5 Modelling of the Brain for Injury Simulation and Prevention
125
animal experimental models (a small and a large one) need to be conducted to
develop a scaling method based on parameters such as age, gender, size, material
properties, etc. Preferably, a third animal model should be tested to validate the
accuracy of the scaling law and to develop a scaling method to estimate the tissue
level injury threshold in humans. Note that this method does not account for
species differences in tolerance and may not provide a direct correlation between
size and tolerance.
Once these animal experiments are completed, there are still numerous hurdles
to overcome in developing fully validated animal brain models. For example, while
material properties available on animal brain tissues are more abundant compared to
human tissues, more studies are still needed to develop proper constitutive laws and
more accurate material properties at strain rates relevant to the impact condition.
Even in a systematic series of investigations, the data are expected to have a fairly
large standard deviation due to biological variations. Presently, FE brain models are
based on a deterministic approach by inserting a set of material properties into the
model which is then subjected to a set of loading conditions to calculate intracranial
responses. Because variations exist in material properties and loading conditions,
a probabilistic approach should be applied to determine the spectrum of response
variables. In other words, there will be a range of model-predicted intracranial
responses reflecting the dissimilarities among the population and the relative risk
of injury.
It can be expected that accurate FE human head models can only be developed
with success after completing the modelling of animal brains through which one can
identify material laws and associated constants, define with accuracy the loading
conditions, and ascertain regional tissue level injury thresholds through 3D injury
mapping. The human head models developed based on this principle can then be
used as the surrogate to design better countermeasures to prevent or mitigate brain
injury. As the human brain is better protected against blunt impact, the incident
rate of TBI should decrease. Nevertheless, complete elimination of TBI may not
be immediately achievable. Thus, newer and higher-quality real-world data need to
be gathered continuously to improve the simulation models so that their capability
in accurately predicting the risk of brain injury under a variety of blunt impact
conditions is augmented.
To improve the prediction power of a model, we will need to represent more
comprehensive anatomical details of the human head. For example, the complex
space between the skull and brain, which includes dura, arachnoid, cerebrospinal
fluid, pia, as well as various border cells between these meninges, is demonstrated
to have a nonlinear, viscoelastic property under tension and shear loading modes
[42, 43, 46]. However, limited by existing computational power and preferred by
users for improving model stability, several of the most advanced human head
models have been developed by adopting a contact algorithm to connect the skull to
the brain. Even though such simplification might not affect the prediction of deep
brain responses, it limits the models’ ability to predict injuries specifically in the
brain-skull interface, such as vessel-damage-induced subdural hematoma. Besides
125
animal experimental models (a small and a large one) need to be conducted to
develop a scaling method based on parameters such as age, gender, size, material
properties, etc. Preferably, a third animal model should be tested to validate the
accuracy of the scaling law and to develop a scaling method to estimate the tissue
level injury threshold in humans. Note that this method does not account for
species differences in tolerance and may not provide a direct correlation between
size and tolerance.
Once these animal experiments are completed, there are still numerous hurdles
to overcome in developing fully validated animal brain models. For example, while
material properties available on animal brain tissues are more abundant compared to
human tissues, more studies are still needed to develop proper constitutive laws and
more accurate material properties at strain rates relevant to the impact condition.
Even in a systematic series of investigations, the data are expected to have a fairly
large standard deviation due to biological variations. Presently, FE brain models are
based on a deterministic approach by inserting a set of material properties into the
model which is then subjected to a set of loading conditions to calculate intracranial
responses. Because variations exist in material properties and loading conditions,
a probabilistic approach should be applied to determine the spectrum of response
variables. In other words, there will be a range of model-predicted intracranial
responses reflecting the dissimilarities among the population and the relative risk
of injury.
It can be expected that accurate FE human head models can only be developed
with success after completing the modelling of animal brains through which one can
identify material laws and associated constants, define with accuracy the loading
conditions, and ascertain regional tissue level injury thresholds through 3D injury
mapping. The human head models developed based on this principle can then be
used as the surrogate to design better countermeasures to prevent or mitigate brain
injury. As the human brain is better protected against blunt impact, the incident
rate of TBI should decrease. Nevertheless, complete elimination of TBI may not
be immediately achievable. Thus, newer and higher-quality real-world data need to
be gathered continuously to improve the simulation models so that their capability
in accurately predicting the risk of brain injury under a variety of blunt impact
conditions is augmented.
To improve the prediction power of a model, we will need to represent more
comprehensive anatomical details of the human head. For example, the complex
space between the skull and brain, which includes dura, arachnoid, cerebrospinal
fluid, pia, as well as various border cells between these meninges, is demonstrated
to have a nonlinear, viscoelastic property under tension and shear loading modes
[42, 43, 46]. However, limited by existing computational power and preferred by
users for improving model stability, several of the most advanced human head
models have been developed by adopting a contact algorithm to connect the skull to
the brain. Even though such simplification might not affect the prediction of deep
brain responses, it limits the models’ ability to predict injuries specifically in the
brain-skull interface, such as vessel-damage-induced subdural hematoma. Besides