Analysis of EEG Mapping Images to Differentiate Mental Tasks in Brain-Computer Interfaces - Foundations on Natural and Artificial Computation - page 247

1. Imagination of repetitive self-paced left hand movements (“left” mental task).
2. Imagination of repetitive self-paced right hand movements (“right” mental
task).
3. Generation of words beginning with the same random letter (“word” mental
task).
The data are provided in two ways: raw EEG signals with a sampling rate of 512
Hz, and precomputed features. To obtain these features the raw EEG potentials
were first spatially filtered with a surface Laplacian and then, every 62.5 ms (16
times per second), the power spectral density (PSD) in the band 8-30 Hz was
estimated over the last second of data with a frequency resolution of 2 Hz.
Fig. 1. 10/20 International System with the electrodes used marked in bold
Using the precomputed data, the EEG images have been obtained. The elec-
trodes used to register the EEG signals are the 8 centro-parietal of the 10/20
InternationalSystem:C3,Cz,C4,CP1,CP2,P3,PzandP4(seeFigure1).
The final EEG sample is a 96-dimensional vector (8 channels with 12 frequency
components).
The samples obtained have been plotted for each three users and sessions
using Matlab. In Figure 2 an example of an EEG map can be seen. Each image
represents an EEG map for a particular frequency and mental task. The axis
show the position of the electrodes and the bar is scaled between 0 and 1 to
improve the difference between each electrode. A total of 324 images (3 users x
3 sessions x 3 tasks x 12 frequencies) for a time interval of 5 seconds have been
plotted. This amount of time has been obtained from a qualitative analysis of
images from 5 seconds to 1 minute, averaging the data processed. This analysis
showed that the minimum amount of time in which the different mental tasks
were possible to differentiate was about 5 seconds.