SSM

SSM provides a three-dimensional statistical shape model (SSM) of the tibia and MATLAB scripts to generate SSMs and segmented surface models of cortical and trabecular bone for quantifying tibial geometric variability.

Key Features:

• Dataset: Lower limb computed tomography (CT) scans from 30 cadavers are used to produce segmented and reconstructed models of cortical and trabecular tibia and a single surface model of the fibula.
• MATLAB scripts: MATLAB scripts are provided to generate SSMs and segmented surface models of cortical and trabecular bone.
• Multiple statistical shape models: Three distinct SSMs are produced: tibia, tibia-fibula, and cortical-trabecular.
• Principal Component Analysis (PCA): PCA is employed to derive the SSMs, retaining principal components that explain approximately 95% of the geometric variation.
• Variability insights: Identified sources of geometric variation include overall size, midshaft thickness, prominence and size of tibial features (e.g., condyle plateau, tibial tuberosity), axial torsion, and curvature characteristics.
• Clinical relevance: Variations such as general tibial thickness, midshaft thickness, tibial length, and medulla cavity diameter are highlighted as factors that may increase risk of tibial stress injuries.

Scientific Applications:

• Risk Assessment: Quantifying tibial geometry to investigate factors contributing to tibial stress fractures.
• Medical equipment design: Providing geometric data to inform design considerations that account for individual tibial variation.
• Clinical diagnosis: Supporting analysis of bone structure and integrity for conditions related to tibial geometry.

Methodology:

CT scans were acquired from the New Mexico Decedent Image Database; the tibia–fibula were segmented into cortical and trabecular sections and reconstructed; statistical shape models were developed from these segmented surfaces using PCA retaining components that explain ~95% of geometric variation, with MATLAB scripts used to generate the models and segmented surface outputs.