Sara Abbaspour - Biomedical Engineering Faculty, Amirkabir University of Technology, Tehran
Ali Fallah
Ali Maleki - Electrical and Computer Engineering Faculty, Semnan University, Semnan

Electromyogram (EMG) is used in different applications such as diagnosis and treatment of diseases. Recorded EMG signals from upper trunk muscles are contaminated byelectrocardiogram (ECG). Removal of the ECG artifact from surface EMG is not simple because their frequency content is muchoverlap. In this paper, we compare the results of adaptive Neuro fuzzy inference system (ANFIS) and real time filtering techniques. Finally performance of these methods is evaluated usingqualitative criteria, power spectrum density and coherence and Quantitative criteria signal to noise ratio, relative error and crosscorrelation. The result of signal to noise ratio, relative error and cross correlation for better results (ANN) is equal to 13.274, 0.03and %97respectively.

ECG artifact, surface EMG, noise removal, ANFIS, real time filtering