Morphological variation in paediatric lower limb bones.

Area of Science:

• Orthopedics
• Biomechanical Engineering
• Medical Imaging Analysis

Background:

• Generating accurate paediatric musculoskeletal geometry is challenging, with current methods like scaling adult geometry being imprecise and medical imaging being costly and time-consuming.
• A need exists for accessible and accurate methods to create paediatric bone geometry for research and clinical applications.

Purpose of the Study:

• To develop and validate a population-based shape model for generating paediatric lower limb bone geometry.
• To improve the accuracy of paediatric bone geometry prediction compared to existing methods.

Main Methods:

• Segmented pelvis, femur, and tibia/fibula from 333 paediatric CT scans (ages 4-18).
• Applied principal component analysis (PCA) to capture bone morphology variation.
• Developed and validated a shape model using demographic and linear bone measurements, with leave-one-out analysis.

Main Results:

• The shape model accurately predicted bone geometry with low root mean squared errors (RMSE): 2.91 mm (pelvis), 2.01 mm (femur), and 1.85 mm (tibia/fibula).
• This model demonstrated significantly higher accuracy than linear scaling of adult geometry (RMSEs ranging from 4.38 to 4.79 mm).
• Principal component analysis successfully captured growth-related changes in bone geometry.

Conclusions:

• A novel shape model effectively captures and predicts paediatric lower limb bone shape variation.
• This method offers a more accurate and accessible alternative to medical imaging for paediatric bone geometry generation.
• The developed model has implications for improving paediatric musculoskeletal research and clinical practice.