Pattern reversal visual evoked responses of V1/V2 and V5/MT as revealed by MEG combined with probabilistic cytoarchitectonic maps

Abstract

Pattern reversal stimulation provides an established tool for assessing the integrity of the visual pathway and for studying early visual processing. Numerous magnetoencephalographic (MEG) and electroencephalographic (EEG) studies have revealed a three-phasic waveform of the averaged pattern reversal visual evoked potential/magnetic field, with components N75(m), P100(m), and N145(m). However, the anatomical assignment of these components to distinct cortical generators is still a matter of debate, which has inter alia connected with considerable interindividual variations of the human striate and extrastriate cortex. The anatomical variability can be compensated for by means of probabilistic cytoarchitectonic maps, which are three-dimensional maps obtained by an observer-independent statistical mapping in a sample of ten postmortem brains. Transformed onto a subject's brain under consideration, these maps provide the probability with which a given voxel of the subject's brain belongs to a particular cytoarchitectonic area. We optimize the spatial selectivity of the probability maps for V1 and V2 with a probability threshold which optimizes the self- vs. cross-overlap in the population of postmortem brains used for deriving the probabilistic cytoarchitectonic maps. For the first time, we use probabilistic cytoarchitectonic maps of visual cortical areas in order to anatomically identify active cortical generators underlying pattern reversal visual evoked magnetic fields as revealed by MEG. The generators are determined with magnetic field tomography (MFT), which reconstructs the current source density in each voxel. In all seven subjects, our approach reveals generators in V1/V2 (with a greater overlap with V1) and in V5 unilaterally (right V5 in three subjects, left V5 in four subjects) and consistent time courses of their stimulus-locked activations, with three peak activations in V1/V2 (contributing to C1m/N75m, P100m, and N145m) and two peak activations in V5 (contributing to P100m and N145m). The reverberating V1/V2 and V5 activations demonstrate the effect of recurrent activation mechanisms including V1 and extrastriate areas and/or corticofugal feedback loops. Our results demonstrate that the combined investigation of MEG signals with MFT and probabilistic cytoarchitectonic maps significantly improves the anatomical identification of active brain areas.