Information Technology Reference
In-Depth Information
Table 2.1
Examples of possible BCI applications for disabled and healthy individuals
Disabled individuals
Healthy individuals
Restoring mobility
—
e.g. to control wheelchair
movement
(
Mainly control of external devices
)
Environmental control
e.g. to control TV,
power beds, thermostats, etc.
—
Mouse control in PC when
ngers are on the
keyboard
Prosthetics control (motor control replacement)
—
Playing musical instruments by thoughts
to control artificial limbs
Rehabilitative (assistive) control
—
to restore
motor control (e.g.: strengthen/improve weak
muscle)
Virtual reality
Computer games (e.g. Mind Pacman)
Flight/space control (pilots, astronauts)
Biometrics
Fig. 2.2
EEG data processing for a BCI
2.2
Electroencephalogram
Acquiring electroencephalogram (EEG) is the
first step in the BCI design. EEG is a
type of oscillating electrical potential recorded from the scalp surface. It is gener-
ated by neuronal activations in the brain (as shown in Fig.
2.3
) and is very small in
amplitude (in the
V range) due to the attenuation caused by the skull and scalp.
Evoked potentials are a speci
μ
c type of EEG evoked during a stimulus like visual,
auditory, etc.
EEG is usually recorded from a number of electrodes on the scalp. A standard
electrode (channel) con
20 electrode system (Jasper
1958
)of19
active electrodes and two mastoids (reference) as shown in Fig.
2.4
. However, it is
common to extend this con
guration is the 10
-
guration and use higher number of channels such as 32,
64, 128 and even 256. The electrode locations are pre
xed by a letter denoting the
cortical area followed by a number (even for the right hemisphere and odd for the left).
The pre
x letter F stands for frontal, similarly C for central, P for parietal and O for
occipital. The electrodes (normally made with Ag/AgCl) are used with gel to increase
the conductance between scalp and electrodes but there are more recent advances in
using dry electrodes made from gold. It is also common to re-reference the EEG using
methods such as common averaging and Laplacian (Neuper and Klimesch
2006
).
