Dezert-Smarandache Theory based Classification of EMG Signals

Abstract: This research paper proposes an intellectual method for the classification of different types of Electromyography (EMG) signals like normal, myopathy and neuropathy signals. Inside the human body, contraction of muscles and nerves occur at every second. And, EMG is a techniqueused to measure this electrical activity. For the analysis of EMG signals, so many methods have been already used.With this research, a new method is proposed in which Dezert-Smarandache Theory (DSmT) based classification technique is utilizedfor the EMG signals analysis. In this, discrete wavelet transform with some features likeenergy, mean and standard deviation are exploited for the features extraction of the EMG signals. After that, classifiers are used in the analysis for the modelling purpose. Then, using these classifiers, DSmT based technique helps in improving the accuracy of the results.It can be seen in the results that DSmT based classification gives the best accuracy (approximately 97%) in comparison to the other classifiers used during this research.

Keywords: DSmT, Electromyography, Wavelet Transform, SVM, SVM-kNN.