Research On EEG Patterns On The Spatial-Temporal Domain

Experimental evidence proposes that complex cognitive functions and behavioral acts are arranged at global levels in the brain and they come up from synchronized operation of localize brain areas. In that sense the estimation of the spatial-temporal organization of the brain activity is a promising approach for studying the integrative activity of the brain. This dissertation focused on the method, developed by Professor Walter Freeman group, to study spatial-temporal patterns of brain activity in signals recorded at the cortical surface (ECoG). The main goal was to adapt the methodology, developed for ECoG signal, for it use on signal recorded at the scalp (EEG). It is important because EEG recording is a non-invasive procedure than can be applied repeatedly to patients, normal adults and children with virtually no risk or limitation and makes it possible to take advantage both of the cognitive and phenomenological awareness of a normal healthy subject.Initially an improved method to study the spatial-temporal patterns in ECoG signal was developed and compared with the classical cone fitting. The improved method significantly reduces the processing time while the coincidence level in time domain obtained by both methods is kept high. Consequently, it is a promising tool to study ECoG in real time. In addition AM patterns can be clearly distinguishable using the improved method. Then the improved method was adapted for use on EEG signals and tested in two databases. Due to the well known differences in scale and resolution between ECoG and EEG, adjustments in the techniques to identify stable frames and estimated frame gradient were necessary.The first database analyzed was a database previously uses to study event related potentials. In this case the experiments were conducted trial by trial; each trial was 1 second length after the stimuli presentation. Stable frames were used as time markers to extract high dimensional feature from the analytic power within the time samples that stable frames showed up. The mean level of classification for EEG was comparable to that for ECoG, which shows for the first time that, owing to the scale-free properties of brain activity, these techniques adapted from ECoG can be employed to extract useful information noninvasively from scalp EEG.The second database was continues recording of brain activity during a perceptual-cognitive task, in which the stimulus was a combination of two modalities, auditory and visual stimulation. The objective was to test the hypothesis that patterns of oscillation related to perceptual categories occur in scalp EEG. Global amplitude patterns in the range of 15-22 Hz from all subjects were classified with respect to stimulus combination type significantly above chance levels shortly after the stimulus onset. At the same time the classification rate in the pre-stimulus period or late post-stimulus period was not statistically different from chance.The results showed that this adapted technique can provide the foundation needed to investigate the cerebral dynamics of learning or at least offer some insights that may help substantially in devising further experimental exploration...