Improved complex demodulation method for frequency analysis

Abstract

AbstractSleep spindle brain waves have an important role in the sleeping mechanism. Recently, computer‐aided automatic analysis methods have been introduced in the form of numerical analysis of frequency, duration, etc., and their rapid progress has been noted. However, the spindle waves are transient nonstationary “phasic events,” and therefore their accurate frequency analysis has caused serious difficulties.Aiming at the accurate frequency analysis of the spindle waves, the authors have already proposed an analysis algorithm based on complex demodulation (CD) in the time domain, but the long processing time required for the analysis has remained a problem.This paper proposes a new analysis algorithm which combines the advantages of frequency domain CD and Hilbert transformation. This method applies Hilbert transformation mainly to speed up the measurement of instantaneous amplitude and, in the case of measuring the instantaneous frequency, the advantage of the CD method using interpolation is exploited to improve the accuracy. Here, the proposed method is explained concretely and the analysis results are investigated in comparison with the time domain CD method published earlier.