coefficients is applied to match the overall head-and-neck kinematics of the computational model to the experimental sled test kinematics. For each experimental study, the global head kinematics of the model was calibrated successfully to mimic the head kinematics.

The model has been modelled to represent subjects with different anthropometric characteristics, involving a novel relationship between intervertebral joint coefficients and anthropometric subject specifications. The observed effect of each change of anthropometric subject characteristic is evaluated independently using time-history diagrams; then the observed effect of multiple changes of anthropometric subject characteristics is assessed using multi-dimensional response surfaces for the response’s highest magnitude.

The analysis of the proposed model has revealed that existing work involving the use of lumped parameter models is not as robust as claimed. This is because existing work has always been evaluated using a low number of validation graphs, i.e. using only the graphs which gave good validation results. The proposed model has been comprehensively evaluated and its limitations are addressed.

The developed model had to merge different studies (different ethnical backgrounds, different subject types, etc.) together to create an adjustable model; this is because of the limited available data. The final model is the most homogeneous model currently possible. In addition, there is also limited relevant experimental data for full validation of the model, which is not ideal. Nevertheless, reliable results for the comparison of global head kinematics compared with several experimental sled test studies have been obtained for the average male subject model. Also, using the proposed model the dynamic effects resulting from anthropometric subject differences have been evaluated; these effects are almost perfectly linear relationships for each subject characteristic change.

Potential applications for the developed model are the injury assessment based on mathematical whiplash injury criteria, head-restraint optimisation to minimise injury risk and the improvement of neck biofidelity in anthropometric test devices.