A prototypical system to virtually reconstruct high energy fracture events

Abstract

This paper presents a new computational system that seeks to estimate the dynamics of fracture event from a 3D CT image of the fractured limb and the (assumed) intact contra-lateral limb. A model of the intact limb anatomy prior to injury is estimated from the image of the intact contra-lateral limb, and a model of the fractured limb anatomy after injury is estimated from the image of the fractured limb. Fracture reconstruction software aids in estimating an intact limb model which characterizes the limb in terms of its soft tissues and fractured bone fragments at the time prior to the fracture event. The system then virtually simulates fracture events by impacting the reconstructed limb model using a hypothesized impacting object, i.e. the strike object, and applying a physics-engine to simulate the dynamics of the fracture event. A maximum likelihood estimation framework is applied to score fracture events based on how well they re-create the fracture pattern observed in the fractured limb image. Initial results are shown for one fracture case which demonstrate that solutions to this heretofore unexplored problem are possible using this framework.