Classification of surface electromyographic signals for control of upper limb virtual prosthesis using time-domain features

Abstract

The development of a training system in the field of rehabilitation has always been a challenge for scientists. Surface electromyographical signals are widely used as input signals for upper limb prosthetic devices. The great mental effort of patients fitted with myoelectric prostheses during the training stage, can be reduced by using a simulator of such device. This paper presents an architecture of a system able to assist the patient and a classification technique of surface electromyographical signals, based on neural networks. Four movements of the upper limb have been classified and a rate of recognition of 96.67% was obtained when a reduced number of features were used as inputs for a feed-forward neural network with two hidden layers.

Metodologia Para Classificação De Movimentos Da Mão Baseada Em Padrões Eletromiográficos

“…O controle de próteses mioelétricas baseia-se na utilização de sinais eletromiográficos (EMG) coletados, principalmente, na superfície da pele e, em geral sobre a musculatura remanescente envolvida, sendo tal técnica amplamente utilizada em próteses de membro superior [1]. Sinais EMG consistem em potenciais elétricos produzidos pela contração de determinado músculo ou grupo muscular e podem ser captados em músculos do antebraço relacionados aos movimentos da mão [2].…”

“…A literatura não apresenta um consenso acerca das melhores características a serem extraídas dos sinais EMG, porém é possível destacar algumas características do domínio do tempo que são amplamente utilizadas: Mean Absolute Value (MAV) [1][2][3][4][5], Zero Crossing (ZC) [1,2,[4][5][6] , Slope Sign Change (SSC) [1,2,[4][5][6], Waveform Length (WL) [1][2][3][4][5][6][7], Willison Amplitude (WA) [3][4][5][6], Variance (VAR) [3,4,7] e Root Mean Square (RMS) [4,6,7].…”

Proposalofasystemforreal-Timeemgacquiring,plotting Anddetectionofcontractions

“…The control of myoelectric prostheses is based on the use of electromyographic (EMG) signals collected mainly on the surface of the skin and, generally, on the remaining musculature, being such a technique widely used in prosthesis control, as in the prostheses of the upper limb [1]. The EMG signals are formed by physiological variations in the state of muscle fiber membranes, that is, they consist of electrical potentials produced by the contraction of a given muscle or muscle group [2,3].…”

“…Such a technique is widely used, for example, associated with Empirical Mode Decomposition for EMG signal filtering and onset detection [5,6]. In the domain of the control of prostheses and applications in HMI, it is of fundamental importance a system that works in real time [1,3]. In this context, onset detection is one of the processes to be executed, as well as feature extraction, pattern recognition and movement classification [1,3].…”

“…In the domain of the control of prostheses and applications in HMI, it is of fundamental importance a system that works in real time [1,3]. In this context, onset detection is one of the processes to be executed, as well as feature extraction, pattern recognition and movement classification [1,3]. For control of prosthetic limbs, the response time, to reach real-time constraints, must be less than 300 ms [3].…”