High-resolution retinotopic maps estimated with magnetoencephalography
Introduction
Among the various methods for non-invasive imaging, magnetoencephalography (MEG) source imaging is known to provide outstanding temporal resolution, while it is typically assumed to have modest spatial resolution (Darvas et al., 2004, Hämäläinen et al., 1993). Consequently MEG imaging is most often used in experiments aimed at measuring temporal fluctuations in neural signals for which the assignment of a precise anatomical source is not critical. Although recent reports have raised the possibility of extracting rich spatial signals from MEG (Cichy et al., 2015), a quantitative estimate of the resolution that can be attained with this imaging modality is lacking.
Here we have examined the capacity of MEG to resolve the well-known retinotopic organization of the primary visual cortex (V1). This representation provides a useful benchmark, because it has been thoroughly and quantitatively characterized using a variety of methods, including electrophysiology (Das and Gilbert, 1995, Hubel and Wiesel, 1977), PET (Fox et al., 1987), optical imaging (White and Culver, 2010) and fMRI (Engel et al., 1997). These approaches have demonstrated a smooth change in the locus of cortical activation for corresponding changes in the position of the retinal stimulus (Dumoulin and Wandell, 2008, Engel et al., 1997, Sereno et al., 1995). Thus to the extent that an imaging modality has high spatial resolution, it should be able to differentiate responses to visual stimuli in different locations. The smallest shift in the locus of cortical activation that can be detected serves as a measure of resolution.
Here we have obtained retinotopic maps from human subjects using MEG in combination with a standard visual stimulation paradigm. We show that surprisingly high spatial resolution maps can be obtained with appropriate choices of visual stimulation and source modeling. In particular, we are able to reliably detect distinct MEG responses emanating from sources separated by 7.0 mm along smooth cortical surfaces and less than 1 mm along the arched gyri.
Section snippets
Participants
Data were recorded from two healthy, right-handed male subjects (one author, one naïve), both of whom had normal or corrected to normal vision. Both subjects gave written consent prior to participation in three sessions, involving structural MRI, functional MRI, and MEG recordings. All experimental protocols were approved by the Research Ethics Board of the Montreal Neurological Institute.
Structural MRI
For the MRI scans, each subject was positioned on his back with a 32 channel surface coil centered over the
Results
We studied the spatial properties of MEG signals emanating from the visual cortex in two human subjects. We elicited visual responses by presenting images comprised of a small number of squares flashed simultaneously on a computer monitor. In this section we analyze the relationship between the positions of individual stimulus squares and MEG source responses, as well as the distribution of these receptive fields across the cortical surface; we use these data to derive an estimate of the
Brief summary of results
In this study we have demonstrated the capacity of individual MEG sources to show selectivity for specific areas of the visual field. We showed that localized visual receptive fields for individual sources (Fig. 3) can be obtained from modest amounts of data (Appendix 3), and that the ensembles of these receptive fields form orderly maps within the occipital lobe (Fig. 4, Fig. 5). These maps are well matched to those obtained with fMRI (Fig. 7). Analysis of correlated responses between pairs of
Conclusion
MEG has traditionally been used in applications requiring excellent temporal resolution. However, its spatial resolution is most often considered to be coarse. We have shown that MEG can recover retinotopic maps with similar shape to those obtained with fMRI, and in some areas with comparable spatial resolution.
Acknowledgements
We would like to thank Elizabeth Bock for MEG training and help in data acquisition, François Tadel for valuable suggestions in using Brainstorm and Drs. Matthew Krause and Theodore Zanos for their comments on the data analysis and Dr. Reza Farivar-Mohseni for his help with fMRI surface rendering. This work was funded by a Molson Neuro-Engineering and a Gerry Sklavounos - MNA Laurier-Dorion Scholarship to K.N., and by a grant from NSERC (341534-12) to C.C.P. S.B. was supported by the Killam
References (84)
- et al.
Chapter 5 - MEG studies of visual processing
- et al.
Magnetic localization of a dipolar current source implanted in a sphere and a human cranium
Electroencephalogr. Clin. Neurophysiol.
(1986) - et al.
Investigating spatial specificity and data averaging in MEG
NeuroImage
(2010) - et al.
Tonotopic organization of human auditory association cortex
Brain Res.
(1994) - et al.
Can visual information encoded in cortical columns be decoded from magnetoencephalography data in humans?
NeuroImage
(2015) - et al.
Phase delays within visual cortex shape the response to steady-state visual stimulation
NeuroImage
(2011) - et al.
Cortical Surface-Based Analysis: I
Segm. Surface Reconstruct. NeuroImage
(1999) - et al.
Mapping human brain function with MEG and EEG: methods and validation
NeuroImage
(2004) - et al.
Receptive-field dynamics in the central visual pathways
Trends Neurosci.
(1995) - et al.
Population receptive field estimates in human visual cortex
NeuroImage
(2008)
Cortical magnification within human primary visual cortex correlates with acuity thresholds
Neuron
Cortical surface-based analysis: II: inflation, flattening, and a surface-based coordinate system
NeuroImage
Magnetoencephalographic investigation of human cortical area V1 using color stimuli
NeuroImage
Inflow effects on functional MRI
NeuroImage
Assessing and improving the spatial accuracy in MEG source localization by depth-weighted minimum-norm estimates
NeuroImage
Two-dimensional substructure of MT receptive fields
Neuron
Consistent and precise localization of brain activity in human primary visual cortex by MEG and fMRI
NeuroImage
Effects of check size on pattern reversal visual evoked magnetic field and potential
Brain Res.
