Gabor Transform and Empirical Mode Decomposition successfully differentiated epilepsy periods (interictal, preictal, and ictal) and extracted patterns from EEG signals.
Gabor Transform and Empirical Mode Decomposition are effective computational methods for extracting features from EEG signals to assist in diagnosing epilepsy periods.
Electroencephalograph (EEG) has been considered as a practical media to explore human brain activities. It is believed that EEG signals have lots of information carried still unknown. The non-stationary, non-linear traits of EEG signals make the information detection a hard task. While time-frequency methods, for their superiority to process such data, were widely studied and applied to this research. EEG information detection is very important during the diagnostics process of epilepsy diseases, because doctors detect abnormal brain activities mainly with their experiences on EEG signals and such subjective method is not so reliable. Here, we try a time-frequency method (Gabor Transform) on EEG signals. The results of Gabor Transform display good performance on both time and frequency scales. The Frequency Band Relative Intensity Ratio (FBRIR) can clearly differentiate the epilepsy periods including interictal, preictal and ictal. Empirical Mode Decomposition (EMD) is also used to extract patterns from the original EEG signals. It shows that EMD can be a valuable practical method for such tasks. The results of the two methods can provide doctors with clinical guidelines.
Chen et al. (2010) studied Epilepsy. Gabor Transform and Empirical Mode Decomposition (EMD) was evaluated on Differentiation of epilepsy periods and pattern extraction. Gabor Transform and Empirical Mode Decomposition successfully differentiated epilepsy periods (interictal, preictal, and ictal) and extracted patterns from EEG signals.