Amplitudenkopplung zwischen Kortexsignalen: Ein bislang ungenutzter Indikator für kooperative Hirnprozesse beim Menschen

The widely distributed information processing in the brain requires extensive dynamic interaction among neuronal processes in different regions or areas of the brain. Activities which are relevant for the processing of sensory stimuli and for higher cognitive functions are often assumed to be reflected in electrophysiological signal components in the higher, so-called gamma-frequency range (30–90 Hz). Searching for specific coupling events among these signal components therefore is an integral part of current brain research. Although neuronal cooperativity in the brain may be expected to occur not only across shorter distances, but also across several centimeters, there is evidence that the coupling range of gamma-signals recorded directly from the cortex or from its surface is usually restricted to few millimeters. As signal theory provides several different techniques for coupling analyses, the question arises whether the commonly used coupling measures may sometimes not be appropriate to the neuronal mechanims which are relevant for long-range interactions. Therefore, one of the objectives of the present work was to systematically compare different coupling measures in the analysis of intracranial brain signals. In the case of diverging results, this comparison was supposed to indicate the neuronal mechanisms possibly underlying certain forms of coupling. Extensive recording of brain signals directly from the human cortex was performed with subdural electrodes that had been implanted for the purpose of presurgical diagnostics in clinical patients at the Bethel Epilepsy Centre (Bielefeld, Germany). Since there had been only little experience with such measurements in experimental science, the first objective of my work was to generally prove the reliability and usability of statements derived from subdural signals. In order to measure brain activity under standardized conditions, I designed an experimental paradigm in which subjects had to perform certain visual-cognitive tasks. They were supposed to compare abstract geometrical patterns with respect to their shape or to their spatial position, respectively. Thus, visual processing pathways in different brain regions should be specifically activated, which in turn should serve to roughly assess the functional meaning and significance of activity and coupling patterns that would occur in the signals. On the basis of a time-resolved spectral analysis of the recorded cortical signals, I was able to identify several foci of activity, which were distinguishable by characteristic time-frequency patterns of spectral amplitude. These patterns showed clear spatial separability, high intraindividual reproducibility, and each of them appeared at homologous sites in several subjects. One of the seven subjects was especially suitable for a case study comprising a detailled coupling analysis, because two functional visual cortical areas were distinguishable in this subject. The most remarkable outcome was a highly specific dynamic interaction between these two areas, which showed as a transient temporal correlation between amplitude envelopes of gamma-signals in one area and slow signal components in the other area. Since this coupling event was not detected by any of the other measures, and since it spanned a distance of several centimeters, I performed a comprehensive coupling analysis including all subjects in order to show general differences among the coupling measures. It turned out that dynamic long-distance signal coupling in the gamma-frequeny range is more often reflected by amplitude-based coupling measures than by measures preserving the signals’ phase structure. The fact that the results were highly specific, spatially structured, intraindividually reproducible, and sometimes remarkably consistent across subjects, suggests that subdurally recorded cortical signals are an informative indicator of neuronal processes. Based on this finding, I was able to show that many neuronal interactions across larger distances do manifest themselves in forms of signal coupling involving gamma-signals, but that these forms of coupling are detected by amplitude-based rather than by phase-based analysis techniques. The reason of this phenomenon might be that during propagation along nerve fibres, neuronal activity is subject to temporal dispersion due to some variability of certain physiological parameters. Such dispersion can be assumed to become larger with increasing distance and may change the phase structure of the measured brain signals especially in the high-frequency range. It can be concluded that the employed coupling measure has considerable influence on the outcome of any brain signal analysis, and the decision for one or the other measure should therefore take into account a mechanistic concept of the neuronal interaction to be expected.