Total knee arthroplasty three-dimensional kinematic estimation prevision. From a two-dimensional fluoroscopy acquired dynamic model.

INTRODUCTION To determine six-degree of freedom of total knee arthroplasty kinematics (TKA), optimized matching algorithms for single fluoroscopic image system may be used. Theoretical accuracy of these systems was reported. Nevertheless, all reports were done under idealized laboratory experimental conditions. The aim of this study was to evaluate the "true" accuracy of a flat panel single plane video-fluoroscopy system based on computed-assisted design (CAD) model matching and compare it to TKA kinematics obtained from optoelectronic measurements as gold standard. HYPOTHESIS The estimation of the error produced by 2D/3D fluoroscopic registration in daily practice is misjudged in most available laboratory reports. MATERIAL AND METHODS The experimental set-up used a TKA implanted into femoral and tibial cadaver bones. Thirty flexions were simultaneously registered using single plane fluoroscopy and an active optical tracking system. Kinematics registered were compared using the root mean square error (RMS), the concordance correlation coefficient and Bland & Altman plot analysis. RESULTS The mean range of motion of flexion during the experiment was 106°. The respective RMS for flexion, varus-valgus and internal-external rotation were 0.68, 0.67 and 1.02°. The respective RMS for antero-posterior, medio-lateral and proximo-distal displacement were 1.3, 2.4 and 1.06 mm. Extreme values of the measured error concerning medio-lateral displacement were -5.4 and 22,1mm. DISCUSSIONS Analysis found some outliners in all degree of freedom with a systematic error and larger standard deviation than already published data. One should make sure that during the experiment the motion of interest is in the in-plane direction. Moreover, this study brings out the true threshold detection of this type of analysis.