Accuracy of reconstructed angular estimates obtained with the Ariel Performance Analysis System

Three-dimensional computerized gait analysis continues to grow in use among physical therapists and other clinical specialists interested in quantitative data regarding human ambulation. This study documented the accuracy of reconstructed angular estimates under static and dynamic conditions using the Ariel Performance Analysis System.

Angular velocity was systematically increased by raising the release position of a T-shaped pendulum. Angular velocities were examined by releasing the pendulum from four angles (0 degree-static, 45 degrees, 90 degrees, and 120 degrees). Twelve reference angles were estimated over 20 autodigitized frames for 10 trials at each release position. Intraclass correlation coefficients (ICCs) and analysis-of-variance (ANOVA) procedures were used to test the hypothesis that the error of angular estimates grows with increasing angular velocity.

Mean errors of the reconstructed angles were consistently within +/- 1.0 degree, regardless of angular velocity. An ANOVA revealed a statistically significant angular velocity effect, characterized by release position. The 90-degree release position produced the greatest error, followed by the 120-, 45-, and 0-degree release positions. The error was not significantly different between the 120- and 45-degree release positions. Intraclass correlation coefficients greater than .90 were found for all frame-to-frame angular velocities, except for the 90-degree release position. The angle estimates consistently underestimated the reference angles, regardless of release position.

The results suggest that clinically accurate angular estimates can be obtained across the range of angular velocities used in this study.