Moving FE-based predictors of femoral fracture closer to the clinics: a multi-perspective analysis of the Bologna biomechanical computed tomography solution

Fragility fractures are the primary clinical manifestation of osteoporosis and a major cause of morbidity and mortality, posing a significant healthcare burden. Currently, the gold standard for osteoporosis diagnosis relies on areal bone mineral density (aBMD) derived from dual-energy X-ray absorptiometry (DXA). However, nearly half of patients classified as non-osteoporotic sustain fragility fractures. To address these limitations, computed tomography (CT)-based finite element (FE) models have been developed. Within this context, the Bologna Biomechanical Computed Tomography–hip (BBCT-hip) methodology was developed to estimate an individual’s absolute risk of hip fracture (ARF0) by combining a mathematical–stochastic fall model to compute subject-specific impact loads, with a patient-specific FE model of the femur to predict load to failure under sideways fall conditions. This PhD aimed to enhance the clinical applicability of BBCT-hip through methodological improvements and validation. A semi-automated femur segmentation procedure was developed and validated, reducing processing time and operator dependence. A phantomless calibration strategy using internal reference tissues enabled accurate density estimation without external phantoms, extending applicability to opportunistic CT scans. Impact force calculation was refined by incorporating CT-derived trochanteric soft tissue thickness (STT), including a method suitable for partially cropped scans. Clinical translation was addressed by identifying low-dose CT protocols that ensure the accuracy of patient-specific FE models. Validation across three independent cohorts (356 subjects: 63 fractured, 293 non-fractured) compared its discrimination performance with the DXA-derived T-score. Two predictive indices were evaluated: ARF0BMI, which included BMI-based estimates of STT, and ARF0STT, which incorporated subject-specific CT-derived STT values. Both effectively discriminated between fracture and non-fracture subjects, but ARF0STT consistently showed superior performance, with higher stratification accuracy than the DXA-derived T-score, confirming that BBCT-hip more accurately assesses fracture risk. Overall, this thesis strengthened the BBCT-hip pipeline and provided robust evidence for its use as a reliable, subject-specific tool for hip fracture risk assessment.