Mean and Median Frequency of EMG Signal to Determine Muscle Force based on Time-Dependent Power Spectrum

Abstract

The analysis of EMG signals can be generally divided into three main issues, i.e., muscle force, muscle geometry and muscle fatigue. Recently, there are no universal indices that can be applied for all issues. In this paper, we modify the global fatigue indices, namely mean frequency (MNF) and median frequency (MDF), to be used as a muscle force and fatigue index. Due to a drawback of MNF and MDF that it has a non-linear relationship between muscle force and feature value, especially in large muscles and in cyclic dynamic contractions. A time-dependence of MNF and MDF (TD-MNF and TD-MDF) is computed for dynamic contractions. Subsequently, a slope of the regression line that fits maximum values of MDF (and MNF) during a number of cyclic contractions is used as a fatigue index. To be additionally used as a muscle force index, some suitable ranges of TD-MNF and TD-MDF should be selected and five effective statistical parameters including mean, median, variance, root mean square and kurtosis, were applied to the selected range. From the experiments, the performance of TD-MNF is definitely better than that of TD-MDF. The results showed that mean and median features of the selected TD-MNF series have a better linear relationship with muscle force (load level) compared to the traditional methods and have a significant difference (p<0.001) between feature values for different loading conditions. Moreover, there is a certain pattern of TD-MNF for all trials and subjects that has not been found in traditional MNF. TD-MNF is optimized when an overlapping consecutive window method is performed with a 384-sample window-size and a 192-sample window-increment. In total, mean and median features of the selected TD-MNF series can be used as a muscle force and fatigue index. In future works, the proposed method can be used instead of using multiple features for the EMG signal analysis.