AU - Hosseini, Seyyed Abed AU - Akbarzadeh Totonchi, Mohammad Reza AU - Naghibi-Sistani, Mohammad Bagher TI - Epilepsy Recognition using Chaotic Qualitative and Quantitative Evaluation of EEG Signals PT - JOURNAL ARTICLE TA - icss JN - icss VO - 17 VI - 2 IP - 2 4099 - http://icssjournal.ir/article-1-338-en.html 4100 - http://icssjournal.ir/article-1-338-en.pdf SO - icss 2 ABĀ  - Introduction: Epilepsy is a disease of the central nervous system presenting by sudden convulsive attacks over a period of time. Method: A key issue in recognition systems is optimal data acquisition as well as accurate labeling based on chaotic qualitative analyses and confirmatory annotations using expert eyes. To this end, chaotic features such as Petrosian fractal dimension, largest Lyapunov exponent and Hurst exponent were used in this investigation. Such features were submitted to the Bayesian classifier in order to have different categories seperated. Results: Our findings confirmed a chaotic behavior in EEG with minimum embedding dimension reduced in ictal state. Similarly, the complexity of the ictal state was reduced. In addition, our results indicated an average classification accuracy of 99.2% for normal vs. pre-ictal states; the average classification accuracy is 99.7% for the normal vs. ictal states and the average classification accuracy is 97.1% for the pre-ictal vs. ictal states. Conclusion: Chaotic analysis appears to serve as a useful method in representation and recognition of the brain activities in epileptic states. CP - IRAN IN - LG - eng PB - icss PG - 43 PT - Research YR - 2015