Cascaded feed forward neural networks and generalized regression for epilepsy risk level classification — A study

Abstract

Characterized by transient, sudden and recurrent electrical disturbances in the cortical regions of the brain, epilepsy can be mentioned as one of the most common neurological disorders. Certain disorders in the nervous systems also cause this fatal condition epilepsy. The activities of the brain can be monitored through Electroencephalogram (EEG) and thus it has become a vital tool for the analysis and diagnosis of epilepsy. Approximate Entropy (ApEn) seems to be a very good measure to understand the non linear nature of the biological signals and therefore ApEn is employed as a feature extraction technique and Cascaded Feed Forward Neural Network (CFFNN) and Generalized Regression Neural Network (GRNN) are utilized as Post Classifiers for the study and Classification of Epilepsy from EEG Signals. The important validation parameters taken here are Performance Index (PI), Quality Value (QV), Time Delay, Sensitivity, Specificity and Accuracy.