Abstract
Epilepsy indicates a short time of abnormalities in the function of brain that can cause serious brain damage. A common technique when analyzing with Epilepsy is based on EEG signals processing, taken from the surface area of human scalp. In this paper, mainly focused on the selection of how much less processing and complexity can be used for features extraction with the ability to provide high accuracy of classification. For this, features are extracted in three steps on the basis of how much computation are required in the processing. These three steps includes features directly from raw EEG, preprocessing EEG and wavelet decomposing. Time domain, frequency domain and time-frequency domain features are extracted for the classification purpose. After that, in each steps three experiments are performed to classify into two particular classes in each experiment. These classification are done by using Feed Forward Neural Network (FFNN). In the 3rd experiment where classification is done between seizure subjects vs. non-seizure subjects, FFNN gives accuracy as high as 99.2%.

This publication has 12 references indexed in Scilit: