Model-based shape matching of orthopaedic implants in RSA and fluoroscopy
A.H. Prins · Leiden Repository (Leiden University) · 2015
Model-based shape matching is commonly used, for example to measure the migration of an implant with Roentgen stereophotogrammetric analysis (RSA) or to measure implant kinematics with fluoroscopy. The aim of this thesis was to investigate the general usability of shape matching and to improve the reliability of shape matching for RSA and fluoroscopy. Several improvements were made to the usability and reliability of model-based shape matching for RSA and fluoroscopy. An automatic contour detection approach was proposed, integrating directly into the estimation of position and orientation of the implant. The accuracy of such estimation was also improved by taking the relation between implant parts into account. The robustness of model-based shape matching was assessed when it is applied fully automatically, demonstrating that manual supervision remains necessary. The standardization of validating new fluoroscopic methods could be a welcome addition to the biomechanical field, making it a lot easier to compare results between methods and experiments. As the technology of fluoroscopy advances, the amount of high quality data is getting larger, and may soon grow beyond the capabilities of manual, supervised pose estimation methods. It is therefore imperative that fully automatic unsupervised methods are developed.