The fluctuation properties of the human electroencephalogram time series are studied using detrended fluctuation analysis. For nearly all 128 channels in each of the 28 subjects studied, it is found that the standard deviation of the fluctuations exhibits scaling behaviors in two regions. Topographical plots of the scaling exponents reveal the spatial structure of the nonlinear electrical activities recorded on the scalp. Moment analyses are performed to summarize the global variability across channels. The correlation between the two scaling exponents in each channel is also examined. Two global measures are found that succinctly characterize the overall properties of the fluctuation behaviors of the brain dynamics for each subject. Together they distinguish the stroke subjects from the normal ones with 90% accuracy, suggesting the possibility that this analysis could lead to an effective diagnostic tool.

• Received 31 May 2001

DOI:https://doi.org/10.1103/PhysRevE.66.021901