Marker displacement data filtering in gait analysis: A technical note
K. Rácz, Rita M. Kiss
Abstract
The filtering process for marker displacement data in gait analysis has been unchanged for almost 50 years, despite technological advances in measurement equipment. Most researchers use a second-order zero-lag Butterworth low-pass filter with a cutoff frequency of 6 Hz, which was established based on TV camera recordings digitised at 96 × 96 pixels and 60 frames per second. The goal of this study is to examine this technique and determine whether it is still adequate. Data from 15 gait analysis measurements of healthy participants between 20 and 50 years of age were examined, along with data from a publicly available dataset. Based on time-domain data, the current filtering technique causes considerable distortion in the displacement, which results in signal peaks being underestimated by more than 10 mm in some cases, and subtle characteristics of movement are lost. A new filtering algorithm is proposed with sixth-order zero-lag Butterworth low-pass filters, which considers the participant’s cadence based on spectral analysis. For vertical and mediolateral directions, a cutoff frequency of 6.25*cadence, while for the anterior-posterior directions, a cutoff frequency of 5*cadence is recommended. This method adapts the widely used filtering method and significantly reduces signal distortion while still providing adequate noise attenuation for displacement data. This would enable researchers to evaluate motion data in more detail.