Experimental analysis of distortion of magnetoencephalography signals by the skull
Clin. Neurophysiol.
Tonotopic organization of the human auditory cortex revealed by transient auditory evoked magnetic fields
Electroencephalogr. Clin. Neurophysiol.
The auditory evoked sustained field: origin and frequency dependence
Electroencephalogr. Clin. Neurophysiol.
Specific tonotopic organizations of different areas of the human auditory cortex revealed by simultaneous magnetic and electric recordings
Electroencephalogr. Clin. Neurophysiol.
Low resolution electromagnetic tomography: a new method for localizing electrical activity in the brain
Int. J. Psychophysiol.
Precise mapping of early visual responses in space and time
NeuroImage
Estimating linear cortical magnification in human primary visual cortex via dynamic programming
NeuroImage
Reverse correlation in neurophysiology
Cogn. Sci.
The advantage of combining MEG and EEG: comparison to fMRI in focally stimulated visual cortex
NeuroImage
Central versus peripheral visual field stimulation results in timing differences in dorsal stream sources as measured with MEG
Vision. Res
Signal-space projections of MEG data characterize both distributed and well-localized neuronal sources
Electroencephalogr. Clin. Neurophysiol.
Discrimination of cortical laminae using MEG
NeuroImage 102
The visual field representation in striate cortex of the macaque monkey: asymmetries, anisotropies, and individual variability
Vision. Res.
Visual field maps in human cortex
Neuron
Phase-encoded retinotopy as an evaluation of diffuse optical neuroimaging
NeuroImage
Electromagnetic brain mapping
IEEE Signal Process. Mag.
Triggered correlation
IEEE Trans. Biomed. Eng.
The psychophysics toolbox
Spat. Vis.
Processing of color, form and disparity information in visual areas VP and V2 of ventral extrastriate cortex in the macaque monkey
J. Neurosci.
Localization of MEG human brain responses to retinotopic visual stimuli with contrasting source reconstruction approaches
Brain Imaging Methods
Is the cortical deficit in amblyopia due to reduced cortical magnification, loss of neural resolution, or neural disorganization?
J. Neurosci.
Improved localizadon of cortical activity by combining EEG and MEG with MRI cortical surface reconstruction: a linear approach
J. Cogn. Neurosci.
The representation of the visual field on the cerebral cortex in monkeys
J. Physiol.
Cited by (24)
Spatiotemporal Precision of Neuroimaging in Psychiatry
2023, Biological PsychiatryDynamics of retinotopic spatial attention revealed by multifocal MEG
2022, NeuroImageCitation Excerpt :Cross-validated Mahalanobis-distance based multivariate analysis achieved reliable separation of signals originating from distinct visual field regions when spatial resolution of the more foveal regions was just around one degree of visual angle. In agreement with recent studies (Kupers et al., 2020; Nasiotis, Clavagnier, Baillet, & Pack, 2017), we found that it is possible to achieve spatial resolving ability comparable with fMRI by combining MEG with multivariate data analysis. Responses for upper visual field stimulation were generally weaker, but still reliably discriminable, unlike reported in (Nasiotis et al., 2017).
A visual encoding model links magnetoencephalography signals to neural synchrony in human cortex
2021, NeuroImageCitation Excerpt :This type of computational model is closest to our approach, but again, differs from our encoding model in that these forward models do not take visual stimuli as an input. Previous forward models have been used to create simulated sensor responses as a benchmark to test specific analyses methods, e.g., brain connectivity analyses in EEG (Haufe and Ewald, 2019), accuracy of volume conduction head models (Henson et al., 2009; Stenroos et al., 2014; Stenroos and Nummenmaa, 2016), guide subdural electrode placement for epilepsy monitoring (Lopes et al., 2020), or provide ‘ground truth’ sources when combined with inverse modeling in both healthy (Sharon et al., 2007; Akalin Acar and Makeig, 2013; Nasiotis et al., 2017) and patient populations (Akalin Acar et al., 2008). Some recent publications of open-source EEG/MEG toolboxes have now advanced to simulating electromagnetic fields with biologically plausible noise (Barzegaran et al., 2019) or cellular-level circuits (Neymotin et al., 2020).
A population receptive field model of the magnetoencephalography response
2021, NeuroImageCitation Excerpt :These differences are likely why we found higher pairwise correlations comparing variance explained sensor maps of the two forward models within subjects compared to across subjects (Supplementary Fig. S7B). Several MEG studies have aimed at reconstructing retinotopy responses on the cortical surface from MEG sensor measurements (e.g. (Moradi et al., 2003; Poghosyan and Ioannides, 2007; Sharon et al., 2007; Brookes et al., 2010; Perry et al., 2011; Cicmil et al., 2014; Nasiotis et al., 2017)). In those studies, instead of a forward model from stimulus to sensors, the cortical sources are estimated by inverse modeling: going from sensors to cortical sources, that is, estimated sources are derived by multiplying the sensor responses by the pseudo-inverse of the gain matrix in the head model.