Traumatic brain injury (TBI) is a great burden for the society worldwide and the statisticsindicates a relative constant total annual rate of TBI. It seems that the present preventativestrategies are not sufficient. To be able to develop head safety measures against accidents ine.g. sports or automobile environment, one needs to understand the mechanism behindtraumatic brain injuries. Through the years, different test subjects have been used, such ascadavers, animals and crash dummies, but there are ethical issues in animal and human testingusing accelerations at injury-level and crash dummies are not completely human-like. In aFinite Element (FE) head model, the complex shape of the intracranial components can bemodeled and mechanical entities, such as pressure, stresses and strains, can be quantified atany theoretical point. It is suggested that the size of the head, the skull-brain boundarycondition, the heterogeneity, and the tethering and suspension system can alter the mechanicalresponse of the brain. It can be seen that the shape of the skull, the composition of gray andwhite matter, the distribution of sulci, the volume of cerebrospinal fluid and geometry of othersoft tissues varies greatly between individuals. All this, suggests the development of patientspecific FE head models.A method to generate patient specific FE head model was contrived based on the geometryfrom Magnetic Resonance Imaging (MRI) scans. The geometry was extracted usingexpectation maximization classification and the mesh of the FE head model was constructedby directly converting the pixel into hexahedral elements. The generated FE model had goodelement quality, the geometrical details were more than 90 % accurate and it correlated wellwith experimental data of relative brain-skull motion. The method was thought to beautomatic but some hypothetically important anatomical structures were not possible to beextracted from medical images. This leads to investigations on the biomechanical influence ofthe cerebral vasculature, the falx and tentorium complex. It was found that biomechanicalinfluence of the cerebral vasculature was minimal, due to the convoluting geometry and thenon-linear elastic material properties of the blood vessels. It suggests that futurebiomechanical FE head model does not necessarily have to include these blood vessels. Theinclusion of falx and tentorium in an FE head model has on the other hand a substantialbiomechanical influence by affecting its surrounding tissue. Therefore, in the investigation ofthe biomechanical influence of the sulci, the falx and tentorium were manually added to theanatomically detailed 3D FE head model. The biomechanical influence of the sulci haspreviously not been studied in 3D and the results indicated an obvious reduction of the strainin the model with sulci compared to the model without sulci in all simulations, and mostinteresting was the consistent reduction of strain in the corpus callosum. The development ofgyri not only produces a larger area for synapses but also forms the sulci to protect the brainfrom external forces.Based on the results, a patient specific FE head model for traumatic brain injury predictionshould at least include the skull, cerebrospinal fluid, falx, tentorium and pia mater, in additionto the brain. With these anatomically detailed 3D models, the injury biomechanics can bebetter understood. Hopefully, the burden of TBI to the society can be alleviated with betterprotective systems and improved understanding of the patients’ condition and hence, theirmedical treatments / QC 20100811

Ce travail de thèse vise à mieux décrire les mécanismes de lésions de la tête humaine en situation de choc en optimisant le modèle par éléments finis de la tête humaine de Strasbourg (SUFEHM) en termes de modélisation mécanique du crâne et du cerveau grâce à de nouvelles données expérimentales et de techniques récentes d’imagerie médicales. Une première étape a consisté à améliorer la loi de comportement de la boîte crânienne, valider son comportement en regards d’éléments expérimentaux sur cadavres et proposer un MEF capable de reproduire fidèlement la fracture crânienne. La deuxième partie consiste en la prise en compte pour la première fois de l’anisotropie dans les simulations par EF d’accidents réels en utilisant l’Imagerie du Tenseur de Diffusion. Après implémentation, une phase de validation a été entreprise afin de démontrer l’apport de l’anisotropie de la matière cérébrale dans un MEF. Enfin 125 accidents réels ont été reproduits avec le SUFEHM ainsi amélioré. Une étude statistique sur les paramètres mécaniques calculés a permis de proposer des limites de tolérances en termes de fracture crânienne et de lésions neurologiques en s’intéressant tout particulièrement à l’élongation axonale maximale admissible, nouvelle métrique proposée. / The principal objective of this study is to enhance the existing finite element head model. A composite material model for skull, taking into account damage is implemented in the Strasbourg University Finite Element Head Model in order to enhance the existing skull mechanical constitutive law. The skull behavior is validated in terms of fracture patterns and contact forces by reconstructing 15 experimental cases in collaboration with Medical College of Wisconsin. The new skull model is capable of reproducing skull fracture precisely. The composite skull model is validated not only for maximum forces, but also for lateral impact against actual force time curves from PMHS for the first time. This study also proposes the implementation of fractional anisotropy and axonal fiber orientation from Diffusion Tensor Imaging of 12 healthy patients into an existing human FE head model to develop a more realistic brain model with advanced constitutive laws. Further, the brain behavior was validated in terms of brain strain against experimental data. A reasonable agreement was observed between the simulation and experimental data. Results showed the feasibility of integrating axonal direction information into FE analysis and established the context of computation of axonal elongation in case of head trauma. A total 125 reconstructions were done by using the new advanced FEHM and the axonal strain was found to be the pertinent parameter to predict DAI.