ABSTRACT
Automatic sensor-to-foot alignment is required in clinical gait analysis using inertial sensors to avoid assumptions about sensors initial positions and orientations. Numerous studies have proposed alignment methods. The current study aimed at describing and accessing the performance of a simple rule to automatically recognize the orientation of the sagittal plane foot angular velocity that can be used with any alignment method and any populations including individuals with severe motor disorders such as patients with cerebral palsy (CP). Fifty-five participants (15 healthy, 15 with CP and 25 with various other motor disorders) wore IMUs on both feet during one or several visits of clinical gait analysis (CGA) with optical motion capture system as reference. The foot coordinate system was determined using acceleration during motionless periods and angular velocity during walking, as previously described in the literature. Based on the foot sagittal plane angular velocity, a novel rule is introduced to determine the latest uncertainty related to mediolateral axis direction which often causes errors. It consisted of massively filtering the signal and applying a simple peak detection, omitting the double peaks with the same sign. The time between the negative and positive peaks can inform on the axis direction. This verification showed excellent results with 99,94% sensibility against the reference. This simple rule could be used to further improve existing sensor-to-segment algorithms with inertial sensors located on the feet, and thus improve pathological gait analysis.
