Publication detail

Does EEG help to identify the epilepsy-related spatial independent component of fMRI data?

LAMOŠ, M. MAREČEK, R. SLAVÍČEK, T. HAVLÍČEK, M. JAN, J.

Original Title

Does EEG help to identify the epilepsy-related spatial independent component of fMRI data?

Czech Title

Může EEG pomoci identifikovat prostorové nezávislé komponenty svázané s epilepsií v fMRI datech?

English Title

Does EEG help to identify the epilepsy-related spatial independent component of fMRI data?

Type

abstract

Language

en

Original Abstract

Independent component analysis (ICA) has advantage of being a completely data-driven approach. However, no idea on the importance of a particular independent component (IC) is available without some prior knowledge about processes that we are looking for [Hyvärinen and Oja, 2000]. A step preceding ICA is data reduction by principal component analysis (PCA). Effects of interest (here the interictal epileptic activity) may be minor compared to other processes in the brain and thus we can remove them accidentally. However, in our retrospective study [Slavíček et al., submitted for publication] we showed that ICA with proper settings applied to fMRI data is capable of finding ICs corresponding to epileptic activity. This contribution aims at the next step – identification of features, which would enable discovering epilepsy related fMRI ICs prospectively.

Czech abstract

Independent component analysis (ICA) has advantage of being a completely data-driven approach. However, no idea on the importance of a particular independent component (IC) is available without some prior knowledge about processes that we are looking for [Hyvärinen and Oja, 2000]. A step preceding ICA is data reduction by principal component analysis (PCA). Effects of interest (here the interictal epileptic activity) may be minor compared to other processes in the brain and thus we can remove them accidentally. However, in our retrospective study [Slavíček et al., submitted for publication] we showed that ICA with proper settings applied to fMRI data is capable of finding ICs corresponding to epileptic activity. This contribution aims at the next step – identification of features, which would enable discovering epilepsy related fMRI ICs prospectively.

English abstract

Independent component analysis (ICA) has advantage of being a completely data-driven approach. However, no idea on the importance of a particular independent component (IC) is available without some prior knowledge about processes that we are looking for [Hyvärinen and Oja, 2000]. A step preceding ICA is data reduction by principal component analysis (PCA). Effects of interest (here the interictal epileptic activity) may be minor compared to other processes in the brain and thus we can remove them accidentally. However, in our retrospective study [Slavíček et al., submitted for publication] we showed that ICA with proper settings applied to fMRI data is capable of finding ICs corresponding to epileptic activity. This contribution aims at the next step – identification of features, which would enable discovering epilepsy related fMRI ICs prospectively.

Keywords

fMRI, EEG, epilepsy, ICA

Released

12.06.2014

Pages count

3

URL

BibTex


@misc{BUT109017,
  author="Martin {Lamoš} and Radek {Mareček} and Tomáš {Slavíček} and Martin {Havlíček} and Jiří {Jan}",
  title="Does EEG help to identify the epilepsy-related spatial independent component of fMRI data?",
  annote="Independent component analysis (ICA) has advantage of being a completely data-driven approach. However, no idea on the importance of a particular independent component (IC) is available without some prior knowledge about processes that we are looking for [Hyvärinen and Oja, 2000]. A step preceding ICA is data reduction by principal component analysis (PCA). Effects of interest (here the interictal epileptic activity) may be minor compared to other processes in the brain and thus we can remove them accidentally. However, in our retrospective study [Slavíček et al., submitted for publication] we showed that ICA with proper settings applied to fMRI data is capable of finding ICs corresponding to epileptic activity. This contribution aims at the next step – identification of features, which would enable discovering epilepsy related fMRI ICs prospectively.",
  booktitle="20th Annual Meeting of the Organization for Human Brain Mapping (OHBM), 2014.",
  chapter="109017",
  howpublished="online",
  year="2014",
  month="june",
  type="abstract"
}