Influence of FE model variability in predicting brain motion and intracranial pressure changes in head impact simulations
Summary (4 min read)
Introduction
- Traumatic head impact injuries occur when the human skull and brain are rapidly subjected to intolerable levels of energy.
- Motor-vehicle crashes are the most common cause of brain injury.
- In particular, a FE model that describes in detail the complex geometry and the multiple material compositions of the brain can be used to calculate internal stress, strain and pressure at all locations and at any given instant during an impact [12] .
- Shugar [16] developed a 3D model of the skull in 1975 and described its method of construction and limitations.
- Both suffered somewhat with problems in creating a well-conditioned element mesh.
Construction of the FE variations 1. Baseline Model
- The baseline model includes the cerebrum, cerebellum and brainstem, intracranial membranes (falx and tentorium), pia, cerebrospinal fluid layer (CSF), dura, a varying thickness three-layered skull (cortical and trabecular bone layers), scalp and the facial bone.
- 7,318 hexahedral elements represent the brain and 2,874 hexahedral elements represent the CSF layer with one element through the thickness.
2. Sliding Boundary Model
- A model was built with a sliding boundary between the pia and the CSF layer.
- The algorithm allows no separation of the contacting pia and CSF layers and thus prevents the formation of a gap at the CSF-cerebrum interface.
- Considering the effect of fibrous trabaculae and the fluid nature of the CSF, material properties assumed for the solid elements which were used for the CSF were the bulk modulus of water and a very low shear modulus.
- The CSF which exists between the falx and the brain, and between the tentorium and cerebrum/cerebellum was also included using the same type of fluid-structure contact algorithm.
3. Grey-White-Ventricular Matter Model (GWV Model)
- The physiological accuracy of the baseline UCDBTM was refined by explicitly distinguishing between grey matter, white matter and the ventricles.
- Hexahedral meshing along the boundaries of such regions would require an unnecessary mesh development effort and the distinction between these types of neural tissue was conveniently made by modifying the baseline finite element mesh in such a way that elements were assigned material properties appropriate to these corresponding regions of the cerebrum.
- Each element then performed a global search to find the MRI voxels in its vicinity, and then performed a more detailed local search to discover exactly what elements were contained within its volume.
- Figure 2 shows how this algorithm operates schematically.
- Any hexahedron element, which is the only type of solid element used in the UCDBTM, can be decomposed into tetrahedrons.
4. 3-element CSF Model
- Instead, a model was developed to consist of three elements through the CSF layer, the outer nodes being tied to the skull and brain.
- In this way the 'no-slip' condition of fluid mechanics could be met.
- Because of the small size of the associated elements, it was decided to use a slightly deeper fluid element model which was developed previously [1] .
- Problems arose when attempts were made to define an Eulerian boundary condition at the foramen.
- This causes there to be a very large number of separate amplitude curves to be defined for the ABAQUS input deck.
5. Projection Mesh Model
- As the mesh density of the circle is increased, the internal angle of these elements approaches 180 degrees.
- These and adjacent elements will be ill-conditioned and this is particularly problematic for analyses where the outer regions are of greatest interest and where numerical accuracy is required to be highest.
- An alternative mesh generation scheme, referred to as the projection mesh method, does not involve this distortion of a square mesh.
- These projected lines, in turn, define the edges of rows of quadrilateral elements (hexahedral in 3D).
- Because of the definite change in mesh structure at the boundary of the square, the quality of the elements in the projection region will not degrade further, regardless of how refined the mesh density is made.
Tentorium (shell elements)
- Brain Skull (rigid in this paper) CSF layer with 3 elements through its thickness.
- This technique of meshing curved objects using multiple blocks is commonly used.
- Normal curves to be constructed from edges of original swept mesh, at a distance user picks -must be sufficient to give room for inner block, and maintain even mesh density as much as possible, also known as Rule 1.
- A further advantage of the projection mesh model is that its element density can be increased to any desired level without loss of element quality; a corresponding increase in element density of the baseline model would reduce the mesh quality at the outer sides of the brain.
- Longer duration impact events (falls, pedestrian and RTAs) do not require meshes of such high element density.
6. Morphed Model
- It was of interest to see how the pressure and displacements of the FE model would change if a different intracranial space (other than that of the visible male's) was modelled.
- Creation of another entire head FE model, even with knowledge of where the decomposition sections should be [1] , was deemed outside the scope of this work.
- This has the basis that a plate with isotropic properties (eg. Steel) can be morphed to fit a homologous shape (a shape that can be formed from the original without joining, crossing or tearing of the original shape) by means of applying forces and constraints to 'bend' the plate to the new desired shape.
- Vectors were constructed by picking points on the coarse grid and their matching landmark point on the new skull ), until every point on the coarse grid had a displacement vector associated with it.
- A static displacement analysis, using all of these displacement vectors as input, was then carried out on this coarse grid (giving it isotropic properties) and the results were read into MSC/Patran [40] .
Constitutive Property Assignment
- The baseline model was previously validated against the pressure information of the cadaver test of Nahum and has been discussed previously [1] .
- It was noted before that changes in intracranial properties affected different simulation results using this model by different.
- The material properties used in this present paper are presented in Table 2 .
1. Predictions of Intracranial Pressure During Impact
- The baseline and five variant FE models were used to simulate Trosseille's cadaver impact test [2] .
- The mechanics of this experiment were more complicated than the cadaver test of Nahum [33] , which had been used previously to validate the baseline model [1] .
