Proposition and Validation of a New Index to Determine the Measurement Change Resolution of Inertial Motion Tracking Systems

Abstract

Orientation data for biomechanical assessment of motion may be obtained from inertial measurement units (IMUs) through the use of a fusion algorithm estimating the orientation of the platform in a fixed and global reference frame from 3D inertial sensors data (accelerometers, gyroscopes, magnetometers). In the current literature, there are evidences that accuracy of the IMUs? estimated orientation varies according to the segment/joint tracked and the movement performed. Typical accuracy studies on IMUs present validation data in the form of root-mean-square difference (RMSD) with a gold standard and/or similarity with recognized gold standard. However, since the error in estimation of the fusion algorithm used by the IMUs is not fully random and is suspected of being somewhat movement-related, this can lead to an over-estimation of the measurement error in a test-retest context. This paper introduces a novel index to determine the Measurement Change Resolution (MCR). The MCR combines the traditional RMSD approach with a reliability index, the Coefficient of Multiple Correlation (CMC) to establish the measurement noise around the actual point of operation of a given IMU. The MCR concept is then tested using orientation data recorded simultaneously with an IMU system and an optoelectronic system in three participants performing repeated gait cycles. Results show that the MCR computed on the maximum range of motion of the knee during walking is a better approximation of the actual resolution of the measure than the traditional error-level estimation using √2 RMSD.