Using a monoplane stereoscopic videoradiography configuration to accurately measure skeletal positions: a...

Biplanar videoradiography is a validated method for measuring skeletal kinematics. However, using two x-ray source-detector pairs introduces geometric constraints that can compromise image quality. The spatial configuration of the two pairs often requires a larger object-to-image distance (OID), leading to greater magnification and geometric blur. Consequently, increased magnification can limit the effective field of view, restricting multi-joint analyses. Additionally, geometric blur degrades image quality, potentially compromising subsequent bone registrations. To address this limitation, this proof-of-concept study evaluated a monoplane stereoscopic videoradiography (MSV) configuration. MSV uses two x-ray sources captured by a single detector, enabling a smaller OID. A pseudo-dynamic experiment was conducted to assess whether MSV would achieve accurate calibration and subsequent three-dimensional (3D) bone positions. A cube and pelvis phantom were rotated 360° on a turntable in 15° increments. At each increment, positions were recorded using both conventional motion capture (MOCAP) and MSV, with MOCAP acting as the reference standard. During MSV image acquisition, a lead barrier alternately blocked one x-ray source, allowing the other to project onto the detector. Calibration was performed using a bundle adjustment technique, and 3D bone positions were calculated using radiostereometric analysis and model-based tracking and compared against MOCAP. Overall, MSV achieved sub-millimeter and sub-degree accuracy. The calibration yielded a mean residual of 0.22 mm, while rotational and translational biases ranged from 0.36-1.16° and 0.22-1.23 mm, respectively, with precisions of 0.09-0.59° and 0.18-0.58 mm. These results support the feasibility of using MSV to accurately calculate skeletal positions. This approach reduces magnification and, by extension, geometric blur, opening avenues for multi-joint analyses and improved imaging of deeper joints. Future work will focus on implementing a true asynchronous MSV acquisition to extend this approach to dynamic,in vivokinematic studies.