Comparison of EMG activity using Spectrum Indices from Biceps Femoris Muscle during Treadmill Walking

Abstract

Under dynamic exercise, the muscle tension is proportional to EMG activity[1], and EMG indices which represent muscle activity can be classified by magnitude indices in time domain and spectrum indices in frequency domain[2-4]. The purpose of this paper is to compare the activities of EMG indices during dynamic exercise, and to examine the spectrum index using spectrum moments which is better and more sensitive than traditional spectrum indices. Ten healthy subjects walked on the treadmill at 4.5 km/h during 30 seconds, and EMG signals are measured from biceps femoris muscle. EMG indices are extracted such as magnitude index including root mean square (RMS), and spectrum indices including mean power frequency (MPF), high-to-low ratio (HLR) and HLR using spectrum moments (SM-HLR). Experimental results shows that the correlation coefficients are not high between RMS and the other spectrum indices but very high between spectrum indices above 0.89, and the variation coefficient of logarithmic SM-HLR is above 3 and 2 times higher than those of logarithmic MPF and HLR, respectively. Therefore, SM-HLR can be a high-sensitive spectrum index to reflect changes of activities of biceps femoris muscle due to walking.