Measurement error in μCT-based three-dimensional geometric morphometrics introduced by surface generation and...

Area of Science:

• Geometric morphometrics
• Comparative anatomy
• Bioimaging

Background:

• Computed tomography (CT)-derived polymesh surfaces are crucial for geometric morphometric studies.
• Processing parameters (scanning, resolution, segmentation) can introduce artefactual variance in 3D landmark data.
• Systematic assessment of measurement error in CT-based morphometrics is needed.

Purpose of the Study:

• To systematically assess artefactual variance in landmark data from CT-derived surfaces.
• To compare the magnitude of this variance to true biological variation.
• To identify key factors contributing to measurement error in geometric morphometrics.

Main Methods:

• Generated multiple CT-derived surface variants of anuran pectoral girdles.
• Systematically varied voxel size, segmentation, and surface simplification.
• Acquired 24 landmarks repeatedly by different observers.
• Utilized random-factor nested permanovas to assess variance contributions.

Main Results:

• Factors other than voxel size significantly contributed to measurement error.
• Intra-observer error was the primary source of artefactual variance (<6.75% total variance), followed by inter-observer error.
• Segmentation and slight surface simplification had minimal impact (<1%).
• Observer consistency is critical, especially with inexperienced individuals.

Conclusions:

• Intra-observer error can be minimized through training and single-session landmark acquisition.
• Careful surface simplification and optimized segmentation aid error reduction.
• CT-derived landmark data are sufficiently precise for anuran pectoral girdle shape studies.
• Systematic measurement error assessment is recommended for all geometric morphometric studies.