The optimal electromyography feature for oral muscle movements

The optimal feature is necessary for the reach of electromyography (EMG) control or human-machine interface. This study aimed to find the optimal feature of time-domain features. Five oral muscles movements were performed in five volunteers. The oral activities were open mouth, protrude mouth, broad smile, straight tongue, and up tongue which were the parts of non-speech oral motor treatments. Sixteen time-domain features were selected to separate five oral muscles movements. The features were chosen using the RES index and evaluated using multilayer perception to confirm the results. The optimal features were different in each channel followed as: slope sign change feature for channel 2, myopulse percentage rate feature for channel 3 and channel 5. In addition, the average percentage of classification was 91.33%.