Gait Phase Discrimination Based on a Foot-Mounted Gyroscope

Abstract

Gait phase discrimination is the temporal gait parameters analyzing issue. In most gait phase discrimination algorithm based on the foot-mounted gyroscope, thresholds are used to detected gait events, which is detrimental to the robustness of detection results. In order to improve the adaptability of the detection of gait events (especially heel strike and toe off), this paper proposes and evaluates a new segmented dynamic threshold method based on local peak detection for gait phase discrimination using the pitch angular velocity. Four typical gait events including heel strike (HS), toe strike (TS), heel off (HO) and toe off (TO) are all detected. In addition, we detect two new gait events, including the top position of backswing (BT) and the top position of forward swing (FT). The results of experiment show that the mean time differences of the detection between the presented algorithm and the reference system are 48.83 ± 4.97ms for HS, 31.12 ± 11.18ms for TS, −65.37 ± 21.35ms for HO and −31.06 ± 12.37ms for TO, respectively. Furthermore, the proposed algorithm can detect HS and TO efficiently at the different walking speed.