a novel multi-planar radiography method for three dimensional pose reconstruction of the patellofemoral and tibiofemoral joints after arthroplasty

determining the 3d pose of the patella after total knee arthroplasty is challenging. the commonly used single-plane fluoroscopy is prone to large errors in the clinically relevant mediolateral direction. a conventional fixed bi-planar setup is limited in the minimum angular distance between the imaging planes necessary for visualizing the patellar component, and requires a highly flexible setup to adjust for the subject-specific geometries. as an alternative solution, this study investigated the use of a novel multi-planar imaging setup that consists of a c-arm tracked by an external optoelectric tracking system, to acquire calibrated radiographs from multiple orientations. to determine the accuracies, a knee prosthesis was implanted on artificial bones and imaged in simulated ‘supine’ and ‘weightbearing’ configurations. the results were compared with measures from a coordinate measuring machine as the ground-truth reference. the weightbearing configuration was the preferred imaging direction with rms errors of 0.48 mm and 1.32° for mediolateral shift and tilt of the patella, respectively, the two most clinically relevant measures. the ‘imaging accuracies’ of the system, defined as the accuracies in 3d reconstruction of a cylindrical ball bearing phantom (so as to avoid the influence of the shape and orientation of the imaging object), showed an order of magnitude (11.5 times) reduction in the out-of-plane rms errors in comparison to single-plane fluoroscopy. with this new method, complete 3d pose of the patellofemoral and tibiofemoral joints during quasi-static activities can be determined with a many-fold (up to 8 times) (3.4 mm) improvement in the out-of-plane accuracies compared to a conventional single-plane fluoroscopy setup.