Inertial measurement unit-based motion capture to replace camera-based systems for assessing gait in healthy young adults: Proceed with caution

Abstract

Gait analysis can identify injury-risk markers indiscernible to the naked eye. Inertial measurement unit (IMU)-based motion capture circumvents optokinetic motion capture (OMC) clinical implementation barriers with its portability, increased affordability, and decreased computational burden. We compared an IMU system to a robust OMC marker set for gait analysis. 10 healthy adults walked at self-selected speeds equipped with Noraxon MyoMotion IMUs and a 24-marker, 5-cluster marker-set in view of 14 OMC cameras. A single calibration was applied. IMU system and OMC calculated joint angles were compared. A single calibration performed similarly to previously reported repeated calibration. IMU and OMC agreement was best in the sagittal plane with IMU axis-mixing affecting off-sagittal plane agreement. System differences were greater than 5° for most motions. Measurement system bias showed at the ankle and knee, however differences varied across participants. IMU kinematics should be interpreted with caution; consistency and accuracy must improve before IMUs can replace OMC.

Keywords: Accuracy; Gait analysis; Inertial measurement unit (IMU); Motion capture; Wearable sensors.

Fig. 1.

Participant equipped with optokinetic markers and IMUs.

Fig. 2.

Participant summary curves for IMU (dashed) kinematic outputs compared with the gold standard OMC (solid) vs % gait cycle. A typical participant with good agreement between the two systems is provided. Curves represent average kinematics across 15 gait cycles and shaded areas are ± one standard deviation.

Fig. 3.

Participant summary curves for IMU (dashed) kinematic outputs compared with the gold standard OMC (solid) vs % gait cycle. A typical participant with poor agreement between the two systems is provided. Curves represent average kinematics across 15 gait cycles and shaded areas are ± one standard deviation.

Fig. 4.

Group summary kinematics for ankle dorsiflexion (A), abduction (B), and external rotation (C), hip flexion (D), abduction (E), and external rotation (F), and knee flexion (G) for IMU (dashed) and OMC (solid) with ± 1 standard deviation in the shaded area.

Fig. 5.

System differences between IMU and OMC averaged all participants with ± 1 standard deviation in the shaded area displayed.

Fig. 6.

Bland Altman plots with average angle between the IMU and OMC systems on the x-axis and difference between the angles calculated by each system on the y-axis for each timepoint in the gait cycle, averaged across all participants for ankle dorsiflexion (A), abduction (B), and external rotation (C), hip flexion (D), abduction (E), and external rotation (F), and knee flexion (G). Agreement between the two systems was assessed by comparing the line of equality, represented with a solid horizontal line, to 0°. The measurements were considered acceptable if the differences fell within a 5° window, represented by dashed lines, around the line of equality for each motion measured.