Image-guided TKR requires a three-dimensional model of the patient, traditionally provided by preoperative CT scans. Recent developments have focused on navigation systems that eliminate the need for preoperative CT scan. This paper aims to assess the comparative accuracy of prosthesis planning using CT-based and CT-free navigation systems.

Four half-body cadavers were implanted with fiducial markers, four per limb, to provide accurate registration points. Eight orthopaedic surgeons then proceeded to plan the anatomy on each limb twice, using CT scan. The CT-free planning involved digitalisation of the fiducial markers, followed by attachment of trackers to the femur and tibia. Several kinematic and digitalisation steps were taken to produce a set of anatomical coordinates for each limb. Again each surgeon repeated this procedure twice on each limb. Calculation of reproducibility of the mechanical axis as defined by both methods was thus possible.

The overall differences related to varus/valgus placements between the two methods were minor, with a mean of 0.04° (−0.20° to 0.28°) for the femur and 0.19° (−0.009° to 0.39°) for the tibia. The mean angular difference in flexion/extension placement was –0.27° (−0.59° to −0.08°) for the femur and −0.08° (−0.40° to 0.24°) for the tibia.

Results for varus/valgus and flexion/extension, as measured by CT-based and CT-free systems, showed a high degree of concordance. There was no observable bias in either system, as shown by the approximately equal spread of data points on either side of the line of equality. The data show a high degree of reproducibility between CT-free navigation systems and CT-based procedures.