Image Based Fracture Prediction Diagnostic Tool for Avascular Necrosis of the Femoral Head

Current methods to diagnose bone diseases like avascular necrosis (AVN) are subjective and a reliable assessment of the fracture risk is not available. A diagnostic fracture prediction tool would aid clinical diagnosis, anticipate disease progression and help with the planning of subsequent interventions. The strength of bones, including the femur, can be calculated using structural mechanics with a view to ascertaining fracture risk. The aim of this thesis was to develop and validate a fracture prediction method based tomographic imaging and beam theory.
In-vitro disease models were created from additive manufacturing, explanted porcine and human femoral heads. The disease models contained a simulated lesion that was either lateral or medial to the fovea to analyse the effects of different lesion positions and to verify the ability of the developed fracture prediction tool. Current classification methods rely on the identification of the lesion volume and location to quantify the fracture risk, an approach that is purely based on geometrical information. The fracture prediction method based on structural stiffness also considered material properties which potentially added predictive capability.
The tool was subsequently validated by predicting the fracture risk of femoral heads from AVN patients to demonstrate the ability to identify necrotic lesions that were likely to progress to fracture. The predicted fracture risk was compared to the current diagnostic gold standard to diagnose AVN. The beam tool was also compared against another novel fracture prediction tool based on FEA to identify possible advantages of beam theory.
The verification tests confirmed that samples with a lesion in the weight bearing area were statistically more likely to fracture at a low load. A low fracture load meant a high fracture risk. However the experimental fracture load of porcine and human femoral heads, even among samples with similar lesions, showed variations indicating that lesion volume and location were not good predictors of fracture risk alone. There was a good correlation between the predicted fracture risk and in-vitro fracture loads of the human femoral head disease model indicating that the developed tool was able to objectively predict the fracture risk. The beam tool had similar good predictive capabilities as current diagnostic methods and fracture prediction methods based on FEA.
An objective in-vivo analysis of the mechanical fracture risk helps identifying patients whose disease is at risk of progressing, as well as stratifying surgical interventions.