Original title: Identification of epileptic activity in electroencephalograms using four techniques of independent component analysis
Translated title: Separace epilepticke aktivity v zaznamech elektroencefalografu pomoci ctyr metod analyzy nezavislych komponent.
Authors: Tichavský, Petr ; Nielsen, Jan ; Krajča, V.
Document type: Papers
Conference/Event: Biosignal 2006, Brno (CZ), 2006-06-28 / 2006-06-30
Year: 2006
Language: eng
Abstract: [eng] [cze]
The presented study aims to evaluate possibility of separation of epileptic activity from the EEG data using two well known and two recently proposed algorithms for independent component analysis (ICA): FastICA, EFICA, SOBI and WASOBI. All these techniques are shown to allow to concentrate an epileptic activityin two epilepsy-related independent components out of 19 channel EEG recordings. Among the techniques, the WASOBI was shown to be a most effective one. V clanku je zkoumana moznost separace epilepticke aktivity v EEG zaznamech pomoci dvou klasickych a dvou nedavno navrzenych metod slepe separace: FastICA, EFICA, SOBI a WASOBI. Tyto metody umoznuji pri zpracovani 19-kanaloveho EEG zaznamu epileptickou aktivitu vice-mene uspesne koncentrovat ve dvou signalovych komponetach. Mezi temito metodami se algoritmus WASOBI jevi jako ten ktery umoznuje nejpresnejsi separaci.
Keywords: blind source separation; electroencephalogram; epilepsy; ictus; independent component analysis
Project no.: CEZ:AV0Z10750506 (CEP), 1M0572 (CEP), 1ET101210512 (CEP)
Funding provider: GA MŠk, GA AV ČR
Host item entry: Analysis of Biomedical Signals and Images. BIOSIGNAL 2006, ISBN 80-214-3152-0

Institution: Institute of Information Theory and Automation AS ČR (web)
Document availability information: Fulltext is available at the institute of the Academy of Sciences.
Original record: http://hdl.handle.net/11104/0133890

Permalink: http://www.nusl.cz/ntk/nusl-35446


The record appears in these collections:
Research > Institutes ASCR > Institute of Information Theory and Automation
Conference materials > Papers