- This impact situation was conveniently simulated with the present finite element models by applying the six time varying components of the linear and angular velocity measured experimentally to the FE models (up to 35 ms of the recorded data).
- EQUATION Figures 9 and 10 compare the experimental and simulated frontal and occipital pressures while Figure 11 compares the pressure at the lateral ventricle (baseline and GWV models only).
- The sliding model exhibits a similar frontal pressure, but this is most likely due to the contact formulation in this area.
2. Predictions of Brain Displacement During Impact
- Experimental data from recent cadaver head impact experiments of Hardy and Kleiven provided vectorised information on the motion of different regions of the brain during impact [3, 46] .
- In Hardy's experiment, two sets of neutral density targets with approximately the same density as brain were inserted in two vertical columns into the brain.
- By assuming that the CG of the FE model was coincident with that of the cadaver head, the kinematics of the FE head would be exactly the same as those of the cadaver head for any given impact [12] .
- The displacements were measured in a coordinate system fixed to the head as shown in Figure 12 , which shows the comparison between the experimental relative displacements (projected onto the X-Z plane) and those predicted numerically by the UCDBTM.
- For ease of viewing, results in Figure 12 show only the results of the GWV model.
Conclusions
- The baseline UCDBTM has previously been validated against the pressure data of Nahum.
- These six variant models were compared against two cadaveric tests: one to measure the variation of intracranial pressure during impact and the second to measure the motion of the brain relative to the skull during impact.
- The most notable difference was observed when comparing the sliding boundary model's pressure response, though this is more likely to be due to the penalty contact formulation used.
- Again little evident difference was noted when the models were compared against the relative motion data.
Did you find this useful? Give us your feedback
Citations
308 citations
219 citations
207 citations
Cites background from "Influence of FE model variability i..."
...The relationship of stress waves to TBI have been well studied using simulations under impact loading such as in sports or motor vehicular injuries (Belingardi et al., 2005; Gilchrist et al., 2001; Horgan and Gilchrist, 2004; Raul et al., 2006; Taylor and Ford, 2006; Zhang et al., 2001; Zhang et al., 2004)....
[...]
...…of stress waves to TBI have been well studied using simulations under impact loading such as in sports or motor vehicular injuries (Belingardi et al., 2005; Gilchrist et al., 2001; Horgan and Gilchrist, 2004; Raul et al., 2006; Taylor and Ford, 2006; Zhang et al., 2001; Zhang et al., 2004)....
[...]
182 citations
173 citations
References
833 citations
Additional excerpts
...[7, 8])....
[...]
556 citations
457 citations
419 citations
366 citations
Related Papers (5)
Frequently Asked Questions (13)
Q2. What have the authors contributed in "Influence of fe model variability in predicting brain motion and intracranial pressure changes in head impact simulations" ?
In total six different model configurations were constructed: ( i ) the baseline model, ( ii ) a refined baseline model which explicitly differentiates between grey and white neural tissue, ( iii ) a model with three elements through the thickness of the cerebrospinal fluid ( CSF ) layer, ( iv ) a model simulating a sliding boundary, ( v ) a projection mesh model ( which also distinguishes between neural tissue ) and ( vi ) a morphed model.
Q3. What is the implications for future modelling activity?
The implications for future modelling activity is that the very large mesh densities that are required to properly simulate the propagation of compressive waves of ballistic impacts using explicit finite element solvers (elements of order 50µm in size [39]) will most likely be developed via projection mesh methods.
Q4. Why do the elements in the projection mesh degrade?
Because of the definite change in mesh structure at the boundary of the square, the quality of the elements in the projection region will not degrade further, regardless of how refined the mesh density is made.
Q5. What was the assumption of a rigid skull?
The assumption of a rigid skull was necessary in order to satisfy the rigid body criteria used to obtain the six acceleration components.
Q6. What is the role of computer models in the investigation of neurotrauma?
With the rapid advances in computer technology, sophisticated computer models of the head can provide useful information in the investigation of neurotrauma due to impact.
Q7. What were the pressures measured in the subarachnoid space and in the ventricular?
Miniature pressure transducers were placed in the subarachnoid space and in the ventricular system to measure intracranial and ventricular pressures.
Q8. What was the emphasis on when constructing the model?
When constructing the model, emphasis was placed on element quality and ease of mesh generation, as this has proved to be a difficulty in other models.
Q9. What is the advantage of cadaver experiments?
Human cadaver experiments have the advantage of accurate anatomy, but the limitation of not properly representing the physiology of the living human (although attempts have been made to simulate vascular lesions by pressurising the vascular system prior to impacting cadaver heads).
Q10. What is the effect of the deformation of a regular square grid?
If a regular square grid is deformed in such a way as to form a circle (Figure 5), it would produce 4 elements, corresponding to the four corners of the block, which are highly distorted.
Q11. What was the process of calculating the CSF layer?
In this formulation, the CSF layer was continually smoothed during the impact simulations, moving the nodes to maintain element quality while mapping the solution from the old mesh positions to the new optimised ones.
Q12. What was the UCDBTM used to determine the relative displacement of the cadaver head?
These position vectors were then converted into a moving coordinate system so as to determine the relative displacement of the targets with respect to the skull.
Q13. Where did the mesh be used to update and improve the quality of the elements?
In the regions of the brain stem and where the midbrain transitions to the temporal lobe, further methods (not reported here) were used to update and improve the element quality while retaining the ability for the user to choose the refinement level.