Psychophysiology, 43:S51, 2006.

Electrical distance as a reference-free measure for identifying artifacts in multichannel electroencephalogram (EEG) recordings

Jürgen Kayser ^a,b, Craig E. Tenke ^a,b

^a Department of Biopsychology, New York State Psychiatric Institute, New York, New York, USA
^b Department of Psychiatry, Columbia University College of Physicians & Surgeons, New York, New York, USA

Abstract

Multichannel EEG, with hundreds of scalp placements now common, has increased the likelihood of recording artifacts in one or more channels within any given time epoch. Reliable artifact detection is mandatory because of distortion in signal topography. This becomes increasingly difficult as the number of sites and epochs increases. A systematic statistical approach for dense electrode arrays uses threshold criteria (amplitude, etc.) and their observed distributions to identify artifacts as median-based deviations (Junghöfer et al., 2000). Whereas this elegant approach eliminates the subjectivity of visual artifact screening, it is based on criteria requiring the choice of a particular recording or offline reference. Moreover, if the reference is contaminated, an artifact will be "detected" in all other channels. These problems can be circumvented and simplified with a reference-free electrical distance measure, which quantifies signal similarity through variances of waveform differences for all pairwise combinations within a given montage. Because spatial (recording sites) and temporal (sample points) proximity warrants highly intercorrelated surface potentials due to volume conduction, a low signal similarity at nearby electrodes strongly implies an unrealistic deviance (i.e., artifact). Electrical distance frequency distributions, both within and across individuals, thereby provide an easy, objective, and reference-free approach to identify recording artifacts, including those affecting the reference site. Implementations are shown for 31-, 67- and 129-channel montages.

Keywords: EEG, ERP, artifact, reference-free, recording reference, dense electrode array, electrical distance, methodology