The RR interval spectrum, the ECG signal and aliasing

Abstract

A reliable spectral analysis requires sampling rate at least twice as large as the frequency bound, otherwise the analysis will be unreliable and plagued with aliasing distortions. The RR samplings do not satisfy the above requirements and therefore their spectral analysis might be unreliable. In order to demonstrate the feasibility of aliasing in RR spectral analysis, we have done an experiment which have shown clearly how the aliasing was developed. In the experiments, one of us (A.G) had kept his high breathing rate constant with the aid of metronome for more than 5 minutes. The breathing rate was larger than one-half the heart rate. Very accurate results were obtained and the resulting aliasing well understood. To our best knowledge this is the first controlled experiment of this kind coducted on humans. We compared the RR spectral analysis with the spectrum of the ECG signals from which the RR intervals were extracted. In the significant for RR analysis frequencies (below one-half Hertz) significant differences were observed. In conclusion we recommend to study the spectral analysis of the ECG signal in the free of aliasing frequency range.