Information Technology Reference
In-Depth Information
motor imagery will consist of two active channels in location C3 and C4 (i.e. motor
cortex area), and EEG is obtained during an imagined movement (say either left or
right hand). The EEG is
filtered in mu and beta bands, and the energy of EEG from
channels C3 and C4 are computed to decide on the movement class:
if energy of C3
EEG
> energy of C4
EEG
: left hand motor imagery
if energy of C4
EEG
> energy of C3
EEG
: right hand motor imagery
if energy of C3
EEG
≈
energy of C4
EEG
: no motor imagery
But this is a crude example and the actual EEG analysis involves several stages
such as determining the appropriate electrode locations, spectral range and use of
features such as band powers and classifiers to obtain accurate detection of the class
of motor imagery.
2.4
EEG-based BCI Paradigm 2
—
SSVEP
SSVEP is a type of EEG that occurs when the visual stimulus
flashes at a frequency
higher than 6 Hz. It is maximal at the visual cortex, speci
cally in the occipital
region. In this paradigm, a target block
flickers with a certain frequency on screen
(the
ashes. The
frequency following effect (sometimes known as photic response) of the brain causes
EEG to oscillate in the frequency of the
flicker can also be achieved using LEDs) and the user looks at the
flickering object. The response is sponta-
neous and does not require any physical effort other than to gaze at the stimulus as
required. In a similar manner, audio-based methods are explored but the results are
not as accurate as the visual-based methods. The detection of the frequency of the
EEG is suf
cient to detect the focused object, though there is a recent study that
showed the possibility of using SSVEP with eyes closed (Lim et al.
2013
).
2.5
EEG-based BCI Paradigm 3
—
P300 VEP
P300 visual evoked potential (VEP) is another type of EEG that is evoked around
300
600 ms after visual stimulus onset (hence the term P300) and is maximal in
midline locations (such as Fz, Cz and Pz). The potential is limited to 8 Hz, and
hence, a low pass
-
filter VEP prior to analysis. It is evoked
in a variety of decision-making tasks and in particular, when a target stimulus is
identi
filter is normally used to
ed, for example when a picture is recognised. A popular paradigm is the
Donchin
s speller matrix paradigm (Donchin et al.
2000
) shown in Fig.
2.6
.It
consists of alphanumeric characters on screen and the rows and columns
'
ash
randomly. The row and column containing the target (focused) character will have a
higher P300 amplitude compared to row or column that contains the unfocused
character. However, this P300 amplitude is not normally detectable in a single trial
due to contamination from higher background EEG and hence require averaging (or
other forms of processing such as PCA and ICA) from a number of trials.
