EOG correction: comparing different calibration methods, and determining the number of epochs required in a calibration average

Objective: A form of EOG correction called the 'aligned-artifact average' (AAA) method has been advanced by the authors. In contrast to many previous methods, this used a correction coefficient (B), based on an average of eye movements rather than raw data, to remove eye movement related contamination from the EEG. The first part of this study was aimed at determining whether a variation of this procedure that is more easily implemented would produce a similar correction. The second pan was designed to determine the number of epochs needed in an average to correct adequately. Methods: Subjects performed a series of eye movement tasks whilst EEG and EOG data were recorded. Data were manipulated according to either the AAA or an alternate new ERP method (NERP) and the resultant Bs were compared in part A. In part B, averages were created from varying numbers of epochs, and the resultant r(2)-values were compared. Results: The AAA and NERP methods produced the same Ss, and averages with at least 40 epochs were required for adequate B estimation for both VEOG and HEOG. Conclusion: There is no difference between the AAA and NERP methods and thus it is acceptable to use the more easily implemented NERP method for EOG correction. Lt is recommended that when applying this procedure, at least 40 epr,chs should be used to make up the averages from which to calculate correction coefficients. (C) 2000 Elsevier Science Ireland Ltd. All rights reserved.