Networking Property During Epileptic Seizure with Multi-channel EEG Recordings
EEG recordings are widely used in epilepsy research. We intend to address a question whether small world network property exists in neural networks when epileptic seizures occur. In this paper, we introduce a bispectrum analysis to calculate the interaction between two EEG recordings; then, a suitable threshold is chosen to convert the interaction of the six channels at five frequency bands to a sparse graph (node: n=30, edge: k=4-7). Through analyzing a real EEG recording, it is found the clustering coefficient is similar to that of regular graph; however the path length is less than that of regular graph. Thus a primary suggestion can be made that neural networks possess small world network characteristic. During epileptic seizures, the small world property of neural network is more significant.
KeywordsRandom Graph Epileptic Seizure Cluster Coefficient Regular Graph Small World
Unable to display preview. Download preview PDF.
- 2.Hoeve, M.J., Jones, R.D., Carroll, G.J., Goelz, H.: Automated Detection of Epileptic Seizures in the EEG. In: Proceedings of the 23rd Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Istanbul, Turkey, vol. 3, pp. 943–946 (2001)Google Scholar
- 12.Swami, A., Mendel, J.M., Nikias, C.L. (Max): Higher-Order Spectral Analysis Toolbox. The MathWorks, Inc. (1998)Google Scholar
- 13.Kochen, M. (ed.): The Small World., Ablex, Norwood, New Jersey (1989)Google Scholar