Abstract
Background: Osteoporosis is a bone disease due to bone mass reduction and skeletal microarchitecture deterioration. In osteoporotic bone, the bone mass absorption is higher than its reproduction, which results in highly porous and low-dense bone. Undetected osteoporotic cases can result in vertebral fractures. These fractures can cause permanent disability. Currently, WHO-recommended two-dimensional bone density measures T-score and Z-score derived from dual X-ray energy absorptiometry (DXA) could not detect the osteoporotic cases accurately, and this method’s efficiency is <50% for the fracture predictions. Recently developed web-based and user-friendly statistical tool - fracture risk assessment tool (FRAX) efficiency in fracture prediction is also low due to multiple limitations. Finite element (FE) based vertebral models have been developed for understanding the biomechanics of the spine. The FE-based methods are accurate, but their application in the clinic is limited due to the requirement to use high-resolution images acquired in research settings. Also, the existing FE methodologies were developed mainly to study only the kinematics of the spine, but the fracture risk prediction was not given much importance. It is crucial to detect the fractures early to provide better health care. The thesis focuses on the feasibility of using non-invasive FE-based patient-specific spine models developed from low-dose and routine clinical data for the early prediction of osteoporosis and subsequent fracture risk. Objective: The primary purpose of this thesis is to develop comprehensive patient-specific vertebrae, functional spine unit, and lumbar spine models using finite element analysis. The research objectives were: i) to understand the applicability of purposely acquired reduced dose data for the modeling of vertebrae, ii) to study the feasibility of using image data acquired in routine clinical settings for the modeling of vertebrae, iii) to predict incidental fractures using the baseline routine clinical data, iv) to model and validate patient-specific functional spine unit using in-vitro image data, and v) to understand the applicability of using clinical data acquired at clinical settings for the modeling of the lumbar spine. Results: In study one, we observed that the dose reduction of 90% through sparse sampling does not affect the FE-based vertebral failure load. A dose reduction of 50% thought tube current reduction showed minimal effect on the vertebral failure load values derived from FE analysis. This study observed that dose reduction through sparse sampling performed considerably better than the tube current dose reduction. In the second study, we also observed a good correlation in FE failure load values between routine clinical data and high radiation intensity data (R2 = 0.87). This study shows that routine data can be used for downstream finite element analysis. In the third study, we observed that compared to BMD alone, the combination of normalized load, displacement, and BMD was able to accurately predict the incidental fractures (AUC = 0.54 for BMD vs. ACU = 0.77). In this study, we have used the baseline routine clinical data to predict the incidental fractures that will occur in the future. The study showed that structural biomarkers like displacement and failure load derived from FE analysis could be used for fracture prediction. The fourth study developed and validated an analysis workflow to model the comprehensive functional spine unit. A strong correlation was observed (r = 0.79) between the experimental bone strength, and FE predicted failure load values for the functional spine unit. Finally, in the fifth study, we have developed a comprehensive lumbar spine model with routine clinical data and validated with the literature values. The lumbar failure load for the healthy subjects was 1361.40 ± 135.91 N is in the range of the experimental values 967 N to 4387 N. Range of motion values of the lumbar spine were also within the range of the literature data. Conclusion: Overall, the FE models of vertebrae, FSUs, and lumbar spine developed from low dose data and routine clinical data were able to predict the bone strength, fracture risk, and ROM accurately. We have also observed that the FE-based biomarkers like failure load and failure displacement along with BMD values could predict the fragility fractures accurately. Thus, the FE patient-specific models can be used in the clinic as an augmentation method to the existing BMD-based methodologies for accurately assessing bone health. Keywords: Finite element analysis, multi-detector computed tomography, bone mineral density, the area under the curve, Fracture risk, and osteoporosis