Removal of ballistocardiogram artifacts from simultaneously recorded EEG and fMRI data using independent component analysis

Wakako Nakamura, Kimitaka Anami, Takeyuki Mori, Osamu Saitoh, Andrzej Cichocki, Shun Ichi Amari

Research output: Contribution to journalArticlepeer-review

76 Citations (Scopus)

Abstract

Simultaneous recording of electroencephalogram (EEG) and functional magnetic resonance imaging (fMRI) has been studied to identify areas related to EEG events. EEG data recorded in the magnetic resonance (MR) scanner with MR imaging is suffered from two specific artifacts, imaging artifact, and ballistocardiogram (BCG). In this paper, we focus on BCG. In preceding studies, average subtraction was often used for this purpose. However, average subtraction requires an assumption that BCG waveforms are precisely periodic, which seems unrealistic because BCG is a biomedical artifact. We propose the application of independent component analysis (ICA) with a postprocessing of high-pass filtering for the removal of BCG. With this approach, it is not necessary to assume that the BCG waveform is periodic. Empirically, we show that our proposed method removes BCG artifacts as well as does the average subtraction method. Power spectral density analysis of the two approaches shows that, with ICA, distortion of recovered EEG data is also as small as that associated with the average subtraction approach. We also propose a hypothesis for how head movement causes BCGs and show why ICA can remove BCG artifacts arising from this source.

Original languageEnglish
Article number1643399
Pages (from-to)1294-1308
Number of pages15
JournalIEEE Transactions on Biomedical Engineering
Volume53
Issue number7
DOIs
Publication statusPublished - Jul 2006
Externally publishedYes

Keywords

  • Biomedical signal processing
  • Electroencephalography
  • Functional magnetic resonance imaging
  • Independent component analysis

Fingerprint

Dive into the research topics of 'Removal of ballistocardiogram artifacts from simultaneously recorded EEG and fMRI data using independent component analysis'. Together they form a unique fingerprint.

Cite this