Automatic and semi-automatic artifact removal methods in EGG and ET coregistration

TORRES BATÁN S; KAMIENKOWSKI JE

Congreso; Latin American Brain Mapping Network (LABMAN); 2017

Latin American Brain Mapping Network (LABMAN) - Organization for Human Brain Mapping

Most of the knowledge of brain basis of several cognitive processes in humans has been studied using foveated stimuli, and thus restricting eye movements. However, this is rarely the case in any human behavior, who tend to move their eyes to explore an image or to read, and have to make an effort to keep their eyes still. Recently, some co-registration studies had started closing this gap and moving towards more ecological tasks. One of the major challenges is to cope with the eye movement artifacts, which are usually several times larger than expected neural effects. Independent Component Analysis is one of the most popular and effective methods for source separation and artifact removal in EEG. However, the identification of artifactual independent components (ICs) still generally relies on an expert?s subjective decision. It have been proposed several algorithms to identify artifactual ICs semi-automatically, such as EyeCatch, which generates a classifier from a large ICs topologies database (Bigdely-Shamlo, et al., 2013); ADJUST, which uses spatial and spectral features (Mognon, et al., 2011); or a method based on comparing variance of ICs pre- and post-saccade (Plöchl, et al., 2012). Albeit, these have never been systematically compared.In the present work, we set up a co-registration experiment in which participants performed large saccades and blinks, combined with foveated visual stimulation. We aimed to compare several of these state-of-the-art methods and some new variants, such as studying the correlation of ICs and eye tracking data, or combinations of these criteria. Moreover, the proposed procedure uses brain responses to foveated visual stimuli to evaluate how selective to artifacts these methods are.We will embed these functions into a framework compatible with EEGLAB and EEG-EYE plugin, as it was meant as a new step towards the spread of these more ecological experimental approaches.