Abstract
Three-dimensional (3D) weight-bearing alignment of the lower extremity is crucial for understanding biomechanics of the normal and pathological functions at the hip, knee, and ankle joints. In addition, implant position with reference to bone is a critical factor affecting the long-term survival of artificial joints. The purpose of this study was to develop a biplanar system using a slot-scan radiography (SSR) for assessing weight-bearing alignment of the lower extremity and for assessing implant positioning with respect to bone. A SSR system (Sonial Vision Safire 17, Shimadzu, Kyoto, Japan) with a custom-made rotation table was used to capture x-ray images at 0 deg and 60 deg relative to the optical axis of an x-ray source [Fig.1]. The SSR system uses collimated fan beam x-rays synchronized with the movement of a flat-panel detector. This system allows to obtain a full length x-ray image of the body with reduced dose and small image distortion compared with conventional x-ray systems. Camera calibration was performed beforehand using an acrylic reference frame with 72 radiopaque markers to determine the 3D positions of the x-ray source and the image plane in the coordinate system embedded in the reference frame. Sawbone femur and tibia and femoral components of the Advance total knee system (Wright Medical Technology, Arlington, TN, USA) were used. Computed tomography of the sawbone femur and tibia was performed to allow the reconstruction of the 3D surface models. For the component, the computer aided design (CAD) model provided by the manufacturer was used. Local coordinate system of each surface model was defined based on central coordinates of 3 reference markers attached to each model. The sawbone femur and tibia were immobilized at extension, axial rotation, and varus deformity and were imaged using the biplanar SSR system. The 3D positions of the femur and tibia were recovered using an interactive 2D to 3D image registration method [Fig.2]. Then, the femoral component was installed to the sawbone femur. The 3D positions of the femur and femoral component were recovered using the above-mentioned image registration method. Overall, the largest estimation errors were 1.1 mm in translation and 0.9 deg in rotation for assessing the alignment, and within 1 mm in translation and 1 deg in rotation for assessing the implant position, demonstrating that this method has an adequate accuracy for the clinical usage.
