Higuchi fractal dimension analysis of surface EMG signals and determination of active electrode positions

Abstract

In this study, fractal dimension which used to analysis complexity of the biomedical signals used to determine active channels when performing 24 fingers and wrist movements. The results compared with root mean square (RMS) and mean absolute value (MAV) features. Higuchi fractal dimension (HFD) method was chosen that has high accuracy and linear with theoretical fractal dimension values nevertheless the method is noise sensitive. The noise sensitivity problem was overcome with filtering the signal. Mean Higuchi fractal dimension feature determined using sliding window and active movements obtained that are above of the chosen thresholds for each channels. Correlation of active channels for each movements which was obtained with RMS and HFD features was discussed. Thus a new method was submitted which based on HFD instead of RMS and MAV features.