ABSTRACT
Model-based tracking (MBT) is a time-consuming and semiautomatic approach, and thus subject to errors during the tracking process. The present study aimed primarily to quantify the effects that interpolation and intra-user variability associated with MBT have on the kinematic and arthrokinematic measurements in comparison to a gold standard radiostereometric analysis (RSA). Cadaveric knee specimens were imaged at 125 Hz while simulating standing, walking, jogging, and lunging motions. (Arthro)kinematic metrics were calculated via MBT without interpolation, MBT with two interpolation techniques when every fifth or tenth frame was analyzed, and RSA. Tracking the same activity multiple times affected (p-value, largest mean difference) the flexion-extension (FE) joint angle during walking (0.03, 0.6°), and the internal-external joint angle during jogging (0.048, -0.9°). Only during jogging for the FE joint angle was there an effect of interpolation (0.046, 0.3°). Neither tracking multiple times nor interpolation affected arthrokinematic metrics (contact path locations and excursions). The present study is the first to quantify the effects that intra-user variability and interpolation have on the (arthro)kinematic measurement accuracy using MBT. Results suggest interpolation may be used without sacrificing (arthro)kinematic outcome measurement accuracy and the errors associated with intra-user variability, while small, were larger than errors due to interpolation.
