Epileptic seizure-classification using probabilistic neural network based on parametric features

Abstract

Neural networks are often used in biological signal processing applications to solve the complexity and to obtain the exact
solution. Likewise, there is a need to integrate science and medicine to make create an effective process. Up to date and
very interesting researches of classification algorithms have been published, but none has effectively focused on
implementing them in brain Electroencephalography (EEG) pattern analyses. In this research, Probabilistic Neural Network
(PNN) is considered for classifying the brain tissue samples by mapping input pattern to a number of classifications. The
dataset is retrieved from the Karunya University for verifying the experiment with10-20 electrodes. Different mental tasks
are considered here to verify the proposed Probabilistic Neural Network based Epileptic Lobe Seizure classifier. Finally,
the experimental results are carried out with several Auto Regression features. Further, the obtained result proves thatthe
proposed PNN model has the maximum accuracy of 96.30%.