Code Converters with City Block Distance Measures for Classifying Epilepsy from EEG Signals

Abstract

This paper aims to compare the performance of code-converter as a feature extraction technique followed by various distance measures such as City Block Distance (CBD) measure and Euclidean Distance (ED) measure for the perfect classification of epilepsy risk levels from Electroencephalography (EEG) Signals. From the extracted parameters such as sharp and spike waves, energy, peaks, duration, variance, events and covariance from the EEG Signals of an epileptic patient, the risk level of epilepsy is classified. The City Block Distance Measure and Euclidean Distance Measure is then applied to the Code Converter's risk level classification output in order to optimize the risk levels for the characterization of the patient. In this study, a group of 10 patients with known epilepsy findings are utilized here. The Performance metrics is computed with the help of parameters like Performance Index (PI) and Quality Values (QV).