Landmark-free geometric methods in biological shape analysis.

Area of Science:

• Computational geometry
• Differential geometry
• Geometric morphometrics

Background:

• Comparing 3D shapes is crucial in various scientific fields.
• Existing methods often rely on landmark identification, which can be subjective and labor-intensive.
• Developing objective and robust shape comparison metrics is an ongoing challenge.

Purpose of the Study:

• To propose a novel, landmark-free approach for computing distances between 3D shapes.
• To introduce a discrete conformal mapping technique minimizing a novel symmetric deformation energy.
• To establish a well-behaved metric on the space of genus zero surfaces.

Main Methods:

• Input: Pairs of topologically equivalent genus zero triangulated surfaces.
• Method: Construct a discrete conformal map minimizing a symmetric deformation energy (Esd).
• Metric: The energy of the minimizing map serves as the shape distance.

Main Results:

• The proposed method yields a well-behaved metric on the space of genus zero surfaces.
• The approach is landmark-free, differentiating it from many existing techniques.
• Demonstrated success in shape recognition and identifying evolutionary patterns in primate skeletal data.

Conclusions:

• The novel discrete conformal mapping approach provides an effective and robust method for 3D shape comparison.
• This landmark-free technique shows significant promise for applications in geometric morphometrics and evolutionary biology.
• The method achieves competitive or superior performance compared to expert observers in analyzing complex biological datasets.