SummaryA finite element modelling pipeline was adopted to predict femur strength in a retrospective cohort of 100 women. The effects of the imaging protocol and the meshing technique on the ability of the femur strength to classify the fracture and the control groups were analysed.IntroductionThe clinical standard to estimate the risk of osteoporotic hip fracture is based on the areal bone mineral density (aBMD). A few retrospective studies have concluded that finite element (FE)-based femoral strength is a better classifier of fracture and control groups than the aBMD, while others could not find significant differences. We investigated the effect of the imaging protocol and of the FE modelling techniques on the discriminatory power of femoral strength.MethodsA retrospective cohort of 100 post-menopausal women (50 with hip fracture, 50 controls) was examined. Each subject received a dual-energy absorptiometry (DXA) exam and a computed tomography (CT) scan of the proximal femur region. Each case was modelled a number of times, using different modelling pipelines, and the results were compared in terms of accuracy in discriminating the fracture and the control cases. The baseline pipeline involved local anatomical orientation and mesh morphing. Revised pipelines involved global anatomical orientation using a full-femur atlas registration and an optimised meshing algorithm. Minimum physiological (MPhyS) and pathological (MPatS) strengths were estimated for each subject. Area under the receiver operating characteristic (ROC) curve (AUC) was calculated to compare the ability of MPhyS, MPatS and aBMD to classify the control and the cases.ResultsDifferences in the modelling protocol were found to considerably affect the accuracy of the FE predictors. For the most optimised protocol, logistic regression showed aBMDNeck, MPhyS and MPatS to be significantly associated with the facture status, with AUC of 0.75, 0.75 and 0.79, respectively.ConclusionThe study emphasized the necessity of modelling the whole femur anatomy to develop a robust FE-based tool for hip fracture risk assessment. FE-strength performed only slightly better than the aBMD in discriminating the fracture and control cases. Differences between the published studies can be explained in terms of differences in the modelling protocol and cohort design.

No abstract

No abstract

No abstract