A STUDY ON THE EFFECTS OF ELECTRODE PLACEMENT ON THE PERFORMANCE OF EMG CLASSIFICATION OF HAND AND WRIST MOVEMENTS

E. F. Schmiele, J. N. Mello, E. Lawrence, A. B. Soares
2018 Anais do V Congresso Brasileiro de Eletromiografia e Cinesiologia e X Simpósio de Engenharia Biomédica   unpublished
Surface Electromyographic signals have been used to control robotic prosthesis by means of carefully designed patter recognition algorithms. There are several challenges yet to be solved considering this field, such as the classifiers' dependency on the correct positioning of the electrodes, since even small displacements of those could cause great differences on the sampled signal. This is highly pertinent considering the difficulty of repeating the same position of an electrode at every
more » ... rode at every fitting of prosthesis socket. In this paper, we investigate those effects in the performance of a well-established classifier and possible strategies to train it to account for drastic changes in the position of the EMG electrodes. For that, three Support Vector Machine (SVM) classifiers were trained with data collected from the upper-forearm, while volunteers performed four hand and wrist movements with a set of four active EMG electrodes positioned around the arm evenly spaced. Each SMV was trained with data collected with the set of four electrodes placed at three different positions around the arm. Among the tested electrode positions, the performance of the SMVs were statistically similar among all three positions for three of the tested movements and among two of the positions for the remaining one.
doi:10.29327/cobecseb.78836 fatcat:y3o2kk2hy5a3fhhbfiwpufd3gq