Use of Sample Entropy Extracted from Intramuscular EMG Signals for the Estimation of Force

Abstract

This study investigates the use of sample entropy as a feature extracted from intramuscular electromyography (EMG) for the estimation of force. Grasping force and intra-muscular EMG) signals were measured in 10 able-bodied subjects. Constraint sample entropy (CSE) was extracted from the EMG signal (window size of 200 ms). The association be-tween the CSE and force was modeled using an artificial neural network. The accuracy of estimation was on average R2 = 0.89 ± 0.05 and root mean square difference (RMSD) = 6.67 ± 2.17 N). It was concluded that sample entropy does capture the dynamics in the intramuscular EMG, and that a single channel of intramuscular EMG can be used for muscle force estimation. The information of muscle force is necessary in proportional myoelectric control.