EEG signal processing for automated epilepsy detection

• المؤلف الرئيسي: Sridharan Srividya
• مؤلفون آخرون: Justin Dauwels
• التنسيق: Theses and Dissertations
• اللغة: English
• منشور في: 2018
• الموضوعات: DRNTU::Engineering::Electrical and electronic engineering
• الوصول للمادة أونلاين: http://hdl.handle.net/10356/73126
• المؤسسة: Nanyang Technological University

Epilepsy is regarded as one among the common neurological disorders accompanied by recurring and sudden episodes of disturbances in sensory activities of brain. Researchers are still working to discover the regions of seizure onset in human brain in order to formulate new methods for effective diagnosis and quick treatment of epilepsy. Although a lot of studies have been conducted to localize the seizures, the existing methods are found to be erroneous in many cases due to inaccurate identification of seizure onset areas. Hence, novel mathematical techniques have been developed to statistically analyze the hyper synchrony of the brain regions. Having briefed the problems, this dissertation focuses on two statistical methods to deduce regions of epileptic activities in the brain. This work subjects the EEG signals to Granger causality and Transfer Entropy analysis to identify the brain regions with high functional linkages in the epileptic patients. The entire analysis is carried out by converting the brain regions into networks using graph theoretical approach. Finally, the results are compared to validate the network formulation and localization of seizures in the brain.