Topography estimation of visual evoked potential by combinational use of mathematical models

To analyze the topographical distribution of visual evoked potentials (VEPs) in electroencephalographic (EEG) record brings about the significant information for visual processing in human brain. This study proposes a method for estimating topographical distribution of VEP from the separated power spectrum components by combining the models. VEPs with various temporal frequencies were recorded from nine healthy adults. Original power spectrum consisting of VEP, background activities, artifacts and other components were then obtained. To the extract the VEP components from the original power spectrum, models corresponding to background activities especially for posterior dominant alpha rhythm, low frequency component and high frequency component caused from electromyographic (EMG) artifact were constructed and those parameters were estimated. Finally, VEP components were calculated by subtracting those components from the original power spectrum. Topographical distribution of both first harmonic (1F) and second harmonic (2F) components of VEP were obtained by the proposed method. Estimated other components except VEPs were also investigated. Merits and usefulness of the proposed method were analyzed by comparing the conventional stimulus-locked averaging method in the time domain. Proposed method has several advanced points for conventional averaging method. Particularly, posterior dominant alpha rhythm and EMG artifact were directly taken into account in the VEP components estimation. Therefore, accurate estimation of VEP components can be done even if those components were contaminated with raw EEG. To obtain the topographical distribution of 1F and 2F components respectively is also possible.