TSALLIS ENTROPY BASED SEIZURE DETECTION

This paper presents EEG signal analysis using Tsallis entropy and then it will make available for comparison with any another method along with KNN classification. Electroencephalogram (EEG) remains the most immediate, easy and rich source of information for accepting phenomena related to brain electrical activities[1]. Important information, about the state of patient under observation, must be extracted from calculated DSD (Decimated signal diagonalization) bispectrum[2]. For this aim, it is useful to delineate an assessment index about the dynamic process associated with the analysedsignal. This information is measure by means of entropy, since the degree of order or disorder of the recorded EEG signal will be replicated in the obtained DSDbispectrum[3]. Tsallis entropy is better than Shannon one because it maximizes the probabilities of the events of the interest through the selection of the entropic index, and so it permits to detect in more perfect way, spikes relatedto epileptic seizure.Then, the signals are classified using K Nearest Neighbourclassifie