A Combined Experimental-Computational Method to Generate Reliable Subject Specific Models of the Knee's Ligamentous Constraint

Abstract

With the advancement of computational models of the knee, the opportunity exists to utilize patient-specific computational models of the knee intra-operatively to assist surgeons. A critical component for evaluation of whole knee mechanics is configuration of the soft tissue ligament structures surrounding the knee. The overarching purpose of the current research was to develop a unique methodology, utilizing both experimental and computational techniques, for efficient development of patient-specific ligament constraint model. To this end, an experimental method to manually assess knee laxity was developed, and used to evaluate changes in knee laxity after total knee replacement in eight cadaveric specimens. A computational model of ligament constraint was developed to complement the knee laxity data collected during the experimental protocol. A sensitivity study performed on the model identified the most critical ligament parameters affecting knee laxity. Subsequently, these ligament parameters were optimized using the simulated annealing algorithm to minimize the difference between the model predicted knee laxity and the experimentally observed knee laxity for four cadaveric specimens. The optimized ligament parameters were used to predict knee kinematics during an experimental assessment in a quasi-static knee loading rig. Knee kinematic predictions using the optimized ligament parameters were compared to predictions using previously published ligament parameters, and subsequently reduced the RMS difference between the predictions and the experimental kinematics by more than 50% for knee rotations.