Normal Cumulative Distribution Function and Dispersion Entropy Based EMG Classification

Abstract

Electromyography (EMG) is used to measure muscle activity. EMG signals are widely used in many biomedical practices such as motion recognition, prosthetic control, physical rehabilitation, and human-computer interfaces. The effective use of EMG in such practices depends on distinctive feature extraction. In this study, Dispersion Entropy (DisEn) and Normal Cumulative Distribution Function (NCDF) methods are used for feature extraction from EMG signals. The suggested method was tested with a data set containing immersion of six different objects. In the experimental studies, the proposed method distinguished the movements with an accuracy performance of 98%. When compared to other methods using the same data set, the suggested method has about 1.2% better performance.