Current Proceedings on Technology

Yazarlar: Yılmaz Kaya, Ramazan Tekin, Necmettin Sezgin, M. Emin Tagluk

Konular:-

Anahtar Kelimeler:EEG,Epilepsy,Rough set theory,Classification

Özet: The detection and classification of epileptic Electroencephalogram (EEG) are significant in the treatment of patients with epilepsy. This paper concerns EEG classification based on Rough Set Theory (RST). RST is an efficient method for reduction of attributes and creation of decision rules, which have successfully been applied in various medical areas. Decision making was performed in 3 stages: the first stage is feature extraction by using discrete wavelet transformation (DWT) and statistical means. The next stage is discretization of these features and the last stage is classification using the classifiers trained on the discretizated features. In this study, RST was implemented for classification of five types of EEG data sets (A, B, C, D and E), each of which with 100 data rows. A and B were obtained from healthy volunteers while their eyes are open and closed respectively. C and D were obtained from epileptogenic zone and hippocampal region across hemisphere of patients prior to surgery respectively. E is obtained from patients during seizures. A few combinations of data samples for testing the performance of RST were used. The experimental results showed that this technique can acquire a high accuracy in classification of epileptic EEG signals. For various combinations of epileptic EEG signals the accuracy was ranged from 98.8% to 100%.

---