MEG Source Localization
Magnetoencephalography Source Localization · Also known as: MEG localization, magnetic source imaging, MSI
Magnetoencephalography (MEG) source localization is the inverse problem of estimating where in the brain neural currents originate from magnetic field measurements at the scalp. Introduced by David Cohen in 1972, MEG offers superior temporal resolution (milliseconds) and spatial specificity compared to EEG, as magnetic fields are less distorted by tissue conductivity, enabling researchers to pinpoint neural activity with high precision.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
MEG source localization is appropriate when high temporal and spatial resolution is required simultaneously, when sources are focal or regionally clustered, and when available resources permit (MEG systems are expensive). Use MEG for studying motor, sensory, and language systems where precise anatomical localization is critical. Avoid MEG when distributed sources are expected (use fMRI instead) or when cost constraints are severe.
Strengths & limitations
- Superior spatial resolution compared to EEG; magnetic fields bypass tissue distortion, yielding more accurate localization
- Exceptional temporal resolution (milliseconds); tracks dynamic source changes in real time
- Magnetic fields are reference-independent; no ambiguity from electrode placement (unlike EEG reference effects)
- Works well for focal sources (e.g., motor cortex, primary sensory cortex); dipole fitting is accurate and interpretable
- Combines advantages of EEG (temporal) and fMRI (spatial): superior to both on complementary dimensions
- Insensitive to currents aligned perpendicular to the scalp surface; radial sources (gyral crowns) often missed
- Expensive and specialized equipment; MEG systems cost millions and require dedicated shielded facilities
- Distributed source inversions are mathematically under-determined; multiple source configurations fit equally well
- Limited sensitivity to deep subcortical sources; strong dependence on source-sensor distance
Frequently asked
Why can't MEG detect all neural sources?
MEG is insensitive to sources oriented radially (perpendicular to scalp). This affects gyral crowns, which generate radial currents. MEG primarily detects tangential (parallel to scalp) currents from sulcal walls. fMRI is complementary, detecting radial sources. Combining MEG and fMRI provides comprehensive source coverage.
What is the difference between dipole fitting and distributed inverse solutions?
Dipole fitting assumes a small number of focal sources (1–5) and estimates location, orientation, and strength. Distributed methods estimate current at many locations simultaneously. Dipole fitting is accurate for focal sources but biased for distributed activity. Distributed methods handle distributed sources but are poorly determined (many solutions fit equally well). Choose based on expected source geometry.
How accurate is MEG source localization?
For focal sources in well-characterized anatomy (motor cortex, visual cortex), localization error is typically 5–10 mm. Error increases for deep sources or distributed activity. Anatomical variability and model misspecification contribute substantially. Always report confidence intervals and validate with independent methods when possible.
Can MEG be used clinically?
Yes, for epilepsy surgery planning and presurgical brain mapping. MEG localizes seizure foci and eloquent cortex (motor, language) with precision comparable to or exceeding fMRI. Clinical MEG requires specialized training and interpretation. Insurance coverage varies; it is widely used but not universally available.
Sources
- Hauk, O., Friston, K. J., & Leff, A. (2019). Functional neuroimaging of language: understanding the complex relationships between localization and function. Journal of Neurolinguistics, 50, 236–250. link ↗
- Halgren, E., Marinkovic, K., & Chauvel, P. (2006). Generators of the late cognitive potentials in auditory and visual oddball tasks. Electroencephalography and Clinical Neurophysiology, 106(2), 156–164. link ↗
How to cite this page
ScholarGate. (2026, June 3). Magnetoencephalography Source Localization. ScholarGate. https://scholargate.app/en/neuroimaging/meg-source-localization
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Dynamic Causal ModelingNeuroimaging↔ compare
- eLORETANeuroimaging↔ compare
- Event-Related Potential AnalysisNeuroimaging↔ compare