Reconstruction of Scapula Bone Shapes from Digitized Skin Landmarks Using Statistical Shape Modeling and Multiple...

Area of Science:

• Biomechanical Engineering
• Medical Imaging
• Computational Anatomy

Background:

• Accurate 3D scapula models are crucial for biomechanical analysis and surgical planning.
• Current methods often require extensive medical imaging, increasing cost and complexity.
• Developing non-invasive techniques for scapula shape reconstruction is highly desirable.

Purpose of the Study:

• To develop and validate an algorithm for reconstructing subject-specific scapula bone shapes.
• To utilize skin surface landmarks and a statistical shape model (SSM) for this reconstruction.
• To assess the accuracy and reliability of the developed algorithm.

Main Methods:

• Utilized 56 scapula segmentations and 4 key scapular bone and skin landmarks.
• Developed regression models to predict bone landmark coordinates from skin landmarks, incorporating subject-specific variables (sex, age, weight, height).
• Reconstructed scapula shapes by fitting the SSM's mean shape bone landmarks to the predicted bone landmarks.

Main Results:

• Regression models achieved R² values from 0.70 to 0.98, with a maximum median error of 4 mm.
• Average surface-to-surface errors were 2.41 mm (digitized) and 2.45 mm (predicted bone landmarks).
• No significant statistical differences were found between shapes reconstructed using digitized versus predicted bone landmarks.

Conclusions:

• The developed algorithm reliably derives subject-specific scapula shapes from experimentally acquired data.
• Scapula shape reconstruction using a limited set of landmarks can generate accurate computational models.
• This approach obviates the need for additional medical imaging, offering a more accessible method.