On the selection of FIR band pass filters for low frequency EEG synchronization applied to A-phase analysis during sleep

Abstract

One way to estimate the instantaneous amplitude and phase of a real-valued signal consists in passing the signal through a bank of quadrature filters. However, a badly designed filter may introduce distortions in the estimations. The objetive of this work is to compare the output phase and amplitude envelope of three bandpass FIR quadrature filters. On the first stage, this study uses as input a synthetic signal whose phase is known, in order to perform a quantitative assessment of the performance of each filter. The bandwidth of the filters is varied in order to see its effect on the output signal. On the second stage, a study of changes in dynamic brain connectivity was made using electroencephalographic data for healthy patients during sleep. Results suggest that the type of filter and its bandwidth must be chosen carefully in order to avoid distortions that may bias the results of any further analysis.