Fuzzy techniques with aggregation operators for classification and optimization of epilepsy risk level from EEG signals

The purpose of this paper is to develop a fuzzy classification model for epilepsy risk level analysis from EEG signals. The parametric values such as energy, variance, and duration, covariance, positive and negative peaks, sharp and spike waves and events are derived in each epoch of two second duration in the EEG signal channels. Fuzzy techniques are used to classify the risk level in each epoch for all the channels, the risk level patterns obtained are found to have low values of sensitivity, specificity, performance index and quality value. In order to increase the classification rate, an optimization technique based on aggregation operators is used and a quality value of 23.78 is achieved when compared to the value 6.25 achieved in the previous case. A comparison of fuzzy techniques without and with optimization is studied. The focal epilepsy problem in normal fuzzy classification is solved using this new approach. A group of ten patients with known epilepsy findings are used in this study. Further research work can be carried out in the classification of epilepsy risk level of a long-term EEG signals about 3 minute's durations. The number of samples may be increased to improve the classification rate.