Wavelet-based fractal analysis of the epileptic EEG signal

The wavelet transform is a natural tool for characterizing self-similar signals. In this work, the spectral exponent ¿ derived from the wavelet-based representation for 1/f processes is used to investigate the self-similarity of electrocorticography (intracranial EEG) signals from an epilepsy patient. An increase in ¿ leads to sample signals with smoother temporal patterns. Our computational results show that during an epileptic seizure ¿ is significantly higher than that associated with other states of the brain, implying that wavelet-based fractal analysis is potentially a useful computational tool for epileptic seizure detection.