Detection of Muscle Synergy by Wrist Electromyogram

Humans have a large number of muscles, and independently control each one. It is believed that enormous amount of computation is required to operate muscles. Therefore, it is estimated that complex human muscle movements can be represented by a small number of functional units called muscle synergies. If the existence of muscle synergy can be explained, it can be utilized effectively in various fields such as rehabilitation and robotics. Therefore, in this paper, we try to explain the existence of muscle synergy. What we did in this paper was to measure seven simple movements of bending and extending the thumb, index finger, and the middle, ring and little fingers simultaneously, neutral movements and five complex movements like “rock” and “scissors” in rock-paper-scissors that can be made by combining these simple movements, using electromyogram at the wrist. Then, we considered the measured simple movements as muscle synergies and implemented them using a neural network so that complex movements can be explained by combinations of simple movements. The results are based on two axes: the first is recognition of the complex movements and the second is the influence rate of the simple movements on the complex movements by looking at the output layer on the simple movements when the complex movements are recognized. The accuracy of complex movements is 92.0%. We believe that as the second result a large value of simple movements comprising complex movements is observed, which can explain to some extent the presence of muscle synergy.