Using visual image measurements to validate a novel finite element model of crack propagation and fracture patterns of proximal femur

Abstract

Listen

Translate article

In this paper, a simple and practical two-dimensional finite element (FE) model coupled to a quasi-brittle damage law has been developed to describe the initiation and progressive propagation of damage of human proximal femur under quasi-static load until complete fracture. In order to validate the model, 10 human proximal femurs were tested till complete fracture under one-legged stance quasi-static load. During each load step, visual image measurements of full-field real-time strain was achieved using a digital image correlation technique consisting in an optical image system with recording cameras linked to a computer with image-processing software. Two-dimensional FE femur models were derived by the projection of micro-computed tomography scans and the specimen fractures were simulated using the same loads and boundary conditions as in the experimental tests. The predicted and optically measured strain field magnitudes and distributions were compared for the 10 specimens. Three femurs were used for calibration of the model and the remaining seven femurs were used for validation. The numerical calibration phase was used to establish the relationship between the FE density and the strain at fracture needed for description of the damage growth. Very good agreement (R2 = 0.89) was obtained between predicted and visualized measured results, indicating that the proposed FE proximal femur fracture model in the quasi-static regime can capture the initiation and propagation of cracks within femurs till complete organ failure. In addition, we show that full-field visual strain measurement provides a much more general and accurate validation than traditional methods based on strain gauges or simple force–displacement curves. The FE model developed here, based on two-dimensional representations of proximal femur geometry and areal bone mineral density distributions, could be applied by clinicians to predict the femur fracture risk of patients using simple and rapid modelling combined with 2D radiographs.