A new fully automated approach for aligning and comparing shapes.

Area of Science:

• Paleontology
• Comparative Anatomy
• Bioinformatics

Background:

• Traditional 3D geometric morphometrics (3DGM) relies on manual landmark placement, introducing researcher interpretation and potential bias.
• Existing 3DGM methods require researchers to designate initial landmarks, influencing shape representation.

Purpose of the Study:

• To present an algorithm for fully automatic placement of correspondence points on 3D digital bone models.
• To enable input into standard 3DGM software and analysis with dimension reduction techniques.

Main Methods:

• An algorithm for automatic correspondence point placement on 3D bone models was developed.
• The algorithm was tested on a dataset of 106 primate calcanei (1,024 points per bone).
• Automated results were compared to a traditional 3DGM study using 27 landmarks per bone, analyzed with morphologika(2.5) and PAST.

Main Results:

• Automated and traditional 3DGM analyses showed strong correlations in principal component scores and similar variance partitioning.
• Generated shape spaces from both methods exhibited similarities.
• Cluster analyses revealed broadly similar patterns, with some notable differences.

Conclusions:

• Automated quantification in 3DGM can produce meaningful shape spaces comparable to observer-based landmarks.
• This approach offers potential for time savings, increased morphological quantification completeness, and elimination of observer error.
• The method allows for broader comparisons of shape diversity across different bone types.