Hexahedral mesh of anatomical atlas for construction of computational human brain models: Applications to modeling biomechanics and bioelectric field propagation

TL;DR

The work addresses the challenge of building accurate computational brain models by converting the Open Anatomy Project SPL/NAC atlas into a conforming, labeled hexahedral FE mesh that supports multi-disciplinary simulations. The authors present a scalable overlay-grid meshing workflow, add missing head structures, and label elements to enable region-specific material properties and boundary definitions, demonstrated through biomechanics and EEG forward problems. Key contributions include high-quality mesh generation with smooth interfaces, atlas-based labeling for context-rich analysis, and seamless integration with open-source tools, culminating in an openly available mesh resource. This atlas-driven approach enables efficient, anatomically informed brain simulations, reducing manual segmentation and enhancing visualization and communication across neuroscience, biomechanics, and bioelectric fields.

Abstract

Numerical simulations rely on constructing accurate and detailed models to produce reliable results - a task that is often challenging. This task becomes notably more difficult when the model is of the human brain. We create an anatomically comprehensive hexahedral mesh of the human brain using an open-source digital brain atlas. Digital atlases are valuable tools currently used by medical professionals, medical students, and researchers for gathering, presenting, and discovering knowledge about the human brain. We demonstrate that the atlas can be used to efficiently create an accurate and detailed hexahedral finite element mesh of the brain for scientific computing. We present two case studies. The first case study constructs a biomechanical model of the brain to compute brain deformations and predict traumatic brain injury risk due to violent impact. In the second case study, we construct a bioelectrical model of the brain to solve the electroencephalography (EEG) forward problem, a frequent simulation process used in electrophysiology to study electromagnetic fields generated by the nervous system. We demonstrate efficient and accurate model construction using the meshed anatomical brain atlas, as well as emphasize the importance of effective communication and contextual analysis of results for enabling multi-disciplinary scientific computing research.

Figures (13)

• Figure 1: Open Anatomy Project's SPL/NAC Brain Atlas. Top is the 3D rendering of the label map and the bottom is the label map overlaid onto the MRI scan
• Figure 2: The Material label map consisting of the scalp, skull, CSF, grey matter and white matter derived from the SPL/NAC brain atlas using the SlicerAtlasEditor extension. The label map is color coded with pink as the scalp, yellow as the skull, blue as the CSF and ventricles, grey as the grey matter and white as the white matter.
• Figure 3: Close-up sagittal plane cut of the labelled meshed anatomical brain atlas showing smoothing of boundaries by the Sculpt meshing procedure. (A) Close-up of the mesh without smooth boundaries and (B) close-up of the mesh with smooth boundaries at the outer scalp, scalp-skull and skull-CSF interface.
• Figure 4: Distribution of mesh quality metrics for 1:1 and 8:1 voxel-to-element ratio meshes. The top row presents mesh quality for the 1:1 ratio mesh, while the bottom row shows results for the 8:1 ratio mesh. For each mesh, the horizontal axis represents the mesh quality metric, and the vertical axis displays the percentage of elements within each quality range.
• Figure 5: The meshed anatomical brain atlas. Top row: (A) 3D cut-out view of the mesh with OAP's SPL/NAC brain atlas labels and (B) 3D view of the mesh without the skull and scalp. Bottom row: (C) 3D cut-out view of the mesh material labels for constructing finite element (FE) models and (D) 3D view of mesh without the skull and scalp.