Patient specific optimization and planning of vertebral augmentation in osteoporosis: Finite element and Artificial Intelligence approaches.
Authors
Affiliations (6)
Affiliations (6)
- Department of Mining Engineering, Indian Institute of Engineering Science and Technology, Howrah, West Bengal 711103, India.
- Department of Aerospace Engineering and Applied Mechanics, Indian Institute of Engineering Science and Technology, Howrah, West Bengal 711103, India.
- Department of Mechatronics and Automation Engineering, NIT Patna, Bihar 800005, India.
- Department of Physiology, Institute of Post Graduate Medical Education and Research (IPGMER), Kolkata, West Bengal 700071, India.
- Department of Mining Engineering, Indian Institute of Engineering Science and Technology, Howrah, West Bengal 711103, India. Electronic address: [email protected].
- Department of Aerospace Engineering and Applied Mechanics, Indian Institute of Engineering Science and Technology, Howrah, West Bengal 711103, India. Electronic address: [email protected].
Abstract
Osteoporotic Vertebral Compression Fractures (OVCFs) are responsible for significant disability, especially among individuals with low Bone Mineral Density (BMD). Although vertebral augmentation methods such as percutaneous vertebroplasty and kyphoplasty offer rapid pain relief and stability, the optimal combination of bone cement properties like volume, elasticity, and distribution remains uncertain. This study aimed to optimize these parameters using a hybrid Finite element (FE) analysis and Artificial Intelligence (AI) framework integrating Artificial Neural Networks (ANN) and Genetic Algorithms (GA). CT-derived thoracolumbar (T12-L2) models from one healthy male cadaveric CT dataset was used to create 261 FE analysis models simulating various cement volumes (2, 4, 6 and 8 ml), elasticity (510, 1434, 3000 and 5193 MPa), geometries (4 shapes) and bone conditions (80%, 70%, 65%, and 60% stiffness), generated by stiffness scaling from a single healthy cadaveric finite element model. Von Mises stress and strain were evaluated with simulations under compressive 400N load and integrated with AI. Results showed that moderate cement elasticity (∼3000 MPa) and volumes (4-6 ml) restored near physiological stress (6 MPa) and strain, resulting in a maximum reinforcement with minimal adjacent-level stress. Excessive elasticity (>5000 MPa) and volume (>6 ml) of bone increased stress transfer, leading the risk of adjacent fractures in severely osteoporosis. Cement volume and modulus had the greatest influence, followed by bone quality and cement geometry. The ANN model trained with the data from FE analysis (R = 0.9327 for strain; R = 0.8943 for stress model) was integrated in GA-based multi-objective optimization to minimize deviation from physiological stress and strain. This integrated FE-AI framework which combines FE analysis, ANN and GA facilitates patient specific intelligent optimization of vertebral augmentation and holds immense potential for future clinical decision support.