Reconstruction of the lower limb bones from digitised anatomical landmarks using statistical shape modelling

Area of Science:

• Biomechanics
• Medical Imaging
• Computational Anatomy

Background:

• Accurate bone geometry is crucial for musculoskeletal models in motion analysis.
• Clinical settings often lack medical imaging for precise bone shape determination.
• Current methods rely on scaling techniques using reference subjects, which can limit accuracy.

Purpose of the Study:

• To evaluate if statistical shape models (SSMs) using skin-based measurements can improve lower limb bone geometry prediction.
• To compare SSM-based predictions with conventional linear scaling methods.
• To assess the accuracy of SSMs in correcting for soft tissue artifacts.

Main Methods:

• Developed SSMs from 35 adult femur and tibia/fibula datasets.
• Reconstructed bone shapes by minimizing landmark distances between models and digitized data (skin or bone).
• Quantified soft tissue artifacts using MRI to predict skin-to-bone landmark distances.

Main Results:

• SSM reconstructions from skin and bone landmarks showed higher accuracy (2.60–2.95 mm median error) than linear scaling (3.66–3.87 mm).
• No significant difference in accuracy was found between SSMs using bone landmarks versus skin landmarks.
• SSM reconstructions from skin landmarks demonstrated accuracy comparable to those derived from medical images.

Conclusions:

• SSM reconstructions provide a more accurate method for determining bone shapes from surface data in motion analysis.
• Skin-based landmark digitization with SSMs offers a reliable alternative to traditional scaling methods, especially without medical imaging.
• This approach enhances the precision of kinematic and musculoskeletal models used in clinical motion analysis.