Empirical Mode Decomposition for EMG Signals Filtering

Abstract

Biomedical signals such as electromyography may present a high noise rate during the acquisition process, this may be due to ambient noise, poor positioning of electrodes, equipment calibration, among others. These noises can affect the results obtained, disturbing the diagnosis of diseases, recognition of movement, recognition of gestures and humancomputer interaction. This article proposes a technique to reduce these artifacts using the Empirical Mode Decomposition (EMD) method. EMG signals were collected from 5 healthy individuals, these signals were decomposed using the respective intrinsic mode functions (IMFs), then thresholds were set to remove noise, and finally the noiseless signal was reconstructed. The noise reduction effectiveness of the technique was quantitatively evaluated using the signal-to-noise ratio (SNR). The mean SNR was 12.427 dB, which means a good signal-tonoise ratio when compared to studies that used other filtering methods. In this sense, the proposed method can later be used in the areas of disease diagnosis, pattern recognition and movement classification.