High-resolution EEG techniques for brain-computer interface applications

Febo Cincotti, Donatella Mattia, Fabio Aloise, Simona Bufalari, Laura Astolfi, Fabrizio De Vico Fallani, Andrea Tocci, Luigi Bianchi, Maria Grazia Marciani, Shangkai Gao, Jose Millan, Fabio Babiloni

Research output: Contribution to journalArticle

Abstract

High-resolution electroencephalographic (HREEG) techniques allow estimation of cortical activity based on non-invasive scalp potential measurements, using appropriate models of volume conduction and of neuroelectrical sources. In this study we propose an application of this body of technologies, originally developed to obtain functional images of the brain's electrical activity, in the context of brain-computer interfaces (BCI). Our working hypothesis predicted that, since HREEG pre-processing removes spatial correlation introduced by current conduction in the head structures, by providing the BCI with waveforms that are mostly due to the unmixed activity of a small cortical region, a more reliable classification would be obtained, at least when the activity to detect has a limited generator, which is the case in motor related tasks. HREEG techniques employed in this study rely on (i) individual head models derived from anatomical magnetic resonance images, (ii) distributed source model, composed of a layer of current dipoles, geometrically constrained to the cortical mantle, (iii) depth-weighted minimum L2-norm constraint and Tikhonov regularization for linear inverse problem solution and (iv) estimation of electrical activity in cortical regions of interest corresponding to relevant Brodmann areas. Six subjects were trained to learn self modulation of sensorimotor EEG rhythms, related to the imagination of limb movements. Off-line EEG data was used to estimate waveforms of cortical activity (cortical current density, CCD) on selected regions of interest. CCD waveforms were fed into the BCI computational pipeline as an alternative to raw EEG signals; spectral features are evaluated through statistical tests (r2 analysis), to quantify their reliability for BCI control. These results are compared, within subjects, to analogous results obtained without HREEG techniques. The processing procedure was designed in such a way that computations could be split into a setup phase (which includes most of the computational burden) and the actual EEG processing phase, which was limited to a single matrix multiplication. This separation allowed to make the procedure suitable for on-line utilization, and a pilot experiment was performed. Results show that lateralization of electrical activity, which is expected to be contralateral to the imagined movement, is more evident on the estimated CCDs than in the scalp potentials. CCDs produce a pattern of relevant spectral features that is more spatially focused, and has a higher statistical significance (EEG: 0.20 ± 0.114 S.D.; CCD: 0.55 ± 0.16 S.D.; p = 10-5). A pilot experiment showed that a trained subject could utilize voluntary modulation of estimated CCDs for accurate (eight targets) on-line control of a cursor. This study showed that it is practically feasible to utilize HREEG techniques for on-line operation of a BCI system; off-line analysis suggests that accuracy of BCI control is enhanced by the proposed method.

Original languageEnglish
Pages (from-to)31-42
Number of pages12
JournalJournal of Neuroscience Methods
Volume167
Issue number1
DOIs
Publication statusPublished - Jan 15 2008

Keywords

  • Assistive technologies
  • EEG-based brain-computer interfaces
  • High-resolution electroencephalography
  • Motor cortex
  • Neuroimaging

ASJC Scopus subject areas

  • Neuroscience(all)

Fingerprint Dive into the research topics of 'High-resolution EEG techniques for brain-computer interface applications'. Together they form a unique fingerprint.

  • Cite this

    Cincotti, F., Mattia, D., Aloise, F., Bufalari, S., Astolfi, L., De Vico Fallani, F., Tocci, A., Bianchi, L., Marciani, M. G., Gao, S., Millan, J., & Babiloni, F. (2008). High-resolution EEG techniques for brain-computer interface applications. Journal of Neuroscience Methods, 167(1), 31-42. https://doi.org/10.1016/j.jneumeth.2007.06.031