Entropy analysis of surface EMG for classification of face movements

Measuring complexity of dynamical systems is a mighty tool for electrophysiological signal processing. There are plenty of entropies for estimating complexity measure. Approximate entropy (ApEn), sample entropy (SampEn), fuzzy entropy (FuzzyEn), wavelet entropy (WE) and wavelet packet entropy (WPE) was used for surface EMG feature extraction for face movements classification. Linear discriminant analysis (LDA) selected for classification. Classification performance was determined by mean square error (MSE) for different window sizes. Fuzzy entropy is the most robust and succeeding method of them. Principal component analysis used to improve classification performance however just results of approximate entropy feature were refined. MSE of wavelet entropy and wavelet packet entropy are also decent methods for this classification problem.