Advanced methodological approaches for EEG/MEG micro-structural analysis

The variability of spectral components in the frequency domain has a quasi-stationary quantity. Such fluctuations can be estimated from the local individual spectra of consecutive epochs (derived by subdividing the original recording). In the phenomenon of the EEG/MEG spectral variability the micro-structure of the signal is reflected.

Adaptive Classification Technique, based on the quantification of brain oscillatory activity (short-term spectral patterns description). The algorithm results in n classes of EEG/MEG segments. The segments within each class are considered effectively generated by the same dynamics and with the same driving force. Whereas segments from different classes are considered to have different driving forces and therefore have been effectively generated by different dynamics. Each segment is labeled by the index of the class to which it belongs. As a result a sequence of labels that represent the sequence of micro-states through which the brain passes is obtained.

A single short-term EEG/MEG spectrum illustrates the particular integral dynamics of tens and hundreds of thousands of neurons in a given cortical area at a particular point in time. Therefore, the absence of variance of a single spectrum during several analyzed epochs proves that in a given cortical area the same macro-regimen of neuronal pool activity is maintained during that period. This phenomenon of a temporary stabilization may be explained by stabilization in the brain oscillatory patterns.

The statistical analysis include the calculation of the mean number of individual EEG/MEG epochs for each channel and averaged across all EEG/MEG channels, which follow like sequences by "n elements" in succession without a change in spectral pattern type, where n - is an integer from 1 to 149 (for 1-min EEG). These sequences reflect the life-span of the particular brain oscillation type and depend on the pathological process. All data are compared with the stochastic models.

The operational logic of unrolling brain functional states during the realization of behavioral or psychological acts, assumes that on the EEG/MEG-level the successive operations of neuronal populations' activity may be traced in the palette of segment dynamics of corresponding EEG/MEG rhythmic components (Fingelkurts et al., 2005).

The non-parametric statistical technique, based on the automatic selection of level-conditions in accordance with a given level of probability of "false alerts", identifies and analyzes the dynamic behavior of neuronal population operations. Operations of different neuronal populations expressed within different oscillatory frequencies are analyzed separately as well in relation to each other.

Since the segments in the EEG/MEG signal underlying inherent elementary operations (Fingelkurts et al., 2005), the estimation of segments synchronization between EEG/MEG channels measure the synchrony of operations between different cortical areas.