Analysis of Complex Network Topological Features for EEG Signals Classification

Electroencephalogram (EEG) signals provide direct insights into the physiological activities of brain, making EEG analysis valuable for understanding cognitive processes and neurological disorders. The main objective of this study is to investigate the effectiveness of extracting combined topological features for epilepsy recognition based on EEG data. We utilize visibility graphs and persistent diagrams to extract both graph features and homological features. These features are then combined to enhance the classification performance, surpassing the results achieved by using graph features alone. Notably, our approach achieves high accuracy rates of 100% for classifying Z and S, 99.75% for N and S, and 97.88% for F and S. These findings significantly contribute to the field of EEG-based epilepsy classification and provide valuable insights for further research and investigation.