On the accuracy of methods identifying gait events using optical motion capture and a single inertial measurement unit on the sacrum

Abstract

Abstract Gait events (i.e., heel strikes and toe-offs) are essential for extracting spatiotemporal parameters and segmenting biological signals. While force platforms and optical motion capture (OMC) are ideal for identifying such events, inertial measurement units (IMUs) are cheaper and more applicable, especially for applications outside traditional lab settings. This study aimed to compare IMU- and OMC-based gait event detection to force plates. Seventeen adults walked on a walkway, stepping on two force plates while an IMU on the sacrum and retro-reflective markers on the calcaneus and 5th metatarsus captured foot kinematics. Gait events were identified using two OMC and two IMU methods (OMC1, OMC2, IMU1, IMU2). OMC1 detects gait events using vertical marker velocity shifts, OMC2 uses sagittal velocity thresholds, IMU1 applies wavelet-based differentiation and IMU2 identifies peaks in acceleration signals. Heel strikes and toe-offs were compared to force plate data, assessing root-mean-square error (RMSE), and intra-subject coefficient of variation (CoV). For heel strike events, OMC1 presented the lowest RMSE (∼14.3ms), significantly differing from IMU1 (RMSE: 50.6ms; p<0,001) and IMU2 (RMSE: 61.1ms; p<0,001). For toe-offs, OMC1 presented the lowest RMSE (∼17.3ms), differing from IMU1 (54.7ms; p<0,001) and IMU2 (74.8ms; p<0,001). IMU2 also showed the highest CoV (∼23.9ms) differing from OMC1 (∼7.1ms; p<0,001) and IMU1 (9.2ms; p<0,001). Lower accuracy and greater variability in IMU methods may stem from the approach used to detect gait events. Thus, OMC methods more accurately detect gait events than sacrum-mounted IMUs. While IMUs offer an alternative, researchers should be cautious of their accuracy and variability limitations.