BoneMorphing versus freehand localization of anatomical landmarks: consequences for the reproducibility of implant positioning in total knee arthroplasty.

OBJECTIVE: This study analyzed the influence of the acquisition method in image-free computer-assisted total knee arthroplasty (CAS-TKA), and the reproducibility of implant planning using BoneMorphing, a 3D morphometric model obtained by a 3D-to-3D elastic registration of statistical models to sparse point clouds acquired directly on the bone surface with a pointer. MATERIALS AND METHODS: Five surgeons (one expert, four trainees) each performed a CAS-TKA hybrid protocol based on morphometric models and landmarks on a cadaveric knee 10 times. In addition, several additional landmarks were digitized during each acquisition. The reproducibility of the implant positioning and sizing, as determined by an implant planning algorithm with morphometric models, was compared to direct digitization accuracy. RESULTS: Femoral and tibial implant positioning parameters with the hybrid protocol resulted in intra-surgeon standard deviations (SDs) of less than+/-1.4 degrees for rotation and 1.9 mm for translation for all surgeons in all directions except for tibial axial rotation (the only parameter determined by a digitized landmark and not recomputed in the 3D model). The variability in individual landmark digitization varied from 2 to 5 mm SD for certain landmarks, with ranges of 15-25 mm across all surgeons. The comparison study showed an improvement in femoral rotation reproducibility with the morphometric model when using the posterior condylar axis. Tibial implant reproducibility for each method was comparable, with the morphometric model giving better results in well-digitized areas such as the tibial plateau. CONCLUSION: A CAS-TKA protocol based on a deformed statistical model offers reproducible implant positioning. Some landmarks, such as distal condyles, show sufficient reproducibility in the direction of interest, while others, such as the anterior tibial tubercle, can lead to hazardous implant positioning. This should be taken into consideration when designing a CAS-TKA system with bony landmarks. In areas where a sufficient number of points have been digitized with good coverage, such as on the distal and posterior femoral condyles or the tibial plateau areas, the information derived from the 3D model is more accurate and reproducible than that derived from digitization. Good training and a guiding user interface are essential to guarantee coverage quality.