Femoral and Tibial Torsion Measurements With 3D Models Based on Low-Dose Biplanar Radiographs in Comparison With Standard CT Measurements

Abstract

The purpose of this study was to evaluate the interchangeability of femoral and tibial torsion measurements obtained with 3D models based on low-dose biplanar radiographs and standard CT measurements by testing the following two hypotheses: that there is excellent agreement between the two methods and that there is excellent interreader agreement. Two independent readers used 3D models based on low-dose simultaneous biplanar radiographs and axial CT images to measure femoral and tibial torsion in 35 patients (mean age, 65 years; range, 46-89 years) with osteoarthritis of the knee who were to undergo prosthesis insertion. The two measurements were compared by means of Bland-Altman plots and descriptive statistics. Interreader agreement was quantified with intraclass correlation coefficients. The average differences between readers on the CT measurements were 1.3° (range, 0°-11°) for the femur and 1.5° (range, 0°-12°) for the tibia. The average differences for the measurements obtained with the 3D model were 0.1° (range, 0°-9°) for the femur and 0.8° (range, 0°-10°) for the tibia. The average differences between the two methods were 0° (range, -5° to 7°) for the femoral measurements and 3° (range, -12° to 5°) for the tibial measurements. Bland-Altman plots showed no relevant differences between the results of the two measurement modalities. Except for one measurement of femoral torsion and one measurement of tibial torsion, all results based on the 3D models were within the 95% limit of agreement (mean ± 1.96 SD). Interreader agreement was statistically significant (p < 0.001) for all measurements with high intraclass correlation coefficients (> 0.9). Femoral and tibial torsion measurements obtained with 3D models based on biplanar radiographs are interchangeable with standard CT measurements in patients with osteoarthritis of the knee.