Examination of Scale-Invariant Characteristics of Multi-channel ECoG Data for Epileptic Seizure Localization

Abstract

One of the most important tasks for epilepsy surgery is to localize a region of the brain involved with the development of epileptic seizure. This study investigates the scale-invariant behavior of brain activity for evidence of epileptic seizure localization. Multi-channel electrocorticogram (ECoG) data of a subject diagnosed with right mesial temporal lobe epilepsy obtained from various regions of the brain and associated with different states of the brain are examined using the wavelet-based fractal analysis. The spectral exponent computed from such analysis is used as a measure for quantifying scale-invariant behavior. The computational results show that the ECoG data corresponding to various regions of the brain and also different states of the brain exhibit distinguishable characteristics of the spectral exponent. Furthermore, a significant increase in the spectral exponent is observed only in the ECoG channels associated with an epileptic seizure event (inside the epileptogenic zone) during the ictal state. The computational results provide preliminary evidence that an epileptic seizure can be both temporally and spatially localized using scale-invariant characteristics.