Human cranium biomechanical simulation

Transcription

1 Human cranium biomechanical simulation P. Perestrelo & P. Bártolo Centre for Rapid and Sustainable Product Development, Leiria, Portugal M. P. Torres, P. Noritomi & J. Silva Division of Three-Dimensional Technologies, Centre for Information Technology Renato Archer, Campinas, Brazil ABSTRACT: The traumatic brain injury (TBI) is a disturbing cause of death that has been growing year after year. In an effort to prevent, detect and understand it, a more audacious approach must be given to this problem. With this purpose, biomechanical and clinical theories must be combined with the objective of improving the technics available. The proposed development of a virtual platform includes the acknowledgement of the human head s anatomy and injuries classification. This platform or virtual model, is developed with the BioCAD protocol, computer aided design (CAD) software and finite element method (FEM) analysis software. The final version of this virtual platform has a demand of being adaptive to the user and/or patient occasioning in an innovative tool most needed to improve the research and prevention of TBI. Hopefully, this will result in a higher level of understanding of this problem. 1 INTRODUCTION Per year there are an estimated number of 10 million people affected by TBI, as referenced by the World Health Organization (WHO). It is also estimated that by 2020 this will be the major cause of death and disability (Gean & Fischbein 2010). Though there have been developments in this area, there is still a long path to be taken to reach an efficient protocol or practice in order to efficiently prevent, detect and treat patients of TBI. Instead of pursuing different and/or separated paths, researchers, engineers and physicians must join forces to develop innovative techniques and further understand the effect of TBI. Linear and angular accelerations have been studied to understand the effect that each one cause on the brain mass. From this, a variety of characteristic injuries were identified, such as, diffuse axonal injury (DAI) and strain rate. Along with this process, analysis and discussions have been conducted to refine this development (King et al. 2003). Still, a change of course is necessary to improve TBI research and prevention. With our proposal of reaching a virtual platform for TBI research, it is intended that situations that can result in head trauma can be recreated and simulated. Thus, in the materials and methods section, a description of the demands, the conceptual theory and the procedures used as a base to the computational model are shown. In the discussion section, flaws from another models and problems encountered in the platform development are explained. As for the simulation section, a preliminary simulation and its results are presented. At the end, in the conclusion section, a discussion of the simulation results along with conclusions of the work made so far, are expounded. 2 MATERIALS AND METHODS With the intention of understanding the basic principles behind the head trauma, a study of the head anatomy must be done. Considering the non-facial area as the area of interest to study for this project, an approach from the outside to the inside is to be understood. Therefore, the first barrier to appear is the scalp with a thickness of 5 to 7 mm, following the skull bone and the meninges. There are three meninges, the dura mater, arachnoid and pia mater. Creating a separation between them it is a cerebrospinal fluid in the subdural and subarachnoid spaces which is important to our study as well because of its shock absorbing capabilities (Brands 2002). Under the elements mentioned above, lays the brain tissue which is separated in several parts. The most important are the cerebrum, the cerebellum and the brain stem. This last part comprises the midbrain, pons and medulla oblongata, connecting to the spinal cord through the foramen magnum. The cerebrum and cerebellum are divided into left and right hemispheres with the accomplishment of this partition being achieved with the falx cerebri and falx cerebelli, invaginations to the dura mater. Another invagina-

2 tion, the tentorium, marks the separation of the cerebrum and cerebellum with the ventricles filled with cerebrospinal fluid (Fig. 1a, 1b) (Brands 2002). Figure 1a. Back view of important components of the human head (Brands 2002). Figure 1b. Side view of important components of the human head (Brands 2002). In the brain matter the cells network and blood vessels distribution is bordered in two types, the gray and white matter. The gray matter is composed of nerve cells (neurons) and the white matter of myelinated nerve cells (axons). The fact that these cells are different can impose an interest to our study (Brands 2002). The classification of the head injuries is described as being opened and closed. Open injuries are associated with skull fracture and can also be associated with soft tissues damage, while in the presence of a skull fracture. Closed injuries are not associated with skull fracture but with damage to the intracranial contents.from these two types of injuries the one with most interest to our project is the closed injury. The closed injury divides itself into focal and diffused. Simplifying, focal injuries can be pointed as restricted to a specific area of the brain and diffused injuries as being dispersed through a region of the brain. Some possible consequences that can be generated by this kind of injuries are subdural hematoma, epidural hematoma, intra-cerebral hematoma, shifts of cerebral tissue, herniation and brain stem compressions. Changes in intracranial pressure can create shear strains and intracranial compression can cause the propagation of a pressure wave, resulting in severe damage (Brands 2002). Looking to the neurophysiology of the concussion as it is a short-lasting functional disturbance of neural function actuated by an exterior aggression or acceleration/deceleration of the head without skull fracture, it can generate all the consequences mentioned in the last paragraph. Consequently, it reveals itself as most challenging to study and leads to an urge to fully understand and study this injury (Shaw 2002). The study of TBI has been done by taking, generally, two approaches. The first one is the physics approach taken by bioengineers, having a special attention to the input variables that cause TBI. Commonly, those are the linear acceleration, angular acceleration, strain and strain rate in the brain (King et al. 2003). The second approach is almost exclusively a clinical one, executed by medical specialists. It consists in a study of the consequences of the trauma, such as, diffuse axonal injury (DAI) and /or brain edema (BE) (Marmaru 2003). A third approach is proposed with our work, differing from the previous two by joining them, therefore, complementing each other. To accomplish this task, an interdisciplinary team was assembled with engineers and a neurosurgeon, so that the whole analysis remains homogeneous. From the beginning, the project has been directed to the creation of a computational model that can serve as an open platform, using a specific patient modeling method (SPM) (Castellano-Smith et al. 2001). Nonetheless, the final intent is to transform it into a general platform with results applicable for various purposes. During the construction of this model, the SPM followed a protocol called BioCAD that was the guideline to the model development. It is a sequence of tasks developed in various softwares, starting with InVesalius 3.0, which is a software developed in the Division of Three-Dimensional Technologies, in the Centre for Information Technology Renato Archer (DT3D/CTI). It follows a computer-aided design (CAD) tool and a finite element analysis (FEA) software. The initial phase of the BioCAD protocol was loading the patient computed tomography (CT) scans into InVesalius 3.0 (Fig. 2) cleaning them for imperfections and soft tissues, leaving only bony structures. When those steps were concluded, a stereolithography (STL) mesh was generated.

3 After the modeling was concluded and a mesh was created, a FEA mesh optimization was initiated followed by results discussion (Fig. 4) (Noritomi et al. 2012). Figure 4. Caption of model mesh in Ansys Figure 2. CT scan file loaded into InVesalius 3.0. Then the STL mesh was exported to CAD software Rhinoceros 4.0 due to its ability to combine surface modeling with complex geometries (Fig. 3). 3 DISCUSSION The concept of BioCAD diverges from the models found in the literature, which lack detail thus, moving to incomplete results or difficulties to simulate. For example, this could be because of a very rough mesh detail (Fig. 5). Figure 5. A cranium model from the literature with rough meshing (Kleiven 2006). Figure 3. STL mesh loaded into Rhinoceros 4.0. Subsequently, having this mesh as a guide to the modeling, one can construct a structure with a variable level of detail maintaining anatomical accuracy. This means that the level of detail can change depending on the area that is being drawn, concerning the purpose of the project and an anatomical study previously done. In the case of our model this opens the possibility of adjusting the mesh refinement to the level the user wants in a specific region of the head. As another example, the case of a voxel based model confirms an excess of detail that can lead to difficulties on the solution process, along with a very high need for computing power. The lack of mesh control, in order to have more detailed areas, is also a problem because when an adjustment is required it has to be done to the whole model, for example (Fig. 6).

4 Figure 8. - InVesalius 3.0 STL mesh with a flaw on the etmoid boné. Figure 6. Voxel model with an excess of detail (Watanabe et al. 2008). A poor representation of the human head anatomy, similar to the crash-test dummies leads to a different energy propagation than with a human head upon an external force application (Fig. 7). Another difficulty is realizing what protuberances or anchorage points need to be modeled to make sure that all important muscles to the head movement are correctly represented for the simulation. This challenging aspect is going to be further discussed with specialists in anatomy and biomechanics, due to their expertise in this area. Beside anatomical problems, the natural struggle to be able to combine the modeling of complex geometries with the level of detail necessary to the platform, is a constant challenge that follows the project development. 4 SIMULATION The actual development state of the computational model represents the human head exterior cortical bone (Fig. 9). Figure 7. Side-by-side comparison between a head model resembling a crash test dummy and an actual crash test dummy (Connolly 2007, Ghajari et al. 2009). As for the difficulties in the development of the platform, there has been a struggle in overcoming some anatomical problems, such as, some areas of the cranium on the CT scan that were not perfectly represented resulting in missing regions in the STL mesh. The case of the ethmoid bone is an example of it (Fig. 8). Figure 9. Exploded view of the current computational model. As a result, in order to learn if the model behaves as intended when in a simulation, an archive in the ACIS (.sat) format was exported from Rhinocerus 4.0 to Ansys Since Ansys 14.0 did not have in its library a material with the properties of cortical bone, it was defined a cortical bone material with a Young s Modulus of MPa and a Poisson Co-

5 efficient of These are the average values found in biomechanical publications for the cortical bone (Akca & Cehreli 2006). It was also assumed that the whole geometry material was a cortical bone with a thickness of 2 mm (Ghajari et al. 2009). Once the material was created the generation of a model mesh was required to proceed to the simulation phase (Fig. 10). This mesh was generated in the automatic mode of the software because of the early state phase of simulations. Nonetheless, with the model complete the mesh will be refined. Figure 12. Colour map of the maximum stress analysis. Figure 10. Computational model mesh generated in Ansys 14.0 software. In order to obtain a simulation that would create a possible situation of a head TBI with an easy interpretation, a fixed support and a force where defined. The intended situation was an excessive load to the malar bone near the eye socket. As can be seen in Figure 11, the fixed support was located in the foramen magnum (skull base) and the force has been applied to a small surface near the eye socket. Figure 11. Caption showing the force and the fixed support applied to the model. With all the considerations in place the situation was simulated. The results were displayed in the format of a colour map concerning the maximum stress analysis in the model, so that a behavior acknowledgement of the geometry would be made (Fig. 12). The numerical results are presented in Figure 13 to display a better visualization. Figure 13. Maximum stress analysis simulation results. 5 CONCLUSIONS After the conclusion of this part of the project, one can firmly state that a model with a poor representation of the human head, with an excess of detail or with a rough mesh, outcomes in results that lack for accuracy. Contrasting with these statements, our proposal is a computational model for a virtual platform developed through a real human cranium. Though it needs to be further developed, its initial creation based in an advantageous anatomical design with a consciousness of the possible TBI injuries that may be caused, adds higher probability of obtaining close to real results. The behavior of the model during the simulation was according to expectations when using constraints and simplification hypotheses. Nevertheless, to have a comparison basis between results, tests in a different software and an experimental validation must be made. However, it is important to understand that these tests do not have a clinical purpose but the objective of confirming that a computational model for a more evolved platform has been achieved. The possibility of obtaining breakthrough results improves with the project development and with the application of BioCAD protocol. The adjustments of the mesh in areas of the head, with significant importance in order to accomplish good results leaving

6 other less essential areas with a rougher mesh, is an example of the value that the BioCAD protocol brings to this kind of modeling. To finish this platform, a complete modeling of the head interior and the brain has to be made. Following, simulations of the whole geometry and structure have to be achieved, preceding results validation. When completed, this platform will prove to be a relevant advance in the area of TBI research and prevention. It will be freely available for anyone that needs such a model for any kind of simulation. 6 REFERENCES Akca K. & Cehreli, M Biomechanical consequences of progressive marginal bone loss around oral implants: a finite element stress analysis. International Federation for Medical and Biological Engineering. Springer. Brands, D Predicting brain mechanics during closed head impact: numerical and constitutive aspects. Eindhoven: University Press Facilities. Castellano-Smith, A., Hartkens, A., Schnabel, T., Hose, J., Liu, D., Hall, H., Truwit, W., Hawkes, C. & Hill, D Constructing patient specific models for correcting intraoperative brain deformation. MICCAI2001 Springer-Verlag. Connolly, C Instrumentation used in vehicle safety testing at Milbrook Proving Ground Ltd. In Sensor Review: 91-98, Vol. 27. Gean, A.D. & Fischbein, N.J Head Trauma. PubMed. Elsevier. Ghajari, M., Deck, C., Galvanetto, U., Ianucci,. & Willinger, R Development of numerical models for the investigation of motorcycle accidents. 7 th European LS-DYNA Conference DYNAmore Gmbh. Hyder, A., Wunderlich, C., Puvanachandra, P., Gururaj, G. & Kobusingye, O Is head injury caused by linear or angular acceleration?. NeuroRehabilitation Journal. IOS Press. Kleiven, S Evaluation of head injury criteria using a finite element model validated against experiments on localized brain motion, intracerebral acceleration and intracranial pressure. IJCrash Woodhead Publishing Ltd. Marmaru, A Pathophysiology of traumatic brain edema: current concepts. Acta Neurchir Springer-Verlag. Noritomi, P., Xavier, T. & Silva, J A comparison between BioCAD and some known methods for finite elemento model generation. In P. Bártolo (ed), Innovative developments in virtual and physical prototyping: Leiden: CRC Press/Balkema. Shaw, N.A Neurophysiology of concussion: Theoretical perspectives. In S. M. Slobounov & W.J. Sebastianelli (eds), Foundations of sport-related brain injuries: Springer. Watanabe, D., Yuge, K., Nishimoto, T., Murakami, S. & Takao, H Head impact analysis related to the mechanism of diffuse axonal injury. High-Tech Research Center Project for Private Universities MEXT Japan.