Automated identification and removal of electroencephalogram artifacts with features based on the Hurst exponent.

This paper is focused on the application of the fractal approach to the problem of preliminary processing of electroencephalogram (EEG) data to detect and remove physical and physiological artifacts. We study the fractal properties of EEG data containing artifacts and present the results of the Hurst exponent estimation obtained by the R/S analysis for various types of artifacts and areas of artifact-free data. We show that the Hurst exponent can be used as an informative feature for algorithms identifying and removing artifacts. [ABSTRACT FROM AUTHOR]