A Spectral Based Forecasting Tool of Epileptic Seizures
A new approach to recognize and predict succedent epileptic seizure by using single channel electroencephalogram (EEG) analysis is proposed. Spectral analysis of a brain time series of the left frontal FP1-F7 (LF) scalp location signal is devoted for seizure prediction and analysis.
Important findings showing the presence of preictal spectral changes in studied brain signal are described.
Spectral features occurring during the preictal epoch are extracted from the application of sliding spectral windows of raw EEG at different moments in time preceding the seizure onset.
The same method is then applied to a couple of Intrinsic Mode Functions (IMF1 and IMF2) of the raw EEG (FP1-F7) decomposed by the algorithm of empirical mode decomposition. The main prediction features are derived from the changes of amplitudes, frequency and the number of spikes which are of diagnostic values.
The sliding spectral windows were computed to trace the amplitude changes of higher harmonics during time interval preceding the seizure onset.
Choosing different moments in time aims to identify the best prediction time of seizure onset. Obviously an early prediction time is always desirable but the seizure may result from an abrupt change and so the spectral ‘signs of an imminent seizure occur during a very short prediction time.
From another viewpoint, it may be advantageous to consider a successive prediction times showing the increase of spike numbers and the predominance of certain waves rather than others when approaching seizure onset.
The common prediction features extracted from the analysis of FP1-F7 signal for both patients were mainly the increasing number of spikes of low frequency waves namely delta and theta waves.
Keywords: Electroencephalogram (EEG), Spectral Analysis, seizure onset, sliding window (FFT), Higher Harmonics (HH)
Download Full-Text
ABOUT THE AUTHORS
Hedi Khammari
Dr. Hedi Khammari received his PhD degree in Electrical Engineering from National School of Engineering, Tunis University in 1999. Currently, he is an Associate Professor at college of computers and Information Technology, Taif University. His research interests are mainly in the area of nonlinear dynamics and the application of nonlinear theory in the field of communication, electric systems and bioinformatics.
Ashraf Anwar
Dr. Ashraf Anwar received his PhD degree in Electronics and Electrical Communication from Faculty of Engineering, Cairo University in 2005. Currently, he is an Assistant Professor at college of computers and Information Technology, Taif University. His research interests are mainly in the area of tracking system, control, and robotics.
Hedi Khammari
Dr. Hedi Khammari received his PhD degree in Electrical Engineering from National School of Engineering, Tunis University in 1999. Currently, he is an Associate Professor at college of computers and Information Technology, Taif University. His research interests are mainly in the area of nonlinear dynamics and the application of nonlinear theory in the field of communication, electric systems and bioinformatics.
Ashraf Anwar
Dr. Ashraf Anwar received his PhD degree in Electronics and Electrical Communication from Faculty of Engineering, Cairo University in 2005. Currently, he is an Assistant Professor at college of computers and Information Technology, Taif University. His research interests are mainly in the area of tracking system, control, and robotics.
