Repeatability, reproducibility, and agreement of three computational methods to approximate the functional flexion-extension axis of the tibiofemoral joint using 3D bone models of the femur

Background: Closely approximating the functional flexion-extension (FE) axis of the tibiofemoral joint in 3D models of the femur is important when computing joint motions which are physiologic. The objectives were to 1) develop methods to approximate the functional FE axis based on fitting circles, a tapered cylinder, and spheres to the posterior condyles, 2) determine the repeatability and reproducibility of each method, and 3) determine limits of agreement between pairs of axes.Methods: For each method, the respective axis was determined in forty 3D bone models of the distal femur. Varus-valgus angles and internal-external axial angles were computed relative to standard planes.Results: Repeatability and reproducibility were comparable for the tapered cylinder-based and sphere-based methods and better than that for the circle-based method. Limits of agreement were tightest when comparing the sphere-based and tapered cylinder-based axes. However, limits of agreement for the internal-external axial angle were wide at +3.6° to −3.9° whereas limits of agreement were tighter at +1.4° to −0.7° for the varus-valgus angle.Conclusion: The tapered cylinder-based and sphere-based methods offer advantages of better repeatability and reproducibility over the circle-based method. However, the sphere-based and tapered cylinder-based axes are not interchangeable owing to wide limits of agreement for the internal-external axial angle. The tapered cylinder-based axis is preferred intuitively over the sphere-based axis because the spheres require fitting in both the sagittal and coronal planes whereas the tapered cylinder requires fitting in the sagittal plane only which is the plane of motion in flexion-extension.