Defining a Fuzzy Rule Based Index for Epileptic Seizure Risks

Hossein Mamaghanian, Biomedical Image and Signal Processing Lab, Sharif University of Technology, Tehran, Iran

Abstract: A method capable of predicting the occurrence of Epileptic seizures is still a challenging area in Epilepsy Research. Many works and studies in this field tried to present a new approach capable of predicting the seizure onset from the recorded Electroencephalogram (EEG) Signals. But none of these could reach optimistic results to be used in practical and clinical application so far. This is very similar to estimating the depth of anesthesia (DOA). The lack of a general measure and feature in representing the DOA, leads to introducing the Fuzzy Rule-Base Index (FRI) for quantizing the DOA. The proposed method is based on the analysis of multi-channel EEG using presented and successful features up to now. Also, we used the “Liley” EEG model as a dynamical model of EEG. Then SIS particle filter is applied forestimating the defined states over time using the recorded epileptic EEG as the observation of the system. The talk will continue about other works in BCI application and a newly proposed method based on generalized Eigen value decomposition to extract single trial EPs from single channel EEG recordings. The extraction of the N75-P100-N135 complex in simulated and actual visual evoked potentials is mainly taken under consideration.

About the speaker: Hossein Mamaghanian was born in 1983. He received His B.Sc. in Electrical Engineering from the University of Tabriz in 2006. In September 2006 he joined to the Biomedical Image and Signal Processing Lab (BISPL) at Sharif University of Technology. Currently he is working on his M.Sc. thesis under the supervision of Prof. M. B. Shamsollahi about "Defining an EEG Index for Seizure Prediction". He has been a researcher in Pasture Institute (2008 till now), Kia Co. (2008) and Moj. Co (2008 till now). His Research Interests include Statistical/Biomedical Signal Processing, Feature extraction and Pattern Recognition.