Individual finger classification from surface EMG: Influence of electrode set.

Abstract

The aim of this work was to minimize the number of channels, determining acceptable electrode locations and optimizing electrode-recording configurations to decode isometric flexion and extension of individual fingers. Nine healthy subjects performed cyclical isometric contractions activating individual fingers. During the experiment they tracked a moving visual marker indicating the contraction type (flexion/extension), desired activation level and the finger that should be employed. Surface electromyography (sEMG) signals were detected from the forearm muscles using a matrix of 192 channels (24 longitudinal columns and 8 transversal rows, 10 mm inter-electrode distance). The classification was evaluated in the context of a linear discriminant analysis (LDA) with different sets of EMG electrodes: A) one linear array of 8 electrodes, B) two arrays of 8 electrodes each, C) a set with one electrode on the barycenter of each sEMG activity area, D) all the recorded channels. The results showed that the classification accuracy depended on the electrode set (F=14.67, p<;0.001). The best reduction approaches were the barycenter calculation and the use of two linear arrays of electrodes, which performed similarly to each other (both > 82% of average success rate). Considering the computation time and electrode positioning, it is concluded that two arrays of 8 electrodes provide an optimal configuration to classify the isometric flexion and extension of individual fingers.