AUTHOR=Sun Wentao , Zhu Jinying , Jiang Yinlai , Yokoi Hiroshi , Huang Qiang 

TITLE=One-Channel Surface Electromyography Decomposition for Muscle Force Estimation

JOURNAL=Frontiers in Neurorobotics

VOLUME=Volume 12 - 2018

YEAR=2018

URL=https://www.frontiersin.org/journals/neurorobotics/articles/10.3389/fnbot.2018.00020

DOI=10.3389/fnbot.2018.00020

ISSN=1662-5218

ABSTRACT=Estimating muscle force by surface electromyography(sEMG) is a noninvasive and flexible way of diagnosing biomechanical diseases and controlling assistive devices, e.g. prosthetic hands. To estimate muscle force by sEMG, supervised method is usually adopted, which requires the simultaneous recording of sEMG and muscle force measured by extra devices to tune the variables. However, recording muscle force of the lost limbs of amputees is challenging, which limits the application of the supervised method. Though unsupervised method does not need to record muscle force, it suffers from low accuracy due to the lack of recorded force as reference. To accurately and easily estimate muscle force by unsupervised methods, we proposed to decompose one-channel sEMG to the constituent motor unit action potentials by two steps: (1) learning an orthogonal basis of the  sEMG using reconstruction independent component analysis; (2) extracting spike-like MUAPs from the basis vectors. Nine healthy subjects were recruited to evaluate the accuracy of the proposed approach on estimating muscle force of the biceps brachii. The result demonstrated that more than 80\% variability of the force can be explained by the decomposed MUAPs at arbitrary force levels. This result is superior to the conventional approach based on amplitude of sEMG which only explains 62.3\% variability of the muscle force on the recorded data. Moreover, we utilized the proposed approach to control grip force of a prosthetic hand, one of the most important clinical applications of the unsupervised method. Experiment results on two trans-radial amputees indicated that the proposed approach can improve the performance of the prosthetic hand in grasping daily-life objects.