Knee deep in Error: Unraveling the accuracy limitations of consumer-grade wearable motion sensors with limited degrees of freedom computing knee joint angles

Abstract

Consumer-grade wearable inertial measurement units (IMUs) with limited degrees of freedom (DOF) are common computing knee motion before/after clinical intervention. Appropriate use requires attaching sensors per manufacturer's instructions. However, IMU misplacement error remains problematic. Thus, we assessed error computing sagittal knee angles across sensor misplacement orientations of one consumer-grade IMU system with 6DOF. We hypothesized error would increase during frontal plane misplacement. Ten young, healthy individuals (20.9 ± 0.7 years) were enrolled. IMUs were adhered/calibrated per manufacturer instructions and IMU-computed knee angles were compared to optical motion capture (MOCAP) during self-selected speed treadmill walking. Shin/thigh IMUs were then moved across sagittal, frontal, and transverse planes (±10°, ±20°). At each orientation, IMU calibration and treadmill walking was re-completed. Error was computed between MOCAP/IMU by calculating the difference and computing the absolute value (i.e. absolute value error). This was averaged across participants at each sensor position (i.e. average absolute value error). Bland Altman analyses assessed MOCAP/IMU agreement. ANOVAs assessed how plane, segment, and falsified IMU angle impacted error. MOCAP/IMUs were highly correlated with IMUs achieving clinically acceptable error (<5°) when placed correctly or with sagittal/transverse plane misalignment. Frontal plane (varus and valgus) and shin sensor misalignment caused significantly greater error (∼7°), likely due to limited IMU DOF. Limited DOF IMUs can accurately compute knee motion even with significant sagittal/transverse plane misalignment. However, frontal plane placement is critical for proper function, implying patients with frontal plane deformities should not use reduced DOF IMUs. Moreover, patient education is critical for appropriate deployment of wearable IMUs.

Keywords: Biomechanics; Error; Inertial measurement unit; Knee; Knee angles; Wearable sensor.