Abstract
Introduction
Clinical outcomes for total knee arthroplasty (TKA) are especially sensitive to lower extremity alignment and implant positioning.1 The use of computer-assisted orthopaedic surgery (CAOS) can improve overall TKA accuracy.2 This study assessed the accuracy of an image-free CAOS guidance system (Exactech GPS, Blue-Ortho, Grenoble, FR) in both a synthetic leg with a normal mechanical axis and legs with abnormal mechanical axis.
Materials and methods
A high-resolution 3D scanner (Comet L3D, Steinbichler, Plymouth, MI) was used to scan varus-deformed (n=12), neutral (n=12), and valgus-deformed (n=4) knee inserts (Mita M-00566, M-00598, M-00567; respectively, Medical Models, Bristol, UK) and collect pre-identified anatomical landmarks prior to using the models to simulate knee surgery. The image-free CAOS guidance system was then used to acquire the same landmarks. After adjusting the position and orientation of the cutting block to match the targets, bone resections were performed, and the knee models were re-scanned. The 3D scans made before and after the cuts were overlaid and the resection parameters calculated using the pre-identified anatomical landmark data and advanced software (UG NX, Siemens PLM, Plano, TX). Data sets obtained from the 3D scanner (see Figure 1A) were compared with data sets from the guidance system (see Figure 1B). Given the accuracy of the 3D scanner (<50μm), its measurements were used as the baseline for assessing CAOS system error.
Results
Table I shows errors in bone resection thickness orientation measurement errors as well as CAOS system confidence intervals (CI) for both the tibia and femur, depending on deformity type.
Regardless of knee deformity and other parameters, the mean error of the CAOS system was systematically less than 0.5 mm for bone resection measurements and 1° for joint angle measurements. The 95% CI were in the range of −1.54 to 0.67mm for bone resection measurements and −0.64° to 1.67° for joint angle measurements. No statistical differences were detected between different deformity groups in the Error Indexes for both the tibia and femur.
Discussion
This study represents an extension of a previous evaluation of the same CAOS system, where only a limited number of neutral models (n=6) were investigated. The current study was performed to reassess the accuracy and precision of the CAOS system using the same methodology with a larger number of knee models (n=28) exhibiting different types of deformities affecting the mechanical axis.
In conclusion, this study demonstrates a high level of in-vitro accuracy for the CAOS system, regardless of leg-alignment deformity type. The mean error of the CAOS system, characterized as the difference between the measured and checked values, was systematically less than 0.5 mm for bone resection measurements and 1° for joint angle measurements.
