Research Area:  Machine Learning
Automated classification of an epileptic seizure is very crucial for efficient diagnosis and treatment management in the health monitoring applications. Finding traces of epilepsy through the visual marking of long Electroencephalogram (EEG) recordings by human experts is a very tedious, time-consuming and high-cost task. It is always a challenging issue for the researchers and neurologist to automatically detect epilepsy disorder from EEG signals which contains huge fluctuating information about the functional behavior of the brain. The Complex network-based time-series analysis approach has the ability to perfectly describe the principal dynamics of the EEG signals. By considering this fact, this paper aims to propose a graph theory based innovative framework and a new complex network feature that is efficient for the automated classification of EEG signals to detect epilepsy. In this study, we introduce a new method for the mapping of time series EEG signals to complex network. We also develop a new feature, named as “Edge Weight Fluctuation (EWF)” that helps to extract sudden fluctuation in EEG signals. The proposed scheme is tested on two benchmark Epileptic EEG databases (Bern-Barcelona EEG database and Bonn University EEG database). In order to check the validity of our proposed methodology, we perform simulation analysis with two different chaotic signals named as Henon map and Logistic map. We also performed the One-Way ANOVA statistical test. The overall accuracy has achieved 99% for Bern-Barcelona database and 100% for Bonn University database. The experimental results reveal that our proposed methodology is more efficient to distinguish epileptic seizure signal from between diverse EEG signals.
Author(s) Name:  Supriya Supriya, Siuly Siuly, Hua Wang, Yanchun Zhang
Journal name:  Applied Acoustics
Publisher name:  Elsevier
Volume Information:  Volume 180, September 2021, 108098
Paper Link:   https://www.sciencedirect.com/science/article/abs/pii/S0003682X21001912