A hybrid approach to EMG pattern analysis for classification of arm movements using statistical and fuzzy techniques

In this paper, a hybrid approach is presented for discriminating a few uppe r limb movements by processing the electromyographic (EMG) signals from sel ected shoulder muscles. Statistical techniques, such as the Generalized Lik elihood Ratio test, the Principal Component Analysis, autoregressive parame tric modeling techniques and cepstral analysis techniques, combined with a fuzzy logic based classifier (the Abe-Lan network) are used to construct lo w-dimensional feature spaces with high classification rates. The experiment al results show the ability of the algorithm to correctly classify all the EMG patterns related to the selected planar arm pointing movements. Moreove r, the structure presented offers promise for real-time applications becaus e of the low computation costs of the overall algorithm. (C) 1999 IPEM. Pub lished by Elsevier Science Ltd. All rights reserved.