Robust analysis of dynamical change in scalp electroencephalogram time series

点击翻译按钮获取中文摘要

In recent years, there have been exciting developments in our understanding of brain functions. However, many problems remain unsolved. Among them is dynamical analysis of the brain that is one of the biggest challenges in science. This analysis helps in understanding, predicting and recognizing brain functions. Scalp electroencephalogram (EEG) time series is a non-invasive recording that reflects physiological states of brain activities. Analysis of dynamical change in EEG time series can be extremely useful in many research and diagnosis applications, such as detecting different sleep stages and pathologic states including epileptic seizure, psychiatric disorders, Alzheimer's, Parkinson's and Creutzfeldt-Jakob's diseases.The objective of this research is to develop a reliable and robust method for EEG analysis in the presence of noise. The method will establish a measure to quantify the degree of complex behavior of the brain based on dynamical change in the components of EEG. Our novel method is based on three steps: decomposition, component characterization and fusion. We will tune and apply the developed algorithms to EEG recorded from patients with epilepsy in order to predict seizure, a physical condition (neurological disorder) which causes sudden bursts of hyperactivity in the brain. It is estimated that 1% of the world population have had, or will have, some form of seizure disorder during their lifetime. From patient's point of view, one of the most disabling aspects of epilepsy is its unpredictability. It impairs patient's quality of life including restrictions on activities, management of medications, safety, and anxiety. Therefore, seizure prediction can improve the quality of life for patients with epilepsy.