An atlas-based geometry pipeline for cardiac Hermite model construction and diffusion tensor reorientation

Area of Science:

• Biomedical Engineering
• Computational Anatomy
• Medical Imaging

Background:

• Accurate patient-specific cardiac models are crucial for understanding heart function and disease.
• Existing methods for generating finite element meshes of the heart are often manual and time-consuming.
• Integrating complex anatomical features and material properties, such as fiber orientations, remains a challenge.

Purpose of the Study:

• To develop and validate a novel, automated pipeline for constructing 3D cubic Hermite finite element meshes of the human heart from tomographic patient image data.
• To enable the incorporation of patient-specific anatomy and diffusion tensor MRI-derived fiber orientations into computational cardiac models.
• To facilitate accurate and efficient patient-specific cardiac simulations.

Main Methods:

• A cardiac atlas was built by segmenting major heart structures and extracting/smoothing boundary surfaces.
• A skeleton-based sweeping method was employed to construct a hexahedral control mesh from surface and path tree data.
• Cubic Hermite finite elements were defined using derivative parameters from the control mesh.
• An optical flow approach was developed for automatic atlas deformation and registration to new patient data.
• Diffusion tensor MRI data was deformably mapped to the patient geometries to include fiber and sheet orientations.

Main Results:

• A novel atlas-based geometry pipeline successfully generated 3D cubic Hermite finite element meshes of the whole human heart.
• The pipeline automatically segmented cardiac structures, identified critical points and skeletons, and constructed a hexahedral control mesh.
• An optical flow-based registration method achieved fully-automatic deformation and alignment of the cardiac atlas with new patient images.
• The method successfully incorporated diffusion tensor MRI-derived fiber and sheet orientations into the finite element models.
• The pipeline provides accurate measurements of atrial and ventricular wall thickness.

Conclusions:

• The presented pipeline offers an automated and robust solution for generating patient-specific 3D finite element models of the human heart.
• This approach significantly reduces manual effort in mesh generation and registration, making it more accessible for clinical applications.
• The ability to incorporate diffusion tensor MRI data enhances the physiological realism of the finite element models, paving the way for more accurate cardiac simulations.