Validation of a single-detector radiostereometric analysis setup for biomechanical research

Introduction Radiostereometric analysis (RSA) enables direct three-dimensional tracking of radiopaque markers placed in internal structures that are inaccessible to external optical measurement systems. The objective was to evaluate the agreement of a custom single-detector RSA system with a validated robotic reference and to compare its measurement performance with that of established optical and tactile measurement systems. It was hypothesized that RSA would show low trial-to-trial variability and measurement performance comparable to the comparator systems. Methods In this controlled laboratory study, programmed translational displacement of 5.00 mm in the anterior (AT), lateral (LT), and superior (ST) directions was performed in a synthetic knee model by a six-degree-of-freedom industrial robot. Ten displacement trials were performed in each direction. Evaluable trial counts were 10 for RSA and a stereo-optical measuring system (SOMS), 7–10 for a tactile measuring system (TMS), and 8–9 for an infrared measuring system (IRMS). Agreement between each system and the robotic reference was characterized by the mean bias with its 95% confidence interval, error standard deviation, 95% fixed-reference error limits, root mean square error (RMSE), and the proportion of errors within ±0.10 mm. Equivalence of the mean bias to zero within ±0.10 mm was assessed using two one-sided tests. Results RSA demonstrated biases of −0.083 mm for AT, −0.133 mm for LT, and −0.028 mm for ST. Error standard deviations ranged from 0.015 to 0.071 mm, and RMSE values ranged from 0.031 to 0.149 mm. Mean-bias equivalence was established for AT and ST, but not for LT. SOMS showed error standard deviations of 0.017–0.029 mm but systematically underestimated displacement in ST. TMS and IRMS showed direction-dependent variability, with error standard deviations of 0.093–0.125 mm and 0.112–0.367 mm, respectively. Discussion The presented RSA system measured static translations with low variability and measurement performance comparable to the established comparator systems. Mean-bias equivalence within ±0.10 mm was demonstrated for AT and ST, but not for LT. These findings support the use of the RSA system for static in vitro biomechanical experiments assessing macroscopic translations.