Estimating the Efficiency of Adaptive Noise Filtering by the Least Mean Squares Method in Relation to Phonocardiographic Signals

The interest in phonocardiography (PCG) diagnostic methods is associated with high information content, the nonperturbing nature of the study, the absence of side effects and relatively low cost. Due to these features, PCG methods are attractive for long-term monitoring of heart activity, including in telemedicine. Modern digital signal processing technologies make it possible to simplify the requirements for the conditions of the diagnostic procedure. The purpose of this work is to evaluate the effectiveness and determine the quasi-optimal parameters of an adaptive noise filter based on the least mean squares algorithm. Based on numerical simulation of an adaptive filter using PCG signals from the PhysioNet database, typical values of the signal-to-noise ratio at the filter output, the filter time constant and filter parameters for different kinds of noises were determined. The presented simulation results allow us to determine the possibilities of adaptive noise filtering by the least mean squares method for PCG signals and formulate requirements for the conditions of telemedicine diagnostics.