Computational Intelligence in Electroencephalogram Analysis - Computational Intelligence in Biomedical Engineering - page 274

FIGURE 6.2 (See color figure following page 204.)
An EEG-monitoring system setup showing wireless EEG electrode connec-
tions to the patient's bedside and a display monitor showing EEG signals.
(From Thakor, N. V., Hyun-Chool, S., Shanbao, T., et al., IEEE Eng. Med.
Biol. Mag. , 25, 20-25, 2006. c

IEEE.)

Hans Berger in 1924, and this proved a significant breakthrough in human
brain research. In this early demonstration, he connected two electrodes to
a patient's skull and was able to detect a very weak current using a gal-
vanometer. Nowadays, clinical EEG machines can be found in many clinics
for routine brain electrical activity monitoring and assisting the physicians
for decision-making processes. Figure 6.2 shows an EEG setup for recording
brain electrical activity using wireless electrodes.
EEG signals are affected by moods such as drowsiness, excitement, and
relaxation. The biopotential recorded from an electrode is a summation of
the electrical activity of the nearby neurons. In clinical tests, recording elec-
trodes are placed on the scalp using international standards EEG geometrical
sites, such as the 10-20 system (Jasper, 1958). EEG recorders can accept sig-
nals from up to 256 electrodes including reference electrodes placed on the
earlobes. Clean contact between the electrode and the skin is essential for
good EEG recording. Conductive gel is usually used to reduce impedance
between the electrode and the scalp. Each electrode is input to a differen-
tial amplifier commonly set between 1,000 and 100,000 amplification. Signal
processing is a combination of low- and high-pass filtering; the low-pass fil-
ter is used to filter the slow electrogalvanic signals, whereas the high-pass
filter is intended to remove any EMG or electromyographic activity from the
EEG waveform.
The EEG signal from the scalp is typically characterized by an amplitude
of approximately 100 µV and a time duration of 0.01-2 s (Geva and Kerem,
1999). Rhythmic sinusoidal activities can be recognized within the EEG signal;
the frequency compositions of the EEG signals commonly used for analysis