Fractal analysis of myoelectric signals

Abstract

In this thesis, we study fractal behavior of myoelectric signals (MESs). Mathematics and definitions behind fractal behavior is presented. Different methods of estimating the fractal dimension are discussed, including time domain-based methods (i.e. Katz method, Box-Counting method) and spectrum based methods (i.e. Power Spectrum Slope Method (PSSM) and General Power Spectrum Method (GPSM)). The GPSM is introduced in context of MESs for the first time. Using simulated MESs effects of its parameters (i.e. number of active motor units (MUs), firing rate and depth of active MUs) on estimated fractal dimension (eFD) and estimated fractal indicators (eFI) are analyzed. Spectrum based methods demonstrate characteristics that suggest superiority in discerning force effects (i.e. number of active MUs and firing rate) and geometric effects (i.e. depth of active MUs). Fractal behavior of MESs during Isometric Constant Force Contractions (ICFC) at different force and joint angles are analyzed. Results of the spectrum based methods suggest that they could possibly be used to estimate the joint angle independent of force. Fractal behavior of MESs during Isometric Voluntary Contractions (IVC) at different force and joint angles are analyzed as well. Results, although highly variable, remain consistent with the simulated results.