Real-Time Gait Events Detection Using IMU Combined With TCNs Model in Transtibial Amputees
IEEE Access · 2026 · 14 · 113223-113234
DOI: 10.1109/access.2026.3715835 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Harit Apidech und 6 weiteren Autor:innen. Die Arbeit nennt 35 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Gait event detection is a fundamental step in gait analysis. It identifies initial contact (IC) and final contact (FC), which are used to segment the stance and swing phases and to extract temporal gait parameters for clinical assessment. This step is especially important for individuals with lower-limb amputation, whose gait is often evaluated to optimize prosthetic alignment. This study proposes an inertial measurement unit (IMU)-driven machine learning framework for gait event detection in transtibial amputees, focusing on the systematic evaluation of sensor configuration and placement. Twelve individuals with unilateral transtibial amputation walked on a treadmill under four speed conditions (comfortable, slow, fast, and self-paced) while wearing 16 IMUs placed at various body segments. The IMU signals (acceleration, angular velocity, and orientation) were used as input features to train a temporal convolutional network for foot-contact classification. The model achieved high classification performance on the amputated side, with F1-scores of 0.91-0.97. Event-level evaluation showed low timing errors, with mean absolute errors of 0.012-0.020 s for IC and 0.011-0.034 s for FC. The temporal gait parameters agreed closely with the reference system, with mean differences near 0.000 s. Optimal performance was achieved with 2- to 4-IMU configurations, particularly at the shank and foot locations. These findings demonstrate the potential of optimized IMU configurations for accurate, efficient gait event detection in transtibial amputees and support the feasibility of the framework for real-world use. Integration with wearable IMU systems could further enable remote rehabilitation, allowing clinicians to monitor progress and adjust interventions without frequent laboratory visits.
Zitieren
APA
Apidech, H., Kusakunniran, W., Samala, M., Jamkrajang, P., Manupibul, U., Niamsang, W., & Limroongreungrat, W. (2026). Real-Time Gait Events Detection Using IMU Combined With TCNs Model in Transtibial Amputees. IEEE Access, 14, 113223-113234. https://doi.org/10.1109/access.2026.3715835
Vancouver
Apidech H, Kusakunniran W, Samala M, Jamkrajang P, Manupibul U, Niamsang W, et al. Real-Time Gait Events Detection Using IMU Combined With TCNs Model in Transtibial Amputees. IEEE Access. 2026;14:113223-113234. doi:10.1109/access.2026.3715835
BibTeX
@article{Apidech2026,
author = {Apidech, Harit and Kusakunniran, Worapan and Samala, Manunchaya and Jamkrajang, Parunchaya and Manupibul, Udomporn and Niamsang, Wisavaporn and Limroongreungrat, Weerawat},
title = {Real-Time Gait Events Detection Using IMU Combined With TCNs Model in Transtibial Amputees},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {113223--113234},
doi = {10.1109/access.2026.3715835}
}
EUVIMED