Error in maximum total point motion of a tibial baseplate is lower with a reverse-engineered model versus a CAD model using model-based radiostereometric analysis.

Because model-based radiostereometric analysis (MBRSA) identifies tibial baseplate designs which increase risk of baseplate loosening, and because registration errors for computer-aided design (CAD) models are large relative to a 6-month stability limit, 3D models more representative of the geometry of implanted baseplates are needed to minimize error. This study tested whether (1) each of three reverse-engineered (RE) models of the same nominal size reduced registration error relative to the equivalent size CAD model, and (2) RE models of multiple sizes reduced registration error relative to CAD models of corresponding sizes. Registration error, quantified as mean artifactual maximum total point motion (aMTPM), was computed between double biplanar radiographs (i.e., two pairs of independent biplanar radiographs from the same day) for thirty-five patients. Double biplanar radiographs were analyzed four times for the most common baseplate size (i.e., size 5) using three RE models and the corresponding CAD model (1st hypothesis) and twice for all patients using one RE model and the equivalent size CAD model (2nd hypothesis). For all three size 5 RE models, mean aMTPM was less than that of the CAD model, though only one RE model reached statistical significance. For multiple size models, mean aMTPM was reduced by 24% when using RE models instead of CAD models, which could mean the difference between categorizing a baseplate as at-risk versus not at-risk relative to a 6-month stability limit. Since error reduction is related to geometry of specific baseplate designs, other baseplate designs should be evaluated using methods presented herein.