Temporal and/or spectral characteristics of the EEG recordings that are useful for classification and/or pattern recognition.
Published in Chapter:
A Review of Automated Methodologies for the Detection of Epileptic Episodes Using Long-Term EEG Signals
Kostas M. Tsiouris (National Technical University of Athens, Greece), Alexandros T. Tzallas (Technological Educational Institute of Epirus, Greece), Sofia Markoula (University Hospital of Ioannina, Greece), Dimitris Koutsouris (National Technical University of Athens, Greece), Spiros Konitsiotis (University of Ioannina, Greece), and Dimitrios I. Fotiadis (University of Ioannina, Greece & Institute of Molecular Biology and Biotechnology, Greece)
Copyright: © 2016
|Pages: 31
DOI: 10.4018/978-1-4666-8828-5.ch011
Abstract
Epilepsy is a chronic neurological condition caused by abnormal electrical activity of the human brain that affects up to 1% of the global population. Since seizures may occur at any time, long-term EEG recordings are more suitable to record ictal activity. To assist the inspection process various automated seizure detection methodologies have been reported demonstrating high performance. In this chapter the majority of such long-term EEG signal processing techniques and methods, used in the seizure detection domain, are presented. Emphasis is particularly given on providing a complete overview of the wide variety of methodologies from the last few years, which were evaluated using two well-known public EEG databases consisting of long-term scalp and intracranial EEG recordings. The purpose of this chapter is to provide an evaluation of the methods' performance under a common reference dataset to assess their suitability for implantable or ambulatory seizure detection devices.