Alberto Malliani, Stefano
Guzzetti.
Centro Ricerche Cardiovascolari, CNR, L.I.T.A. di Vialba, Medicina Interna II,
Ospedale "L. Sacco", Universita degli Studi, Milan, Italy
|
|
Numerous exhaustive accounts have been recently
published on power spectrum analysis of HRV1,6,7. In
this context suffice it to say that in our studies we have mainly used autoregressive
algorithms that indicate the existence in both the heart period and arterial pressure
variabilities of three spectral components, respectively centered around 0. Hz (the very
low frequency component, VLF), 0.1 Hz (the LF component) and 0.20-0.35 Hz (the HF
component). Usually only the LF and HF components can be adequately studied in time series
of 200-500 events. With this approach each spectral component is assessed in terms of its
center frequency and of its absolute power (i.e. the area), expressed in squared units
(e.g., msec2).
The normalization procedure
In view of the physiological model1 of a reciprocal
relation between LF and HF components, their estimation independent of total power has
proven to be crucial. This can be easily obtained by computing their values in normalized
units (nu), which are obtained by dividing the power of a given component by the total
power (from which VLF has been subtracted) and multiplying by 100. Usually the sum of LF
and HF falls short of 100 because of the frequent presence of some small noise components.
As an example of the importance of the normalization procedure for the interpretation of
data, one should consider that the sympathetic excitation that leads to tachycardia is
often accompanied by a reduction in total power of HRV: this can cause a paradoxical
reduction of LF in absolute units, while this component appears clearly increased if
expressed in nu.
|