A Kinematic Assessment of Knee Prosthesis from Fluoroscopy Images

We have developed a technique for estimation 3D motion of knee prosthesis from its 2D perspective projections. Our estimation algorithm includes some innovations such as a two-step estimation algorithm, incorporative use of a geometric articulation model and a new method to solve two silhouettes’overlapping problem. Computer model simulations and experiments results demonstrated that our algorithms give sufficient accuracy. Next, with the cooperation of medical surgeons, we assessed the algorithm’s clinical performance by applying it to moving fluoroscopy images of patients who had just undergone TKA recently. Our experiments were done in four steps; first we have taken the moving X-ray pictures called fluoroscopy images of the knee prosthesis at different knee motions; second, introduced the absolute positions/orientations for both components, third, introduced the relative positions/orientations between the femoral and the tibial components and finally, introduced the contact points trajectories between the femur and the tibial insert. We drew the estimation results graphically and made the CAD model pictures of the prosthesis, thereby helping us to assess how the relative motions between the femoral and the tibial components were generated. Estimation results of the clinical applications demonstrated that our algorithm worked well as like as heoretical