FRACTAL ANALYSIS OF EPILEPSY

Abstract

A complex system has constituents interacting in a nonlinear way. A complex biological system is the human brain, which function is based on the communication between neurons, if there is a mistake in the communication between them, it can cause epilepsy. Epilepsy affects from 1% to 2% worldwide population. Fractal Geometry is used to analyze electroencephalograms (EEG) from epileptic patients to determine where the epileptogenic region is located to make a surgery to the epileptic patient. In this work fractal geometry is applied (using four self-affine trace methods: R/S analysis, Roughness-Length, Variogram, and Wavelets), to study temporal series generated by EEG from data that refers to people that do not have epilepsy but have had a neurological problem and epileptic patients. After doing the fractal analysis, it is concluded that for all complex signals (EGG) under studying, the best methods to analyze epilepsy are R/S and Variogram, because they were the solidest to every analyzed channel, in addition to be the nearest to the average values from the four self-affine trace methods.