Finding complete 3D vertex correspondence for statistical shape modeling.

Area of Science:

• Computational anatomy
• Medical imaging
• 3D shape analysis

Background:

• Accurate statistical shape models require dense correspondences across 3D shapes.
• Identifying sufficient landmarks on 3D anatomical meshes, like aortic roots, is challenging due to smooth surfaces.
• Existing methods struggle with establishing complete vertex correspondences from sparse landmarks.

Purpose of the Study:

• To develop a method for establishing dense vertex correspondences in 3D aortic root meshes using sparse landmarks.
• To enable accurate generalization of shape variations within a family of 3D aortic root models.

Main Methods:

• Non-rigidly transforming a source mesh to a target mesh to establish vertex correspondences.
• Utilizing a sparse set of initially identified landmarks to guide the dense correspondence mapping.
• Deforming the source mesh to match the target, ensuring every target vertex has a corresponding point on the deformed source.

Main Results:

• Achieved complete vertex correspondence across a set of 3D aortic root meshes.
• Demonstrated accurate mesh registration with an average Hausdorff distance of 3.65mm.
• Obtained an average point-to-mesh distance of 0.41mm, indicating high precision (within one voxel).

Conclusions:

• The proposed method effectively establishes dense correspondences from sparse landmarks in 3D aortic root meshes.
• This approach enhances the accuracy of statistical shape modeling for complex anatomical structures.
• The technique provides a robust solution for vertex correspondence challenges in medical mesh analysis.