ヒト関節角度の測定の精度改善: Improving the accuracy of human joint angle measurements

We report a method to reduce the root mean square error (RMSE) of human joint angle measurements using a personal wireless sensor network. The action is based on the virtual rotation of a wireless sensor worn by the user, focusing on the arm, whose position is measured in five degrees of freedom (DOF). The wireless sensor uses an inertial magnetic unit to measure the alignment of the arm with the Earth's gravity and magnetic field. Due to the biomechanical properties of human tissue (e.g., skin elasticity), the orientation of the sensor shifts, and this shift affects the accuracy of the measurement. In the proposed method, the change in orientation is first modeled from a linear regression of data collected from 15 participants at different positions of the arm. Of the eight body indices measured by dual-energy X-ray absorptiometry, the body fat percentage (II) was found to have the greatest correlation with the rate of change of the sensor orientation. These findings allow us to estimate the change in sensor orientation from the user's body fat percentage. Finally, the algorithm rotates the sensor using quaternion theory with the aim of reducing the error. The proposed method is validated by experiments with five different participants. For the DOF, the error was reduced the most, with the RMSE decreasing from 2.20° to 0.87°, an improvement of 60%, and for the reduced degrees of freedom, the RMSE decreased from 1.64° to 1.37°, an improvement of 16%. The average RMSE improved by 44%. Copyright 2016 The Institute of Electrical and Electronics Engineers, Inc. All Rights reserved. Translated from English into Japanese by JST