Utilização de wavelets no processamento de sinais EMG

Abstract

This study proposes an approach to analyze EMG signals using wavelets transformed as a method of signal features extraction. The adopted methodology is based on the study of the aggregated power envelope and the aggregate power spectrum envelope, which are obtained from the distribution of energy of a certain signal, based on the potency of wavelet coefficients, showed like wavelets spectrograms or from a wavelet scalegram. EMG signals were captured in the surface of the human skin and came from the right leg rectus femoris muscle in a static condition (isometric), also from the flexor muscle form the right hand in dynamic contraction (isotonic) and also form a train of motor unit action potential (MUAP) form the First Dorsal Interosseous muscle during dynamic contraction. Having those signals, there were taken two research phases: extraction of the feature based on the analytical wavelet transformed (AWT) in muscles during contraction (isometric and isotonic) and the phase of detection of MUAPs. In the AWT phase, considering the calculation of the envelopes in the timefrequency chart (spectrogram), the results shoed that the wavelet transformed can be applied for extraction of spectral content of the signal and also showed the possibility of verifying the potency signal spectrum and the energy of such signal intimae. Those variables were according to the expected features for EMG signal, reported by literature. In the second phase, MUAP detection, it was used the calculation of the envelopes based on the scalegram, having as a main wavelet the Daubechies of 4 (db4), Coiflet of 4 (coif4) and Symlet of 5 (sym5) . The result showed that the method allowed to locate in time of MUAPs and showed that it is sensible enough to detect signals form motor units, far from the sensor, which contribute to formation of the EMG signal. The use of the wavelet Db4 showed to be better to detect the muscle activity on the beginning of it ( set-on ), because the Db4 is similar to a MUAP. This work proposes that future studies can be based on the research of families of wavelets, using of the method of the aggregated power envelope to control proteases for arms, or hands for example. It is also proposed studies for detection of MUAPs as an important tool for muscles evaluation, in diagnosis of miopathologies and neuro-muscle disjunctions, envelope features extraction process for other biomedical signals, such as EEG and ECG.