Iterative Subspace Decomposition for Ocular Artifact Removal from EEG Recordings

In this study, we present a method to remove ocular artifacts from electroencephalographic (EEG) recordings. This method is based on the detection of the EOG activation periods from a reference EOG channel, definition of covariance matrices containing the nonstationary information of the EOG, and applying generalized eigenvalue decomposition (GEVD) onto these matrices to rank the components in order of resemblance with the EOG. An iterative procedure is further proposed to remove the EOG components in a deflation fashion.

• Type: Conference
• Year: 2009
• Where: IDA
• Authors: Cédric Gouy-Pailler, Reza Sameni, Marco Congedo, Christian Jutten
• Keywords: Artificial Intelligence, EOG Activation Periods, EOG Components, IDA 2009, Reference Eog Channel