Multivariate analysis of ERP: a spatial ERP component extraction method and its multisenory ERP analysis application

摘要

The present study addressed a new Event-related Potential (ERP) spatial component extraction method. Using the average of single trial data upon specific time segment as samples, electrodes as variates, first spatial Principle Component Analysis (PCA) is performed. Then amplitude of each spatial component loading over the time segment is to be estimated by Multiple Regression Analysis (MRA). Thus a new ERP spatial component extraction method is formulated. With data obtained from a newly devised Stimulus Onset Asynchrony (SOA) detection multisensory ERP experiment, the method was tested and comparison between the spatial component topographies and corresponding grand mean mappings along the time course was conducted. The consistency of analysis results with former studies demonstrated its effectiveness and usefulness. This new ERP spatial component extraction method provides information of spatial components for voltage topography mapping, which is never provided and cannot be provided by any ERP study with high-density electrode array before.