Spectral analysis of normal resting ECGs demonstrates that waveform amplitude information extends up to 200 Hz, while duration information lies below 60 Hz.
Electrocardiographic (ECG) signals were recorded from ten normal resting male subjects using two bipolar lead configurations: sternal and axillary. The recordings were made under low-noise conditions with a recording system bandpass extending from dc to 10 kHz. The ECG data were digitized and spectrally analyzed using a fast Fourier transform (FFT). The spectral analysis included both single ECG cycles and averaged ECG cycles. Slight differences were found between the averaged sternal and axillary ECG spectra in the low-frequency highamplitude range. An overall mean spectrum was produced as a bound for the normal resting male ECG spectrum as encountered in this study. The maximum value of this spectrum occurred at the fundamental (1.22 Hz); it had a -20-dB point at 36 Hz, -40-dB at 105 Hz, and -60-dB at 300 Hz. To investigate the effects of band-limiting on the ECG, an ideal digital filter was implemented using the FFT. It was found that waveform notching is caused by frequencies over a wide passband, waveform amplitude information extends only to 200 Hz, and waveform duration information lies below 60 Hz.
Golden et al. (Sat,) studied this question.