Pose-induced effects of femur and tibia on accuracy of model-based radiostereometric analysis

Abstract The radiostereometric analysis (RSA) technique has been used to monitor intercomponent position and polyethylene wear after total knee arthroplasty. RSA uses anteroposterior (AP) and mediolateral (ML) X-ray images to predict the three-dimensional (3D) relationship of prostheses. However, overlapping images can potentially reduce the accuracy of the predicted RSA results. Overlapping images often occur from improper poses of the lower limb. This study evaluates the pose-induced errors of knee, ankle and hip joints by taking simulated X-ray images of knee prosthesis. The poses include knee flexion only, knee/dorsal flexion, knee flexion and hip abduction and knee/dorsal flexion and hip abduction. Predicted RSA accuracy and pose-induced effects on the overlapping images are evaluated. Two comparison indices are chosen among the poses: percentage of image overlap and rotational and translational RSA errors. The results show that dorsal flexion induces more image overlap than knee flexion. Hip abduction induces image overlap on the ML rather than on AP view. Knee/dorsal flexion and hip abduction induce the worst image overlap and least RSA accuracy. If dorsal flexion is over 10°, the overlap on images gradually increases and the coverage begins to converge at 30°. RSA accuracy substantially decreases when the coverage is over 20%. Furthermore, RSA error increases by 0.3–10 times when overlapping images occurs on the AP and ML views. In conclusion, dorsal flexion is the predominant factor to induce overlap image and decrease RSA accuracy. During taking X-ray images, the ankle should be kept vertical and only knee flexion allowed to ensure the quality of the projected images and RSA results.