Application to biomedical signals - Practical Biomedical Signal Analysis Using MATLAB - page 185

detrended fluctuation analysis, modified measures of entropy, Poincare plots.
4.2.2.4.1 Empirical mode decomposition Empirical mode decomposition (EMD)
is based on the decomposition of the original signal into components called instanta-
neous mode functions (Sect. 2.4.2.2.9). The method may be compared to band-pass
filtering, but the sub-bands are not predetermined. The difference relies on the fact
that high versus low frequency discrimination applies only locally, so selection of
modes corresponds to adaptive, signal dependent, time-variant filtering. EMD found
application in separation and tracking of rhythms present in HRV [Echeverria et al.,
2001, Souza Neto et al., 2004].
FIGURE 4.34: Empirical mode decomposition of a real, short-term HRV signal
from a healthy young adult involving standing up movement at 300 s. Top graph
presents original R-R interval series, and C1, C2, C3, C4 were used to describe the
first four components obtained by EMD. Also shown is reconstructed (C1-C4) series
obtained by first four components of decomposition. Plot in bottom graph represents
Hilbert amplitude ratio of third and first components. From [Echeverria et al., 2001].
In [Echeverria et al., 2001] a series of simulations was performed illustrating the
behavior of the method for stationary, non-stationary, and chirp signals. The cal-
culations were also performed for experimental data involving measurements during
rhythmic breathing and for HRV connected with the change of posture from seated to
standing. First, the decomposition into instantaneous mode functions was performed,
then the Hilbert transform was calculated for the consecutive components. Next, the
instantaneous frequency was estimated from the phases of component time series