Analysis of Muscle Fatigue Conditions in Surface EMG Signal with A Novel Hilbert Marginal Spectrum Entropy Method

Surface Electromyogram is the superposition of action potentials generated during muscle contraction that provides useful indices for biomechanics. Such signals are nonstationary and the measure of its time varying components are essentially needed to identify the progression of fatigue. In this work, an attempt has been made to identify the muscle fatigue condition using Hilbert-Huang Marginal Spectrum. The novelty of the proposed framework is that the Marginal Spectrum is computed by ordering the instantaneous frequency of an Intrinsic Mode Function obtained for a signal to be analyzed. For this purpose, signals are recorded from the biceps brachii muscles of 50 healthy volunteers using isometric and dynamic contraction exercise protocols. Initially, each signal is equally partitioned into 10 segments, where the 1st, 5th and 10th segments are analyzed. Further, metrics estimated from marginal spectrum such as area under the curve, skewness and kurtosis are calculated for the considered three segments. The preliminary results show that the estimated metrics are able to distinguish the fatiguing conditions. The recorded signals show low amplitude and high frequencies in nonfatigue region and vice-versa. The obtained results are statistically significant with p < 0.005. It appears that the instantaneous frequency based marginal spectrum estimator is able to measure the fatigue index. Therefore, the proposed method can be useful in analyzing fatigue condition of skeletal muscles.