Dimension and entropy analysis of MEG time series from human α-rhythm

Magnetoencephalograms (MEG) from human brain were measured by means of a 37 channel SQUID magnetometer (KRENIKON). Correlation integrals were calculated from time series exhibiting strong α-rhythm in order to give estimates of correlation dimension andK2 entropy. The results are discussed regarding the length and the stationarity of the data. It is shown that low spurious correlation dimensions andK2-entropies may easily be obtained as artefacts due to time correlations in phase space and data length. When time correlations are avoided and the length of time series is taken into account, estimates of correlation dimension andK2 entropy indicate no evidence of the existence of low dimensionality.