Accurate and Reliable Gait Cycle Detection in Parkinson's Disease.

There is a growing interest in the use of Inertial Measurement Unit (IMU)-based systems that employ gyroscopes for gait analysis. We describe an improved IMU-based gait analysis processing method that uses gyroscope angular rate reversal to identify the start of each gait cycle during walking. In validation tests with six subjects with Parkinson disease (PD), including those with severe shuffling gait patterns, and seven controls, the probability of True-Positive event detection and False-Positive event detection was 100% and 0%, respectively. Stride time validation tests using high-speed cameras yielded a standard deviation of 6.6 ms for controls and 11.8 ms for those with PD. These data demonstrate that the use of our angular rate reversal algorithm leads to improvements over previous gyroscope-based gait analysis systems. Highly accurate and reliable stride time measurements enabled us to detect subtle changes in stride time variability following a Parkinson's exercise class. We found unacceptable measurement accuracy for stride length when using the Aminian et al gyro-based biomechanical algorithm, with errors as high as 30% in PD subjects. An alternative method, using synchronized infrared timing gates to measure velocity, combined with accurate mean stride time from our angular rate reversal algorithm, more accurately calculates mean stride length.