A novel feature extraction method for epilepsy EEG signals based on robust generalized synchrony analysis

A feature extraction method for Epilepsy diagnosis is proposed in this paper, which can be incorporated in automatic/semi-automatic epilepsy diagnosis systems to improve diagnosis efficiency from multi-channel Electroencephalogram signals. This method calculates the Robust Generalized Synchrony between pairs of Electroencephalogram channels in the first step. Then six character parameters are extracted from the Robust Generalized Synchrony values for the whole brain and the sub-brain regions. A set of Electroencephalogram data including 20 normal objects and 20 epileptic patients in interictal states were used to test the proposed method The results demonstrate that these features are effective to differentiate between epilepsy patients and the normal objects with the p-values smaller than 0.01.