Characterizing the Impact of Sampling Rate and Filter Design on the Morphology of Lower Limb Angular Velocities

The angular velocity of lower limb segments during walking is useful for calculating temporal-spatial parameters and joint kinematics in the inertial measurement unit (IMU)-based gait analysis. While many IMU-based methods for gait analysis have been proposed, configurations of data acquisition and signal conditioning parameter have not been standardized across studies. This study examined the effect of sampling rate and filter design on the characteristic angular velocity waveform obtained from IMUs donned on the lower limbs. Specifically, the frequency analysis of this waveform was performed, and the difference in timing and magnitude of the signal at gait events was evaluated relative to the initial signal. A minimum sampling rate of 35 Hz and a lowpass filter (LPF) cutoff of greater than 10 Hz were found to be appropriate parameters for acquisition and conditioning of the lower limb angular velocity signals. The required low minimum sampling rate suggests that gyroscope data are appropriate for use in bandwidth and battery life limited mobile health applications. Our results also showed that while the commonly employed LPF cutoffs of 5 Hz or less may be appropriate for mid-stance detection, they were not optimal for calculation of heel-strike, toe-off, mid-swing, or angular displacement, obtained with heel-strike and toe-off as delimiters.