Biomedical Engineering Reference
In-Depth Information
quency in the power spectrum. Spectral power as a function of time is used for track-
ing the changes of EEG during anesthesia. The methods described in Section 2.4.2
may be applied for this purpose. Another measure used in anesthesia monitoring is
bicoherence (Sect. 2.3.2.1.4).
A measure based on the concept of entropy, so called spectral entropy, is applied
during anesthesia and critical care. Spectral entropy SE for a given frequency band
is given by the formula [Rampil, 2009]:
f 2
f i =
SE
(
f 1
,
f 2
)=
P n
(
f i
)
log P n
(
f i
)
(4.5)
f 1
where P n
is a normalized power value at frequency f i . This measure may be
normalized in the range
(
f i
)
(
0
,
1
)
taking into account the number of data points n
(
f 1
,
f 2
)
in the band
(
f 1
,
f 2
)
. Normalized spectral entropy SE N is given by the formula:
SE
(
f 1
,
f 2
)
SE N
(
f 1
,
f 2
)=
(4.6)
log n
(
f 1
,
f 2
)
4.1.7 Analysis of epoched EEG signals
In the previous section we discussed methods of signal analysis that are suitable
for getting insights into ongoing, spontaneous brain activity.
If we consider the brain as a system, we can learn a lot about it, observing and
analyzing reactions of the system to specific perturbation of the spontaneous activ-
ity. The perturbations are usually caused by different stimuli. Studies of reaction of
the brain to the stimuli require focusing the analysis on a period of time surrounding
the moment of stimulation. The methods developed for this type of analysis try to
quantify the changes in the EEG/MEG signal that are provoked by the external or
internal stimuli. These changes can be globally referred to as event-related potentials
(ERP), a subset of them related to sensory (visual, auditory, somatosensory) stimuli
are commonly called evoked potentials (EP). In case of MEG signals, the respective
names are: event-related fields (ERF) and evoked fields (EF). For the sake of sim-
plicity we shall refer further in this section to the EEG signals: ERP and EP, keeping
in mind that the described methods can be applied equally well to the analysis of
MEG signals.
The voltage deflections observed in the ERP reflect the reception and processing
of sensory information, as well as higher level processing that involves selective
attention, memory updating, semantic comprehension, and other types of cognitive
activity. In clinical practice ERPs are viewed as a sequence of components which are
defined by their positive or negative polarity (in respect to the reference electrode
potential), their latencies, their scalp distribution, and the relation to experimental
variables. The clinical and research interest in ERPs relies on the fact that they are
linked in time with physical or mental events [Duncan et al., 2009].
In clinical neurophysiology the ERP or EF is described as the response averaged
across tenths or hundreds of repetitions. The diagnostic strength of analysis of ERP
 
Search WWH ::




Custom Search