Inter-point procrustes: identifying regional and large differences in 3D anatomical shapes

Area of Science:

• Medical imaging
• Computational anatomy
• Biomedical engineering

Background:

• Standard Procrustes analysis struggles with non-Gaussian residuals in 3D anatomical structure alignment.
• Large and localized shape differences pose challenges for existing alignment techniques.

Purpose of the Study:

• To develop a robust alignment method for 3D anatomical structures with substantial shape variations.
• To improve the interpretation of regional differences in anatomical shapes.

Main Methods:

• A novel approach identifies and displaces influential points to create an intermediate template.
• Pose-invariant shape variables guide point detection and displacement.
• The intermediate template facilitates robust estimation of final pose parameters.

Main Results:

• The proposed technique demonstrates robustness against up to 50% regional shape differences.
• Validation on synthetic and real datasets confirms the method's effectiveness.
• The approach shows potential for accurate shape classification.

Conclusions:

• The new method offers a robust solution for aligning 3D anatomical structures with significant shape variations.
• It effectively highlights regional differences and has potential applications in shape classification.