Visualization and quantification of in vivo femorotibial contact in total knee arthroplasty by X-ray fluoroscopic image analysis

Abstract

In this study, we present a novel technique to evaluate in vivo femorotbial contact in total knee arthroplasty (TKA) during dynamic motion. Knowledge of joint contact and its area in TKA during dynamic motion could not only provide kinematic information dependent on articular shape but also assist in improving the design of TKA that include tibial polyethylene insert. In previous studies, joint contact in TKA has been evaluated using pressure sensitive film or digital tactile sensor. However these techniques are limited because they are only able to operate under static in vitro or quasi-dynamic conditions, and cannot be physically applied to TKA patients under dynamic in vivo conditions. To materialize evaluation of femorotibial contact under dynamic in vivo conditions, we employ a computer vision technique called 2D/3D registration, which estimate the 3D pose of radiopaque metallic femoral and tibal components using X-ray fluoroscopic image and computer-aided design (CAD) model of the implant. Although the polyethylene insert is radiolucent and does not appear on fluoroscopic image, in fixed-bearing TKA, the insert can be assumed to be fixed on the tibial tray. Therefore, the 3D pose of the insert can be easily determined, as the pose of tibial component was estimated. To visualize and quantify femorotibial contact, the proximity between surfaces of femoral component and insert is calculated, and mapped onto the insert surface model. The femorotibial contact was visualized and quantified as the region on the insert surface where the proximity is less than 0.5 mm threshold, and the threshold value was determined from the result of 3D pose estimation accuracy by phantom experiments. In clinical applications, the evaluation of femorotibial contact was performed using images from a fixed-bearing TKA patient during dynamic motion. Dynamic states of the contact were visualized and quantified including axial rotation and unilateral loading during knee flexion, and post-cam contact of posterior stabilized TKA. The present technique provided us new information and enabled us to better understand or predict the location, translation and size of the contact region during in vivo dynamic motion.