FEATURE EXTRACTION FROM INTERICTAL EPILEPTIC AND NON- EPILEPTIC PATHOLOGICAL EEG EVENTS FOR DIAGNOSTIC PURPOSES USING LVQ1 NEURAL NETWORK

• Published in: Mathematical Methods In Scattering Theory And Biomedical Engineering pp. 390 - 398
• Main Authors: PAPAVLASOPOULOS, S., POULOS, M., EVANGELOU, A.
• Format: Book Chapter
• Language: English
• Published: WORLD SCIENTIFIC 01.08.2006
• Subjects: Biomedical Engineering
• ISBN: 9812568603; 9812773193; 9789812773197; 9789814477598; 9814477591; 9789812568601
• DOI: 10.1142/9789812773197_0038

PURPOSE: In this study, the problem of discriminating between interictal epileptic and non-epileptic pathological EEG cases, which present episodic loss of consciousness, investigated. The proposed method produces cross-correlated coefficients from an appropriate, selected EEG segment. The characteristic features of these coefficients based on a prototype spectral variation of an extended spectrum in which alpha, beta and gamma activities are included. For the diagnostic test of the discrimination between these two EEGs categories, we used a suitable constructed artificial supervised LVQ1 neural network. RESULTS: Two (2) data, types recorded.,.42 interictal epileptic EEGs from diagnosed epileptic individuals and 44 EEGs from diagnosed pathological cases, who had presented loss of consciousness, recorded. The results showed that the LVQ1 neural network classified correctly the two categories in 86 percent. CONCLUSIONS: In this experiment, we ascertained that the LVQ1 neural network classified better the above categories in the frequency 8-40 Hz than our previous study.