213
Electroencephalogram
amplitude exceeding a threshold that is at least six times the average amplitude of the
recording over the preceding 10 s. This criterion needs to be combined with boundary conditions of the algorithm that ensure the capture of the entire artifact.
Electric signals originating from muscles usually have a steeper slope than the
average EEG signal. The use of slope threshold or steepness threshold can be used
to minimize the influence of these types of muscle artifacts. The slope of a curve
is found by estimating the first derivative of the signal. The first- and second-order
derivatives of the slope are also used to form a complexity measure from the EEG
signal. These measures are highly similar to signal complexity and signal mobility
measures introduced in Part I of the book.
Other complexity measures such as fractal dimension and entropy are also heavily used as features of the EEG signal. A very important rule regarding the fractal
dimension of the EEG signal indicates that the fractal dimension of EEG falls by
age. This means that older people often have much smaller fractal dimensions compared to younger people. An intuitive justification of this phenomenon is based on
the internal complexity of the brain. From the definition of fractal dimension, it is
evident that systems with complex modular structure will provide a higher fractal
dimension. More specifically, a system, where many active subsystems generate signals, will produce an overall signal that reflects the intrinsic hierarchical or modular
structure in the form of a highly fractal signal. In a young brain where all parts of
the brain are active, the hierarchical and modular structure is more significant. But,
as the person grows older, some parts of the brain become considerably less active,
which in turn reduces the complexity of the EEG signal. The same decreases in other
complexity measures are recorded in EEGs of older people.
It is also reported that all complexity measures of the EEG signal and, in particular, the fractal dimension decrease in diseases such as epilepsy and Alzheimer’s.
This is again due to complexity reduction of the brain due to the reduction in modularity of brain activities. The reduction of almost 30% fractal dimension of EEG in
epilepsy is used as a diagnostic criterion to detect epilepsy.
An efficient method of EEG analysis is designed based on the coherence of the
recorded signals, as described in the following.
10.7.3.1 Coherence Analysis
Another type of EEG characterization is based on the synchronicity of any pairs of
signals in the 20 channels. The measure of coherence between channels can reveal
details on the efficiency of the brain functions. One common method of EEG synchronicity analysis is achieved by comparing the recordings of the electrodes in the
left brain to the corresponding electrodes in the right brain. A powerful tool for this
type of comparison is averaging, gated by specific stimuli such as the ERP methods,
which will be covered in the following subsections.
Coherence analysis is also performed between the specific waves (frequency
bands) of the EEG signals. In the description of the alpha wave spectrum, it was indicated this spectrum represents a significant level of consciousness. Since the patient
is alert during both alpha and beta rhythms, especially during the beta rhythm, these
wave patterns will display a certain level of coherence. In contrast, since the theta
waves are generated under relaxation with little or no sensory input. These waves do
Electroencephalogram
amplitude exceeding a threshold that is at least six times the average amplitude of the
recording over the preceding 10 s. This criterion needs to be combined with boundary conditions of the algorithm that ensure the capture of the entire artifact.
Electric signals originating from muscles usually have a steeper slope than the
average EEG signal. The use of slope threshold or steepness threshold can be used
to minimize the influence of these types of muscle artifacts. The slope of a curve
is found by estimating the first derivative of the signal. The first- and second-order
derivatives of the slope are also used to form a complexity measure from the EEG
signal. These measures are highly similar to signal complexity and signal mobility
measures introduced in Part I of the book.
Other complexity measures such as fractal dimension and entropy are also heavily used as features of the EEG signal. A very important rule regarding the fractal
dimension of the EEG signal indicates that the fractal dimension of EEG falls by
age. This means that older people often have much smaller fractal dimensions compared to younger people. An intuitive justification of this phenomenon is based on
the internal complexity of the brain. From the definition of fractal dimension, it is
evident that systems with complex modular structure will provide a higher fractal
dimension. More specifically, a system, where many active subsystems generate signals, will produce an overall signal that reflects the intrinsic hierarchical or modular
structure in the form of a highly fractal signal. In a young brain where all parts of
the brain are active, the hierarchical and modular structure is more significant. But,
as the person grows older, some parts of the brain become considerably less active,
which in turn reduces the complexity of the EEG signal. The same decreases in other
complexity measures are recorded in EEGs of older people.
It is also reported that all complexity measures of the EEG signal and, in particular, the fractal dimension decrease in diseases such as epilepsy and Alzheimer’s.
This is again due to complexity reduction of the brain due to the reduction in modularity of brain activities. The reduction of almost 30% fractal dimension of EEG in
epilepsy is used as a diagnostic criterion to detect epilepsy.
An efficient method of EEG analysis is designed based on the coherence of the
recorded signals, as described in the following.
10.7.3.1 Coherence Analysis
Another type of EEG characterization is based on the synchronicity of any pairs of
signals in the 20 channels. The measure of coherence between channels can reveal
details on the efficiency of the brain functions. One common method of EEG synchronicity analysis is achieved by comparing the recordings of the electrodes in the
left brain to the corresponding electrodes in the right brain. A powerful tool for this
type of comparison is averaging, gated by specific stimuli such as the ERP methods,
which will be covered in the following subsections.
Coherence analysis is also performed between the specific waves (frequency
bands) of the EEG signals. In the description of the alpha wave spectrum, it was indicated this spectrum represents a significant level of consciousness. Since the patient
is alert during both alpha and beta rhythms, especially during the beta rhythm, these
wave patterns will display a certain level of coherence. In contrast, since the theta
waves are generated under relaxation with little or no sensory input. These waves do