122
K. H. Yang and H. Mao
apparently outliers. Obviously, there were inherent problems associated with these
experiments.
Gurdjian et al. (1953) once wrote: ‘Due to the error introduced resulting from
deformation of the scalp, muscles, skull, and striker itself, it appeared more accurate
to measure the acceleration of the skull.’ As a result, the most commonly used head
injury criteria are all based on acceleration without considering the intracranial
responses. Explanations for the unreasonable measurements reported by Nahum
et al. [71] may include data acquisition system errors, variations in the cadaveric
anthropometry, and the unspecified locations of pressure transducer placements.
Similar problems can also be found in newer laboratory-generated experimental data
reported by other authors. For example, Hardy et al. [30] reported large standard
deviations observed in the motions of the brain with respect to the skull using the
same cadaveric head under very similar loading conditions. Lastly, using real-world
accident reconstruction data for model validation presents even more problems
because the only known variable is the injury outcome.
To correct this deficiency, experiments specifically designed to generate data for
model validation are needed. One such dataset has been generated to validate human
neck models subjected to rear-end impact [100]. In this series of experiments, highresolution CT scans were conducted on each cadaver to document the cervical spine
geometry and to generate subject-specific neck models, while cervical spine kinematics were obtained by a high-speed biplanar x-ray system for model validation.
Obviously, experiments of this kind will be very expensive and require pooling of
financial resources from all stakeholders to achieve the goal. In the recent Warrior
Injury Assessment Manikin (WIAMan) project [4], every cadaver tested were CT
scanned to provide geometric information needed for potential modelling efforts. If
FE models are to be used as a tool for mitigating the societal problems associated
with TBI, we sincerely hope that government and industrial leaders will step up
their efforts by organising systematic studies to identify material properties and to
conduct experiments specifically designed for model validations, just like that has
been done for the WIAMan project.
5.4 Revamp FE Modelling of Human Head: A Look
into the Future
We hypothesise that successful development of human head models for preventing
blunt impact-induced TBI depends to a great extent on animal experiments. There
are several reasons for this. First, freshly dead human heads (and brains) are very
hard to obtain, and studying the TBI mechanism and threshold requires a large
number of cadavers because of the inherent variability of cadaveric data. Second,
for those cadavers available for impact testing, it is very difficult to prepare them to
imitate in vivo conditions. Third, cadaver experiments will not reveal critical brain
injury besides contusion or tearing. Animal studies, on the other hand, enable the
K. H. Yang and H. Mao
apparently outliers. Obviously, there were inherent problems associated with these
experiments.
Gurdjian et al. (1953) once wrote: ‘Due to the error introduced resulting from
deformation of the scalp, muscles, skull, and striker itself, it appeared more accurate
to measure the acceleration of the skull.’ As a result, the most commonly used head
injury criteria are all based on acceleration without considering the intracranial
responses. Explanations for the unreasonable measurements reported by Nahum
et al. [71] may include data acquisition system errors, variations in the cadaveric
anthropometry, and the unspecified locations of pressure transducer placements.
Similar problems can also be found in newer laboratory-generated experimental data
reported by other authors. For example, Hardy et al. [30] reported large standard
deviations observed in the motions of the brain with respect to the skull using the
same cadaveric head under very similar loading conditions. Lastly, using real-world
accident reconstruction data for model validation presents even more problems
because the only known variable is the injury outcome.
To correct this deficiency, experiments specifically designed to generate data for
model validation are needed. One such dataset has been generated to validate human
neck models subjected to rear-end impact [100]. In this series of experiments, highresolution CT scans were conducted on each cadaver to document the cervical spine
geometry and to generate subject-specific neck models, while cervical spine kinematics were obtained by a high-speed biplanar x-ray system for model validation.
Obviously, experiments of this kind will be very expensive and require pooling of
financial resources from all stakeholders to achieve the goal. In the recent Warrior
Injury Assessment Manikin (WIAMan) project [4], every cadaver tested were CT
scanned to provide geometric information needed for potential modelling efforts. If
FE models are to be used as a tool for mitigating the societal problems associated
with TBI, we sincerely hope that government and industrial leaders will step up
their efforts by organising systematic studies to identify material properties and to
conduct experiments specifically designed for model validations, just like that has
been done for the WIAMan project.
5.4 Revamp FE Modelling of Human Head: A Look
into the Future
We hypothesise that successful development of human head models for preventing
blunt impact-induced TBI depends to a great extent on animal experiments. There
are several reasons for this. First, freshly dead human heads (and brains) are very
hard to obtain, and studying the TBI mechanism and threshold requires a large
number of cadavers because of the inherent variability of cadaveric data. Second,
for those cadavers available for impact testing, it is very difficult to prepare them to
imitate in vivo conditions. Third, cadaver experiments will not reveal critical brain
injury besides contusion or tearing. Animal studies, on the other hand, enable the