27.11.2014 Views

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 ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Summary<br />

Since the rsts recordings in humans performed in 1929, the <strong>EEG</strong> has become one <strong>of</strong><br />

the most important diagnostic tools in clinical neurophysiology, but up to now, <strong>EEG</strong><br />

<strong>analysis</strong> still relies mostly on its visual inspection. Due to the fact that visual inspection<br />

is very subjective <strong>and</strong> hardly allows any statistical <strong>analysis</strong> or st<strong>and</strong>ardization, several<br />

<strong>methods</strong> were proposed in order to quantify the information <strong>of</strong> the <strong>EEG</strong>. Among these,<br />

the Fourier Transform emerged as a very powerful tool capable <strong>of</strong> characterizing the<br />

<strong>frequency</strong> components <strong>of</strong> <strong>EEG</strong> <strong>signals</strong>, even reaching diagnostical importance. However,<br />

Fourier Transform has some disadvantages that limit its applicability <strong>and</strong> therefore,<br />

other <strong>methods</strong> for extracting \hidden" information from the <strong>EEG</strong> are necessary.<br />

In this work, I described, extended <strong>and</strong> compared <strong>methods</strong> <strong>of</strong> <strong>analysis</strong> <strong>of</strong> dierent<br />

types <strong>of</strong> <strong>EEG</strong> <strong>signals</strong>, namely time-<strong>frequency</strong> <strong>methods</strong> (Gabor Transform <strong>and</strong> Wavelet<br />

Transform) <strong>and</strong> Chaos <strong>methods</strong> (attractor reconstruction, Correlation dimension, Lyapunov<br />

exponents, etc.).<br />

<strong>Time</strong>-<strong>frequency</strong> <strong>methods</strong> provided new information about sources <strong>and</strong> dynamics <strong>of</strong><br />

Gr<strong>and</strong> Mal (Tonic-clonic) seizures, something very dicult to obtain with conventional<br />

<strong>methods</strong>. Gr<strong>and</strong> Mal seizures were rst dominated by alpha (7:5 ; 12:5Hz) <strong>and</strong> theta<br />

(3:5;7:5Hz)rhythms, these rhythms later becoming slower in correlation with the starting<br />

<strong>of</strong> the clonic phase. The dynamics <strong>of</strong> the <strong>frequency</strong> patterns during these seizures<br />

was very interesting in relation to processes <strong>of</strong> neuronal fatigue, neurotransmitter disbalance,<br />

similarity with animal experiments <strong>and</strong> computer simulations. The <strong>analysis</strong><br />

with Chaos theory showed a decrease in parameters as the Correlation Dimension or<br />

the maximum Lyapunov exponent, parameters that characterize the complexity <strong>and</strong><br />

\chaoticity" <strong>of</strong> the signal. These results showed a transition to a more simple system<br />

during epileptic seizures.<br />

In order to study basic features <strong>of</strong> brain oscillations, I analyzed event-related responses<br />

(i.e. alterations <strong>of</strong> the ongoing <strong>EEG</strong> due to an external or internal stimuli) with<br />

recent <strong>methods</strong> <strong>of</strong> time-<strong>frequency</strong> <strong>analysis</strong>. In this context, the study <strong>of</strong> event-related<br />

alpha oscillations (i.e. event-related responses in the alpha range) showed that these<br />

responses were distributed along the scalp with signicant dierences in their delays<br />

between anterior <strong>and</strong> posterior electrodes. This result implied that several sources were<br />

involved in the origin <strong>of</strong> the event-related alpha oscillations. Furthermore, their independence<br />

on the performance <strong>of</strong> a cognitive task, the best denition in occipital locations<br />

<strong>and</strong> the short latency <strong>of</strong> the responses pointed towards a relation between event-related<br />

alpha oscillations <strong>and</strong> primary sensory processing.<br />

The study <strong>of</strong> the responses upon bimodal stimulation (simultaneous visual <strong>and</strong> audi-<br />

1

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!