A new method of decompose SEMG into SFAP based on RBFNN

This paper proposes a new method of decomposition of surface EMG (electromyograms) signals into their constituent single fiber action potentials (SFAPs). As the complexity of decomposition, the problem of sEMG decomposition is translated into problems of curve fitting and parameter clustering of the same SFAP. A new decomposition technique, based on genetic algorithm (GA) and radial basis function neural network (RBFNN) for curve fitting and Kohonen neural network for parameter clustering, is proposed in this paper. Compared with the method of curve fitting by using Hopfield neural network, the use of RBFNN increased the decomposition correctness, and also the learning speed. The significance of such solution is that it enables a physician a non-invasive manner for diagnostic purposes or other medical applications.