Purpose: Accurate measurement of dynamic joint motion remains a clinical challenge. To address this problem, we have developed a low-dose clinical procedure using the Roentgen Single-plane Photogrammetric Analysis (RSPA) technique. A validation study was performed in a clinical setting, using a conventional digital flat-panel radiography system.

Method: To validate the technique, three experiments were performed: assessment of static accuracy, dynamic repeatability and measurement of effective dose. A knee joint phantom, imbedded with tantalum markers, was utilized for the experiments. Relative spatial positions of the markers were reconstructed using Radiostereometric Analysis (RSA). A digital flat-panel radiography system was used for image acquisition, and the three-dimensional pose of each segment was determined from single-plane projections by applying the RSPA technique. All images were processed using software developed in-house. To assess static accuracy, the phantom was mounted onto a three-axis translational stage and moved through a series of displacements ranging from 0 to 500 μm. Images of the phantom were acquired at each position. Accuracy was calculated by analyzing differences between reconstructed and applied displacements. To assess dynamic repeatability, the phantom was mounted on a six-axis robot, programmed to apply a flexion-extension movement to the joint. Multiple cine acquisitions of the moving phantom were acquired (30 fps, 4 ms exposure). Repeatability was calculated by analyzing the variation between motions reconstructed from repeated acquisitions. The effective dose of the procedure was measured using an ion-chamber dosimeter. The ion chamber was positioned between the phantom and x-ray source, facing the source. Entrance exposure was measured for multiple acquisitions, from which the effective dose was calculated.

Results: The accuracy determined from the static assessment was 25 μm and 450μm at the 95% confidence intervals for translations parallel and orthogonal to the image plane, respectively. Repeatability of the motion reconstructed from dynamic acquisitions was better than ± 200 μm for translations and ± 0.1 for rotations. The average effective dose for a 6 second dynamic acquisition was approximately 2μSv.

Conclusion: The proposed clinical procedure demonstrates both a high degree of accuracy and repeatability, and delivers a low effective dose.