Abstract
Introduction
3D-to-2D model registration technique has been used for evaluating 3D kinematics from 3D surface models of the prostheses or bones and radiographic image sequences. However, no studies have employed these techniques to evaluate in vivo hip kinematics under dynamic weight-bearing conditions. The purposes of this study were to evaluate kinematics of healthy hips and also hips with osteoarthritis (OA) prior to total hip arthroplasty (THA) during four different weight-bearing activities using 3D-to-2D model-to-image registration technique.
Measurement
Dynamic hip kinematics during gait, squatting, chair-rising, and twisting were analyzed for six healthy subjects and eleven patients with osteoarthritis (OA). Continuous anteroposterior radiographic images were recorded using a flat panel X-ray detector (Fig. 1), and each hip joint was scanned by computed tomography (CT). The 3D positions and orientations of the pelvis and femur in movement cycle were determined using a 3D-to-2D model-to-image registration technique. A matching algorithm maximizing correlations between density-based digitally reconstructed radiographs from CT data and the radiographic images was applied (Fig. 2). The relative positions and orientations of the pelvis with respect to the world coordinate systems were defined as pelvic movements (anterior-posterior tilt, contralateral-ipsilateral rotation, Fig. 3b and c), and those of the femur with respect to the world coordinate systems were defined as femoral movements (flexion-extension, internal-external rotation, Fig. 3d). We also defined the relative positions and orientations of the femur for the pelvis as hip movements (flexion-extension, internal-external rotation, Fig. 3e and f).
Accuracy evaluation experiment
The pelvis and femur of a pig carcass fixed to a stage were rotated and translated to known values. The 3D-to-2D model-to-image registration process was performed for the radiographic images at each position to determine the relative pose of each bone. The root-mean-square (RMS) errors of the pelvis and femur were calculated.
Result
For gait, chair-rising, and squatting, the maximum hip flexion-extension of OA patients (average: 22°, 63°, and 65°, respectively) was smaller than those of healthy subjects (30°, 81°, and 102°, respectively), but the minimum hip flexion-extension was not significantly different between healthy (1°, −3°, and 0°, respectively) and OA (2°, 3°, and −3°, respectively) hips. The pelvis of OA patients tended to tilt more anteriorly (−9°) for gait and more posteriorly (18° and 24°, respectively) for chair-rising and squatting than that of healthy subjects (−6°, 12°, and 11°, respectively). For twisting, OA patients demonstrated smaller internal and external hip rotation (0° and 16°, respectively) compared to healthy subjects (29° and 31°, respectively). The RMS errors of the pelvis and femur were 0.21 mm and 0.15 mm in the in-plane direction, 0.14 mm and 0.23 mm in the out-of-plane direction, and 0.25° and 0.23° in rotation, respectively.
Conclusion
3D-to-2D model registration techniques could evaluate accurately in vivo hip kinematics during weight-bearing activities. The current study demonstrated that limited hip range of movement of OA patients was compensated by pelvic tilt during gait and squatting. OA patients demonstrated restriction in hip internal rotation even under dynamic conditions during twisting. Pathological changes due to OA may influence post-THA hip kinematics.
