Development of neural network classifier for classification of epileptic seizures in EEG signals

Epilepsy is a type of neurological disorder which is prevalent in many people next to stroke. Electroencephalogram (EEG) recordings give valuable information about epileptic disorders. It plays a vital role in understanding the brain activity. A slight variation in EEG signals shows a definite type of abnormality in the brain. The objective of this work is to design techniques which could be used for treatment of patients with epilepsy. This paper presents various neural network architectures to identify if a person has epileptic seizure or not. This is done by extracting different statistical features like mean, median, standard deviation from the collected EEG signals. The three network architectures like Feed Forward network, Elman Network, Cascade Forward Backpropagation Network are used and compared using various performance measures. The results show that Elman network is superior to other networks.