Quantitative analysis of EEG signals: Time-frequency methods and ...
Quantitative analysis of EEG signals: Time-frequency methods and ...
Quantitative analysis of EEG signals: Time-frequency methods and ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
vertical lines, is showed a transient with a <strong>frequency</strong> clearly lower than the range <strong>of</strong> alpha<br />
b<strong>and</strong>. The digital ltering does not resolve this transient <strong>and</strong> it spuriously \interpolates"<br />
alpha oscillations in continuity with the ones that precede or follow the transient. On<br />
the other h<strong>and</strong>, the wavelet coecients show a decrease in this time segment, this<br />
phenomenon being also visible in the reconstructed form. Something similar occurs in<br />
sweep #3 with the transientmarked with an arrow. In fact in this last case, the transient<br />
is due to the cognitive P300wave typically obtained upon target stimuli. With wavelets<br />
it is visible that, as in the original signal, there is no important contribution <strong>of</strong> alpha<br />
oscillations in this time range, the digital ltering having not enough resolution for<br />
resolving this. The better time-<strong>frequency</strong> resolution <strong>of</strong> wavelets (in this case a better<br />
<strong>frequency</strong> localization for a certain time range) can be also seen in sweep #4. In the<br />
original signal, in between the vertical dashed lines there is a marked oscillation <strong>of</strong><br />
about 4 ; 6Hz, corresponding to the theta b<strong>and</strong>. The digital ltered signal shows an<br />
alpha oscillation not present in the original signal. On the other h<strong>and</strong>, due to its better<br />
resolution the wavelet coecients <strong>and</strong> the reconstructed signal show a clear decrease<br />
for this time range. Finally, sweep #5 shows a ringing eect (i.e. spurious oscillations<br />
appearing before the stimulation time point due to time resolution limitations). The<br />
oscillation before the stimulation time, marked with an arrow, appears in the digital<br />
ltered signal with more amplitude than in the original signal, this eect being overcome<br />
with wavelets.<br />
4.5.3 Results<br />
The gr<strong>and</strong> average wideb<strong>and</strong> ltered (0:5 ; 100Hz) event-related potentials are shown<br />
in gure 19. The P100 response is clearly visible upon all stimuli types at about 100ms,<br />
best dened in occipital locations, where it reaches amplitudes about 8V . In the case<br />
<strong>of</strong> target stimulation, a marked positive peakappears between 400 ; 500ms, according<br />
to the expected cognitive (P300) response (see sec. x x1.2).<br />
Figure 20 shows the wavelet components in the alpha b<strong>and</strong> (for brevity, \alpha<br />
responses") for the subject JA (for a better visualization <strong>of</strong> the responses, the signal<br />
reconstructed from the alpha b<strong>and</strong> wavelet coecients is shown). One second pre- <strong>and</strong><br />
one second post-stimulation <strong>EEG</strong> are plotted. Alpha components corresponding to the<br />
pre-stimulus <strong>EEG</strong> have about 5V <strong>and</strong> upon all stimulus types, amplitude enhancements<br />
are clearly marked in posterior locations reaching values up to 20V . Furthermore, in<br />
posterior electrodes responses upon TARGET stimulation are prolonged compared with<br />
the other two stimulus types.<br />
Results for the gr<strong>and</strong> average <strong>of</strong> the 10 subjects (g. 21) are similar to the ones<br />
outlined for the rst subject. Amplitude increases were distributed over the whole scalp<br />
54