The random effect of electrical noise on T wave alternans

This study was designed to examine the effect of random noise on the nonparametric power spectrum estimation of beat-to-beat alternations in the electrocardiogram (ECG). It has been widely assumed that random noise is additive to the power spectrum contribution of a beat-to-beat alternating signal. In this study, alternans of the T wave was used to simulate a subtle beat-to-beat alternating signal present in the ECG. Pseudorandom noise was then superimposed on the alternans and a nonparametric power spectrum estimation technique, the Welch method, was used to quantify the amount of T wave alternans present. The study showed that in 34 of 100 pseudorandom noise simulations, the degree of alternans of the T wave detected was decreased when compared to the case where only alternans of the T wave was present without added noise. Thus, assuming that the random noise power spectrum component (at 0.5 cpb) is always additive to the T wave alternans power spectrum component may lend to significant errors in interpretation of the results of power spectrum estimation when applied to the analysis of subtle beat-to-beat fluctuations in the electrocardiogram.<>