206
Biomedical Signal and Image Processing
Since the stimuli are not very specific, the ERPs are extremely small in amplitude.
The standard mode of operation in such situations is to perform repetitive measurements and to average the signals. Ideally, all the signals in these repeated experiments
must have a similar response (in the strict theoretical sense); often in practical applications, the measured signals are stochastic signals that are in the best condition
stationary processes. As discussed in the analysis of stochastic processes, averaging
the actual responses would increase the signal-to-noise ratio; however, the unique
response of a single stimulus may be lost. In typical analysis of EPs and ERPs, stochastic
analysis is often preferred over the deterministic signal processing.
The investigation of EPs and ERPs usually focuses on establishing characteristic
changes in typical signal components. These characteristic features usually receive
a designation based on the polarity of the initial signal deflection, either positive
or negative, in combination with the time lag in milliseconds. For instance, P300
stands for a positive peak after 300 ms. As another example, an auditory stimulus
will have a typical response signal with approximately 100 ms latency compared to
the applied stimulus. Such a response is typically a negative deflection and is referred
to as N100, due to the 100 ms delay and the negative nature.
ERP EEG measurements are often used to investigate neurophysiologic correlation between factual knowledge, awareness, and attention. ERP measurements
can also be used to identify specific components or patterns in the electromyogram
(EMG) signal, which is the electric activation of muscle tissue. The EMG will be
discussed later in a separate chapter.
10.5 DISEASES OF CENTRAL NERVOUS SYSTEM AND EEG
In clinical applications, EEG is commonly used to diagnose diseases in the CNS.
While the cerebellum, thalamus, and spinal cord do not offer large enough signal
amplitudes to make clinical observations, signals originated at the cortex are heavily
used for biomedical diagnostics. A typical use of EEG, as discussed later, is the diagnosis of epilepsy. This is the application that made EEG a routine clinical test of the
CNS. Some of the important diagnostic applications of the EEG are described next.
10.5.1 EPILEPSY
Epilepsy is a chronic illness identified by irregular occasions of unconsciousness,
sometimes associated with violent convulsions. It is a pathological condition of a
group of nerve cells firing harmoniously with a certain frequency. With the default
location of the nerve cells in the cortex, this will manifest in a measurable EEG
signal. It is estimated that approximately 1 out of 2000 people per year will develop
some type of epilepsy. In addition, it is speculated that 1 out of every 20 people will
have at least one spontaneous epileptic occurrence in their lifetime.
From the standpoint of diagnostics, it is a definite inconvenience that epileptic
attacks are relatively unpredictable. However, sometimes epilepsy can be induced
under laboratory conditions. In general, epilepsy may be induced by periodic events
such as flickering lights in a discotheque or while traveling down a road lined with
trees, providing an alternating sun-shade sequence. In the clinical environment,
Biomedical Signal and Image Processing
Since the stimuli are not very specific, the ERPs are extremely small in amplitude.
The standard mode of operation in such situations is to perform repetitive measurements and to average the signals. Ideally, all the signals in these repeated experiments
must have a similar response (in the strict theoretical sense); often in practical applications, the measured signals are stochastic signals that are in the best condition
stationary processes. As discussed in the analysis of stochastic processes, averaging
the actual responses would increase the signal-to-noise ratio; however, the unique
response of a single stimulus may be lost. In typical analysis of EPs and ERPs, stochastic
analysis is often preferred over the deterministic signal processing.
The investigation of EPs and ERPs usually focuses on establishing characteristic
changes in typical signal components. These characteristic features usually receive
a designation based on the polarity of the initial signal deflection, either positive
or negative, in combination with the time lag in milliseconds. For instance, P300
stands for a positive peak after 300 ms. As another example, an auditory stimulus
will have a typical response signal with approximately 100 ms latency compared to
the applied stimulus. Such a response is typically a negative deflection and is referred
to as N100, due to the 100 ms delay and the negative nature.
ERP EEG measurements are often used to investigate neurophysiologic correlation between factual knowledge, awareness, and attention. ERP measurements
can also be used to identify specific components or patterns in the electromyogram
(EMG) signal, which is the electric activation of muscle tissue. The EMG will be
discussed later in a separate chapter.
10.5 DISEASES OF CENTRAL NERVOUS SYSTEM AND EEG
In clinical applications, EEG is commonly used to diagnose diseases in the CNS.
While the cerebellum, thalamus, and spinal cord do not offer large enough signal
amplitudes to make clinical observations, signals originated at the cortex are heavily
used for biomedical diagnostics. A typical use of EEG, as discussed later, is the diagnosis of epilepsy. This is the application that made EEG a routine clinical test of the
CNS. Some of the important diagnostic applications of the EEG are described next.
10.5.1 EPILEPSY
Epilepsy is a chronic illness identified by irregular occasions of unconsciousness,
sometimes associated with violent convulsions. It is a pathological condition of a
group of nerve cells firing harmoniously with a certain frequency. With the default
location of the nerve cells in the cortex, this will manifest in a measurable EEG
signal. It is estimated that approximately 1 out of 2000 people per year will develop
some type of epilepsy. In addition, it is speculated that 1 out of every 20 people will
have at least one spontaneous epileptic occurrence in their lifetime.
From the standpoint of diagnostics, it is a definite inconvenience that epileptic
attacks are relatively unpredictable. However, sometimes epilepsy can be induced
under laboratory conditions. In general, epilepsy may be induced by periodic events
such as flickering lights in a discotheque or while traveling down a road lined with
trees, providing an alternating sun-shade sequence. In the clinical environment,