Abstract:
This paper presents an automated framework for generating subject-specific, all-hexahedral finite element (FE) meshes of the brain from medical image volumes, tailored for biomechanical applications such as traumatic brain injury (TBI) simulation. An initial atlas mesh is created via a scripted multi-block hexahedral meshing approach; subsequent subjects’ MR images are rigidly registered to this atlas using mutual information, their segmented brain surfaces transformed into atlas space, and the same meshing script applied—after which the mesh is inversely mapped back to subject space. Tested on 11 athlete subjects, the method yields meshes with >99.5 % of elements having scaled Jacobian >0.6, average surface‐to‐mesh boundary errors of 0.07 mm (<1 % >0.5 mm), and no inverted elements, all generated in under 4 minutes per case without any mesh repair. By preserving anatomical fidelity and mesh quality through image-based registration rather than direct mesh morphing, this technique ensures biofidelic FE models that support accurate load transmission and deformation predictions in computational biomechanics studies.
