Statistical shape models that predict native glenoid width based on glenoid height are inaccurate in their current...

Area of Science:

• Orthopedics
• Radiology
• Biomedical Engineering

Background:

• Unclear measurement errors in statistical shape models for predicting native glenoid width.
• Need to assess accuracy of models using glenoid height for determining anterior glenoid bone loss.

Purpose of the Study:

• Create and validate a statistical shape model for glenoid width using 3D-CT measurements.
• Determine measurement errors of existing 3D-CT statistical shape models.

Main Methods:

• Retrospective cross-sectional study of patients with traumatic anterior shoulder dislocation.
• 3D segmentation and measurement of glenoid height and width from CT scans.
• Calculation of measurement error as percentage deviation from native glenoid width.

Main Results:

• Strong correlation (r=0.80) between glenoid height and native glenoid width.
• Developed model: native glenoid width = 0.75 × glenoid height - 0.61, with 5% ± 4% average error.
• Existing models showed higher errors: Giles et al. (10% ± 7%), Chen et al. (6% ± 5%), Rayes et al. (9% ± 6%).

Conclusions:

• Statistical shape models for estimating glenoid width based on height have unacceptable measurement errors.
• Caution is advised when using these models for glenoid bone loss assessment.
• Using the contralateral side as a reference is preferred to minimize morphologic differences and errors.