Related Experiment Video
Updated: Jul 16, 2026
11:25
Simultaneous Scalp Electroencephalography (EEG), Electromyography (EMG), and Whole-body Segmental Inertial Recording for Multi-modal Neural Decoding
Published on: July 26, 2013
MoCap-Referenced Neck-Shoulder sEMG-IMU Decoding for Discrete Assistive Commands: A Pilot Study
Ameer H Majeed1, Farah Masood1, Hussein A Abdullah2
1Biomedical Engineering Department, Al-Khwarizmi College of Engineering, University of Baghdad, Baghdad 17635, Iraq.
Sensors (Basel, Switzerland)
|July 15, 2026
Summary
This study introduces a new neck and shoulder surface electromyography (sEMG) system for hands-free control. The novel framework shows high accuracy in decoding user commands, offering a promising alternative for assistive technology.
Area of Science:
- Biomedical Engineering
- Rehabilitation Technology
- Human-Computer Interaction
Background:
- Hands-free command interfaces are crucial for individuals with motor impairments.
- Neck-shoulder surface electromyography (sEMG) offers a viable alternative control method.
- Previous validation methods may overestimate accuracy; false-trigger behavior needs quantification.
Purpose of the Study:
- To present a motion-capture (MoCap)-referenced decoding framework using sEMG and inertial measurement units (IMUs).
- To evaluate the accuracy and generalization of this framework for decoding seven to eight distinct user commands.
- To quantify false-trigger rates during a REST state and assess trade-offs.
Main Methods:
- Utilized four bilateral upper trapezius (UT) and sternocleidomastoid (SCM) sEMG channels with integrated IMUs.
- Employed optical MoCap for kinematic reference and quality control.
- Classified features using LDA, kNN, and SVM, evaluated with within-subject, trial-wise, and leave-one-subject-out (LOSO) testing.
Main Results:
- Achieved high within-subject accuracy (96.02% - 96.35%) and pooled trial-wise accuracy (90.5% - 92.1%).
- LOSO accuracy decreased to 60.4% - 63.8%, indicating generalization challenges.
- REST false activation rate (FAR) ranged from approximately 9.8% to 25.6%.
Conclusions:
- Demonstrated controlled offline pilot feasibility of the MoCap-referenced sEMG decoding framework.
- Quantified key generalization trade-offs and REST false-activation rates.
- Provides a foundation for future validation in diverse clinical populations.
