Psychophysiology, 2012, S28. [Paper presented at the 65th Annual Meeting of the Society of Psychophysiological Research (SPR) in New Orleans, LA, September 19-23, 2012.]
Incidence of electrode bridges in publicly available EEG data: An exploratory survey
Daniel M. Alschuler1, Craig E. Tenke1,2, Gerard E. Bruder1,2, Jürgen Kayser1,2
1 Division of Cognitive Neuroscience, New York State Psychiatric Institute, New York, NY
2 Department of Psychiatry, Columbia University College of Physicians and Surgeons, New York, NY
Abstract Electrode bridging in multichannel EEG montages may result from electrolyte spreading between adjacent electrode sites or from physical damage to wires or connectors. This sensor bridging can result in an uncontrolled and possibly systematic distortion of EEG topography, but its prevalence in EEG and ERP data is unknown. We examined several publicly-available EEG datasets for evidence of bridging using a simple screening method (Tenke & Kayser, 2001). Continuous EEG data with a minimum of 20 scalp sites and 15 60-s recording intervals were analyzed, and epochs (1 s) were artifacted and averaged. To screen for bridged channels, the electrical distance (temporal variance of all pairwise difference waveforms) was used to identify and quantify the similarity of the nearest electrical neighbor (intrinsic Hjorth). Bridged channels were identified visually as having essentially flat difference waveforms. Bridging was expressed as the percentage of the total number of channels in the montage. The mean percentage of bridged channels across the four datasets was 9.3 ± 15.5% (M ±SD%; A: 9.0 ±9.6; B: 25.0 ±27.1; C, D: no bridging found in any recording interval). Given that electrical bridging was prevalent in half of the examined EEG datasets, which were acquired at different laboratories, with different equipment, and from different populations, electrode bridging is probably more widespread than commonly thought. Key Words: electrolyte bridging; public dataset; high-density EEG |