Proceedings of The Physiological Society

Physiology 2012 (Edinburgh) (2012) Proc Physiol Soc 27, PC251

Poster Communications

Source reconstruction of MEG human brain responses to retinotopic visual stimuli.

N. Cicmil1, H. Bridge2, A. J. Parker1, M. W. Woolrich3,2, K. Krug1

1. Dept. Physiology, Anatomy and Genetics, University of Oxford, Oxford, United Kingdom. 2. Oxford Centre for Functional MRI of the Brain (FMRIB), University of Oxford, Oxford, United Kingdom. 3. Oxford Centre for Human Brain Activity (OHBA), University of Oxford, Oxford, United Kingdom.


Retinotopic maps are an established feature of early visual areas and are useful for evaluating the source localization of neuroimaging techniques. Magnetoencephalography (MEG) is a physiological technique that allows recording of human brain activity at high temporal resolution. However, the accurate localization of brain sources of the MEG signal remains a challenge, as the signal alone cannot constrain a unique solution without additional assumptions. We evaluated three MEG source reconstruction methods by comparing retinotopic activity obtained with MEG with retinotopic maps obtained with functional magnetic resonance imaging (fMRI). Six participants placed in a MEG scanner (Elekta Neuromag) viewed black/white checkerboard stimuli, located in visual field quadrants and annular rings to map angular and eccentric locations respectively. Brain sources were reconstructed with three contrasting approaches: i) an assumption of multiple sparse priors (MSP); ii) a beamformer (spatial filter); and iii) a minimum norm estimate (MNE) that assumes independently distributed sources of minimal amplitude. We calculated the percentage of MEG sources localized to the subregions of early visual areas V1, V2 and V3, defined retinotopically with fMRI in the same individuals. For quadrants, all reconstruction methods localized sources better than chance level (p < 0.05, t-test), but the MNE approach was significantly more accurate than the beamformer or MSP (F = 7.3, p < 0.01, ANOVA). For complete rings, no reconstruction method localized activity better than chance. However, when ring stimuli were shown in quarters, eccentricity of lower field stimuli was accurately represented by MNE and beamformer but not MSP. Our results suggest that MNE and beamformer approaches are the most suitable for localizing early visual activity, but that there are limitations to using MEG to localize stimuli that transect quadrant boundaries. This may be due to the folding and orientation of the cortical surface of early visual areas.

Where applicable, experiments conform with Society ethical requirements