Updating a 3-D vertebral body finite element model using 2-D images

Abstract

In osteoporotic patients, vertebral strength is often evaluated in the clinical setting using bone densitometry methods, such as dual energy X-ray absorptiometry. Finite element models based on computed tomography (CT) have been shown to outperform such methods in predicting fracture strength, but repetitive use of CT scans may be impractical due to cost, availability, and radiation exposure. We propose a method of generating a vertebral model at an initial time point which can subsequently be updated using several digital radiographs by using an algebraic reconstruction technique (ART) to find the three-dimensional density distribution. The effectiveness of the algorithm was measured by comparison of the error of the reconstructed model to the error incurred by using the outdated model. Application of the ART was able to reduce density errors from 30% to under 7% and to reduce errors in calculated vertebral stiffness from 125% to under 10%. This preliminary study demonstrates that the method is valid and could potentially improve fracture risk diagnostics drastically.