Validation of the in vivo volumetric wear measurement for total knee prostheses in model-based RSA

Implant failure related to polyethylene wear remains an important issue in total knee arthroplasty. Polyethylene wear is usually assessed in vivo by measuring the remaining insert thickness on X-ray images of the knee. To reflect the amount of wear debris more accurately, a 3-dimensional overlap measurement has been suggested, which is based on implant component models which are matched on calibrated stereo X-ray images using model-based roentgen stereophotogrammatic analysis. The goal of this study was to determine the influence of pose estimation, insert thickness deviation and variation in the femoral-tibial contact location on the accuracy and precision of the measurement using simulations and a phantom experiment. We found that the pose estimation was the largest source of variation. The 95% prediction interval varied between 111 and 283 mm(3), which is approximately 100-200% of the detected volumetric wear. Insert thickness variation resulted in prediction intervals of 74-174 mm(3). Variation of the femoral-tibial contact location in the phantom experiment gave a prediction interval of 40 mm(3). Large differences in the detected wear volume were found for different flexion angles. At most 56% of the true wear volume was detected (129 of 230 mm(3), 30° of flexion). In summary, both the accuracy and precision of the volumetric wear measurement were low. The prediction interval of the volumetric wear measurement is at least as large as the measurement outcome itself. This is an important limitation to the applicability of the volumetric wear measurement in clinical practice and further clinical validation is required.