Abstract
Osteoporotic vertebral fractures (OVFs) are a significant determinant of quality of life, especially in the elderly. Without early intervention, OVFs disrupt daily lifestyle on an individual level since many patients barely return to their previous functional status. The pathological changes induced by osteoporosis remain largely ‘silent’ until a radiological scan is carried out or a fracture occurs. OVFs can be considered as the predominant clinical symptom of osteoporosis. Existing diagnostic methods such as the use of bone mineral density (BMD) derived from dual X-ray energy absorptiometry (DXA) has been proven to be insensitive to the structural deterioration that occurs during the progression of osteoporosis. Thus, patients do not get classified accurately with their severity of osteoporosis. Unfortunately, the clinical attention is on the treatment of OVFs, by when it is too late as patients have already sustained or are highly susceptible to OVFs. The current focus of the thesis is to discover an early prediction workflow using an effective structural biomarker that can predict OVFs and assess the risk of fracture in patients employing non-invasive computational analysis.This thesis presents an investigation on the prediction of vertebral strength using patient-specific finite element (FE) analysis. The primary aims were: i) to design, develop, and validate a vertebral strength assessment workflow using patient-specific FE models of vertebrae with osteoporosis, ii) to assess OVF risk using the developed methodology with patients diagnosed with secondary osteoporosis, and iii) to evaluate the effect of radiation dose reduction strategies on predicting vertebral bone strength and the secondary aims were: iv) to evaluate advanced radiation dose reduction strategies applied on the workflow in predicting vertebral bone strength and v) to investigate comprehensive FE methodologies such as the inclusion of intervertebral discs (IVDs) in the prediction of OVFs.An analysis workflow was developed and validated firstly for FE model generation of rigid vertebral bodies and secondly for functional spinal units (FSUs) based on multi-detector computed tomography (MDCT) images. Compared to experimentally measured vertebral strength, FE models of vertebral bodies (R2 = 0.85) exhibited a better predictor of vertebral strength than bone mineral density (BMD) estimated from MDCT images (R2 = 0.75). This FE workflow was then implemented in a multiple myeloma (MM) cohort, who are at high risk for OVFs. The FE models were able to differentiate between vertebrae with and without OVFs by observing regions of erratic vertebral strength. In order to establish more regular follow-ups, the FE workflow was also evaluated with different dose reduction strategies. Statistical iterative reconstruction (SIR) based FE models (R2 = 0.95) estimated vertebral strength better than the one based on standard filtered back-projection (FBP) (R2 = 0.89). Tube current reduction of mid-vertebral specimens from the highest (500 mAs) to the lowest (80 mAs) had no significant impact on FE models, which led to a further study on patients with standard-dose MDCT of their lumbar vertebrae and a 75% dose reduction was achieved virtually. Compared to experimental values of FSUs, FE models (R2 = 0.93) better predicted vertebral strength than FE models of the respective vertebral bodies, demonstrating the significant role of intervertebral discs. Overall, the FE workflow of both vertebral bodies and FSUs were proven to be an accurate tool for vertebral strength prediction.