Blind Source Separation Electrical Characteristics Extraction Method of Surface Electrical Impedance Myography Mixed Signals

Surface electrical impedance myography (sEIM) is important for evaluating muscle imbalance and muscle diseases. The mixed signals of impedance in the subcutaneous multilayer tissue captured by surface electrodes contain diverse-redundant components. Taking the mixed signals captured by sEIM as the blind signals and the muscle layer impedance as the source signal, this paper proposed a method for separating muscle layer impedance based on the impedance equivalent analysis and blind source separation, in order to improve the sensitivity of sEIM detecting changes in target muscle state. Firstly, a limb multilayer cylindrical finite element model for simulation was constructed. Secondly, a sensitivity method was employed to calculate the impedance contribution of each tissue layer for excluding redundant weak signals, which was equated to a blind source separation problem targeting the muscle layer. Finally, the separation effects of independent component analysis, principal component analysis, and equivariant adaptive separation via independence (EASI) were compared by numerical simulation and in vivo experiments to obtain the optimal solution and verify its feasibility. The results show that the correlation coefficient is above 0.98, the noise immunity is approximately 0.8, and the error cross talking (ECT) converges to 0.876 using the EASI-based method for separating muscle layer impedance. The separated muscle layer impedance in the in vivo experiments is consistent with the law of human impedance characteristics, indicating the method for separating muscle layer impedance with EASI can better enhance the sensitivity of sEIM to detect changes in the target muscle state.