Application to biomedical signals - Practical Biomedical Signal Analysis Using MATLAB - page 200

heart rhythm; the shape of the complexes was not considered. After removal of ma-
ternal MCG four approaches for fetal QRS identification were tested: Hilbert trans-
form, ICA, ICA followed by Hilbert transform, and filtering. The Hilbert transform
(HT) yields an analytic signal whose amplitude is always positive independently of
signal polarity (Sect. 2.4.1). The value called rate of change of HT amplitude ( RHA )
was defined as:

2 2

( )= ( − ) +( − )

RHA n x n + 1 x n y n + 1 y n (4.43)

where x n and y n are real and imaginary parts of the transformed signal at point n .The
series RHA

is always positive and signals from different channels may be easily
averaged. The next step was application of FastICA. FastICA (Sect. 3.6.2.3) finds the
columns of separating matrix by maximizing the absolute value of kurtosis, which
works particularly well for QRS extraction due to the high value of kurtosis for QRS
complexes. The third approach used in [Wilson et al., 2008] relied on application
of HT to ICA components. The fourth method involved manual selection of QRS
complexes in a signal filtered in the 1-100 Hz band. The output signals obtained by
the above four methods are shown in Figure 4.46. The next step in analysis involved
correction for missed beats which was performed in a way similar to the method
described in the section on fECG. In the quantitative analysis of results the Hilbert
method scored best, yielding the smallest number of total errors and efficiency of
99.6% in comparison to ICA (94.4%) and ICA-HT (96.4%).
Adaptive ICA was used in real-time fetal heart monitoring. The developed moni-
toring system [Waldert et al., 2007] consisted of a real-time access to fMCG data, an
algorithm based on independent component analysis, and a graphical user interface.
The authors reported that the algorithm extracts the current fetal and maternal heart
signal from a noisy and artifact-contaminated data stream in real-time and is able to
adapt automatically to continuously varying environmental parameters.
In spite of the numerous successful clinical applications based on the measurement
of heart magnetic fields the advantages and disadvantages of ECG and MCG are
still a matter of debate [Hren et al., 1998, Stroink, 2010]. Obviously the advantage
of ECG is lower cost and easier support of the devices. The long years of clinical
experience in ECG have some meaning as well. It seems that in future the decisive
improvement may bring the fusion of both magnetic and electric data.

( n )

4.3 Electromyogram
Electromyogram (EMG) is a record of electrical muscle activity. In this sec-
tion, only EMG of striated muscles will be discussed, the electrical activity of spe-
cific smooth muscles such as intestines—electroenterogram (EEnG) or stomach—
electrogastrogram (EGG), will be discussed in the next section.