Decorrelation of multichannel EEG based on Hjorth filter and graph theory

This paper addresses the spatial domain decorrelation problem in the compression of multichannel Electroencephalograms (EEGs). Multichannel EEGs recorded from a large array of electrodes have correlations among channels, which should be removed in order to gain a higher compression ratio. In this paper, we present a novel decorrelation method by applying the Hjorth filter and graph theory. An algorithm is derived which successively predicts the EEG value in a particular channel based on its neighboring channels. The decorrelated EEGs are more suitable for further compression in the temporal domain. Our results show that, by applying our method, the compression ratio is considerably improved over the traditional methods.