Independent component decomposition of around ear EEG data to detect artifacts

Electroencephalography (EEG) allows monitoring people’s brain activity. Beyond the lab EEG recordings in everyday situations for extended periods of time are possible using mobile EEG hardware, concealed electrode positioning around the ears and smartphone-based signal acquisition. Given the susceptibility of EEG to non-brain signal contributions, the interpretation of EEG recordings acquired in these situations is challenging. To avoid misinterpretation of the recorded signal a clear separating of brain and non-brain signal contributions is necessary. Here, we study how different non-brain signal sources contribute to the signal as recorded by ear electrodes (cEEGrid). In a concurrent cap and ear-EEG recording, we first identified different artifacts on the cap data using wellestablished procedures based on independent component analysis (ICA). Using the corresponding artifact time courses we searched for these artifact profiles in the ICA decomposition of the cEEGrid data. For eye-blinks, lateral eye movements and the heartbeat we could identify a matching ICA component in the cEEGrid data. Similar to cap ICA components the cEEGrid ICA components have a characteristic topographic representation. Our results give an indication that automatized artifact detection based on ICA decomposition, well established for cap EEG, may also be feasible for cEEGrid ear-EEG.