Automated subject-specific, hexahedral mesh generation via image registration

Area of Science:

• Biomechanics
• Computational anatomy
• Finite Element Analysis (FEA)

Background:

• Generating subject-specific, all-hexahedral meshes is crucial for biomechanical research.
• Existing automated methods often degrade mesh quality due to distortion during atlas-to-subject morphing.

Purpose of the Study:

• To present an automated meshing technique that produces high-quality, accurate, all-hexahedral meshes without requiring mesh repair.
• To evaluate the performance and efficiency of this novel meshing approach.

Main Methods:

• Developed an atlas mesh using a script.
• Generated subject-specific meshes by transforming subject geometry to atlas space via rigid image registration, then back-transforming the mesh.
• Applied the technique to mesh the brain in 11 subjects.

Main Results:

• Achieved satisfactory mesh quality: 99.5% of elements had a scaled Jacobian >0.6.
• Demonstrated high accuracy: average boundary distance of 0.07 mm.
• Reported low computational cost: combined registration and meshing under 4 minutes.

Conclusions:

• The automated technique effectively generates subject-specific, all-hexahedral meshes with excellent quality and accuracy.
• This method shows promise for meshing diverse anatomical structures in various biomechanical research applications.