CARDIOVASCULAR VARIABILITY SIGNAL PROCESSING: A CHALLENGE BETWEEN NOISE AND CHAOS

Prof. Dr. Sergio Cerutti

It is well known that heart rate variability (HRV) signal is a remarkable probe to assess the properties of cardiovascular control elicited by neural mechanisms (mainly related to autonomic nervous system, ANS), by mechanical stimuli and humoral factors as well.

Various methods have been applied in the last 30 years to enhance information from HRV signal and significant results have been obtained both for physiological purposes and for clinical applications. Which is more important, in fact, is that various parameters obtained as a "clever" post-processing of HRV signal have been suggested for diagnostic aims and are currently employed in advanced ECG interpretation. The combination of new developments of medical apparatus technology towards more compacted and powerful ECG equipment (for short-term and long-term analysis, for ambulatory patients, for intensive care units, etc.) joined with sophisticated signal processing algorithms has allowed the implementation of "smart" ECG machines, characterized by a very broad spectrum of applications.

Various steps have singled out the development of HRV signal processing algorithms: from the so-called "variance era", basically carried out in the time domain, we passed through the "power spectrum era" in frequency domain. Further, more recent applications have considered time-frequency or time-scale approaches for the detection of transient characteristics which may last few beats, up to the studying of non-linear dynamics with long-range and short range parameters which open to new findings in the interpretation of the cardiovascular functioning. Many non linear parameters are extremely suggestive to support innovative ideas in clinical medicine (like, for example the concept of "dynamical diseases"), even if they need to be more throughly validated from the medical standpoint.

On the other hand, it is indeed through the various methods employed in the analysis of HRV signal that it seems now possible to separate from the signal the deterministic components (characterized by "rhythmic" activity and other non linear properties connected to the complexity of the phenomenon under studying) and the stochastic components (typical of noise superimposed or to the statistical differentiation of the complex biological system).