Abstract
A dependable means of describing the electroencephalograph (EEG) is going to have to overcome many of the complexities that commonly plague analysis of biological systems. Here we cover various attempts at developing a reliable method for automated detection of seizure events in EEG using nonLinear dynamics. Such techniques include Correlation Integral (CI), and Approximate Entropy (ApEn). ApEn algorithm is proposed as a new measure of complexity in the neural activity in scalp EEG data. It is demonstrated that ApEn is considerably less sensitive to signal noise, and amplitude variations and remains a more robust measure of complexity under normalization of EEG signals. That ApEn has proven quite promising for distinguishing the amplitude increasing artifacts such as eye movements from seizure activity is demonstrated.Epilepsy centers like the Comprehensive Epilepsy Center of Miami Childrens Hospital are facing a critical need for software suitable for handling the large amounts of patient data they are presented with. As a solution to this medical need, we present software with both a research and clinical focus, that allows for detection, analysis, and storage of seizure events in EEG using methods, not limited to those described here. The software, the Application for Seizure Detection, is web-based and facilitates the sharing and communication of patient data and results throughout the entire authorized medical community. Together with methods used in nonlinear dynamics and the software to support them we present a comprehensive solution for EEG analysis.