Analysis of electromyograms recorded using invasive and noninvasive electrodes: a study based on entropy and Lyapunov exponents estimated using artificial neural networks

Abstract

Electromyograms are electrical signals produced by the muscles. Electromyography (EMG) is an efficient technique for analyzing the normal and abnormal states of the neuromuscular system. However, many factors such as crosstalk, fatigue, type of electrodes and electrode movements, affect the quality of the acquired EMG signals. In this work, the information content of EMG signals acquired using three different electrodes namely, monopolar needle, concentric needle and high spatial resolution surface electrodes, was analyzed using Tsallis entropy and Renyi entropy. Further, Lyapunov exponents were utilized to analyze the chaotic nature of the adopted EMG signals obtained from three different electrode types. The signals were modelled using artificial neural networks (ANN) and the Lyapunov exponents were estimated using the developed ANN models. Results demonstrate that, the Tsallis entropy and Renyi entropy of EMG signals acquired using concentric needle electrodes is higher when compared to the entropy of EMG signals acquired using monopolar needle electrodes and surface electrodes. Also, the Lyapunov exponents of EMG signals measured using concentric needle electrodes is higher when compared to the EMG signals measured using monopolar needle and surface electrodes. Further, the information content of EMG signals acquired using needle electrodes is higher when compared to the surface electrodes. This work seems to be of high clinical relevance since the efficient recording of EMG signals is highly useful for development of diagnostic support tools for identification of muscle disorders.