CT-based finite element simulating spatial bone damage accumulation predicts metastatic human vertebrae strength and...

Area of Science:

• Biomechanics
• Medical Imaging
• Computational Modeling

Background:

• Pathologic vertebral fractures in spinal metastases significantly impact patients.
• Current understanding of the mechanical factors causing these fractures is limited, hindering predictive capabilities.

Purpose of the Study:

• To develop a damage-based finite element (FE) framework for predicting vertebral strength and stiffness.
• To evaluate the accuracy of specimen-specific versus global material calibration in FE models derived from CT data.

Main Methods:

• Developed a damage-based FE framework utilizing X-ray computed tomography (CT) data.
• Evaluated global and specimen-specific material calibration methods on osteosclerotic, osteolytic, and mixed lesion vertebrae.
• Used a machine learning approach to derive vertebral lesion types.

Main Results:

• The FE framework with global calibration accurately predicted vertebral stiffness (R² = 0.90) and strength (R² = 0.83).
• Specimen-specific calibration yielded near-perfect predictions for both stiffness and strength (R² = 0.99).
• Simulations revealed distinct spatial damage patterns between osteosclerotic and osteolytic vertebrae.

Conclusions:

• The developed FE framework reliably predicts pathologic vertebral strength and stiffness.
• This approach shows potential for predicting fracture onset and location in image-based diagnostics.
• Understanding damage evolution pathways can guide the development of predictive tools for impending vertebral fractures.