Feature Extraction and Classification of EEG Signals. The Use of a Genetic Algorithm for an Application on Alertness Prediction - Guide to Brain-Computer Music Interfacing - page 193

Fig. 9.8 Graphical user interface for the CNV display

an user can easily plot the CNV curve for a given participant. The top right of
Fig. 9.8 is a topographic map. At a given period, it represents the electrical activity
recorded on the scalp of a participant. It allows to view the appearance of the CNV
on the scalp and thus to locate brain regions involved in the CNV appearance.
The study of CNV was performed on the 44 participants of the study and 13
participants were kept for further analysis. Thus, an important number of partici-
pants are rejected. The stress due to the experiment and the duration of the
installation of the cap may be factors that deteriorate the ef

ciency of the relaxation
session. To limit the duration of the cap wearing, the relaxation session is relatively
short. Thus, it is possible that the duration of the relaxation session (20 min) is too
short to achieve fully relax these subjects. The participants selected are those that
have special abilities to relax in stressful conditions and in a relatively short period
of time. Those points can explain the high proportion of rejected participants in our
study.

9.3.2 Data

Finally, the data consist of 26 records of 3 min of raw EEG signals from the 13
selected participants (one EEG for each partici-

pant). Each record contains variations of electric potential obtained with a sampling
frequency of 256 Hz with 58 active electrodes placed on a cap (ElectroCap). Using
this sampling frequency, each signal recorded by an electrode for a given subject in
a given alertness state contains 46,000 data points. A representation of the data
matrix is given in Fig. 9.9 .

“ normal ” EEG and one “ relaxed ”