Nonlinear Dynamics Of The Emg Analysis

As the use of EMG is very convenient and fast, it is now becoming increasingly a powerful measure to get information and to diagnose about the muscular and nervous systems. But up to now, most methods of EMG are still based on traditional linear and statistical analysis. Only a few people dealed with nonlinear principle and method. So far as our opinion, most physiological processes are nonlinear, most probably so does EMG. Therefore we think in the study of EMG, it is important to know whether EMG is nonlinear in its character or not. In the past two decades, some people examined it, but the methods they used were all too simple to draw convincing conclusion. In this paper, we use a fairly comprehensive nonlinear dynamical analysis of four EMGs data of an adult woman. At first, construct ten sets of surrogate data for each EMG data, then calculate the correlation dimension (Dcorr), correlation time, maximum Lyapunov exponent (A ,), L-Z complexity and approximate entropy (ApEn) of both the original EMG data and surrogate data, and then compare them. We find that all the results are quite different between each original EMG and its ten surrogate data, i.e. EMG is not a linear random noise, but a nonlinear deterministic signal (though it does not like a low dimensional chaos). EMG is also assessed with recurrence plot analysis (RPA), recurrence quantity analysis (RQA), iterated function system (IPS) dumpiness test, singular-value decomposition (SVD), and False Nearest neighbors (FFN). Synthesizing all of our results and analysis, we conclude that EMG obeys nonlinear deterministic law. It is probably high dimensional chaos with DCOTT of the attractor greater than 4 and smaller than 5.