Magnetoencephalography (MEG): Principles and Lab Methods

Notice: This guide is intended for academic research, laboratory operations, and neuroimaging protocol development. Magnetoencephalography is also used in clinical neurophysiology (such as presurgical functional mapping); however, this article does not provide medical advice, diagnosis, or clinical care instructions.

Short answer

Magnetoencephalography (MEG) is a non-invasive neuroimaging technique that records magnetic fields produced by synchronized post-synaptic electrical currents in cerebral pyramidal neurons. Offering millisecond temporal precision alongside millimeter-scale spatial localization, MEG tracks rapid cortical dynamics without signal distortion caused by intervening skull and scalp tissues, enabling high-resolution source localization when co-registered with structural MRI.

Definition and scope

Magnetoencephalography measures the tiny magnetic flux densities—typically ranging from 10 to 1000 femtotesla (10⁻¹⁵ to 10⁻¹³ T)—generated outside the scalp by collective synaptic activity in cortical gray matter. Within the broader framework of human neuroscience research, MEG bridges the gap between the high temporal resolution of electrophysiology and the spatial localization of functional metabolic imaging.

Biophysically, the primary currents driving the MEG signal originate from excitatory postsynaptic currents (EPSCs) flowing longitudinally down the apical dendrites of pyramidal cells oriented perpendicularly to the cortical mantle. Because magnetic fields decay as the square or cube of distance and magnetic curl requires coherent dipole alignment, MEG is predominantly sensitive to tangential currents located along cortical sulci, complementing the radial sensitivity of surface electrophysiology.

How it works or how it is used in practice

Operating a research MEG facility requires specialized instrumentation, shielding, and analytical workflows divided across distinct laboratory stages:

• Magnetic Shielding: Earth’s geomagnetic field (~50 μT) and urban low-frequency noise (~10–100 nT) dwarf cerebral signals by 6 to 8 orders of magnitude. Recordings must occur inside a magnetically shielded room (MSR) constructed from high-permeability nickel-iron alloys (mu-metal) to attenuate static and low-frequency fields, coupled with aluminum layers to suppress high-frequency electromagnetic interference via eddy currents.
• Sensor Technologies: Conventional clinical and research systems utilize arrays of 100 to 300 Superconducting Quantum Interference Devices (SQUIDs). These sensors require cryogenic cooling via liquid helium inside vacuum-insulated dewars maintained at approximately 4.2 K (−269 °C). Emerging wearable systems instead deploy Optically Pumped Magnetometers (OPM-MEG), which monitor laser absorption through heated vapor cells containing rubidium-87 atoms under the spin-exchange relaxation-free (SERF) regime at room temperature.
• Subject Preparation and Co-Registration: Accurate source estimation requires precise spatial mapping of sensor locations relative to the participant’s cranial anatomy. Technicians record three anatomical fiducials (nasion, left preauricular point, and right preauricular point) and scalp surface points using a 3D magnetic digitizer (such as a Polhemus tracker). Head Position Indicator (HPI) coils attached to the scalp emit low-frequency magnetic tones to track continuous head position during recordings.
• Artifact Suppression and Preprocessing: Raw sensor time series undergo noise attenuation using Signal Space Separation (SSS) or temporally extended SSS (tSSS) to isolate magnetic fields generated inside the cranial volume from external disturbance vectors. Independent Component Analysis (ICA) identifies and subtracts physiological artifacts from cardiac electrical activity (ECG) and ocular blinks (EOG).
• Forward and Inverse Source Modeling: To map sensor-level sensor flux back onto cortical coordinates, researchers first solve the forward problem by computing lead-field matrices using Boundary Element Models (BEM) or Finite Element Models (FEM) derived from individual T1-weighted anatomical MRIs. The mathematically ill-posed inverse problem is then solved using deterministic algorithms such as equivalent current dipole (ECD) fitting, distributed source models (Minimum Norm Estimation [MNE], dynamic Statistical Parametric Mapping [dSPM], or standardized low-resolution brain electromagnetic tomography [sLORETA]), or adaptive spatial filters (Linearly Constrained Minimum Variance [LCMV] beamformers).

Types or related concepts

Understanding MEG requires contrasting sensor architectures, magnetic field pickup geometries, and complementary recording modalities:

Sensor / System Architecture	Operating Temperature	Pickup Geometry / Physical Mechanism	Key Research Utility
SQUID Magnetometers	Cryogenic (~4.2 K, Liquid Helium)	Single superconducting pickup loop coupling magnetic flux directly into a SQUID sensor.	High sensitivity to deep and distant cerebral sources; susceptible to distant environmental interference.
SQUID Planar Gradiometers	Cryogenic (~4.2 K, Liquid Helium)	Figure-eight figure loop measuring spatial derivative (dB/dx) of magnetic flux.	Maximizes signal directly above superficial cortical sources while cancelling uniform distant noise.
SQUID Axial Gradiometers	Cryogenic (~4.2 K, Liquid Helium)	Two concentric coils separated along the z-axis (dB/dz) in baseline configuration.	Balanced sensitivity across depth profiles; standard configuration in CTF and legacy systems.
Optically Pumped Magnetometers (OPM)	Room Temperature (~20–25 °C ambient)	Rubidium-87 vapor cell interrogated by circularly polarized laser under SERF regime.	Wearable, customizable montages with 4–6 mm scalp proximity, accommodating head movement and pediatric cohorts.

Where it matters for labs, clinics or supply decisions

MEG plays a vital role across cognitive neuroscience investigations and translational neurology facilities:

• Cognitive and Systems Neuroscience: Investigating the millisecond temporal choreography of sensory perception, motor execution, language parsing, and inter-areal phase synchronization. Unlike micro-electrode recordings, MEG captures whole-cortex macro-circuit oscillations non-invasively, providing a macroscopic counterpart to cellular investigations such as optogenetic neural interrogation.
• Presurgical Epilepsy and Functional Mapping: Clinical centers employ MEG to localize epileptogenic zones in pharmacoresistant focal epilepsy and map eloquent motor and receptive language cortex prior to surgical resection, reducing post-operative deficits.
• Data Sharing and Interoperability: Modern MEG pipelines adhere to standardized community specifications. Datasets formatted under the MEG-BIDS extension are archived on public platforms like OpenNeuro and BIDS neuroimaging data sharing, facilitating reproducible pipelines across MNE-Python, Brainstorm, FieldTrip, and SPM packages.
• Laboratory Facilities and Helium Stewardship: Traditional SQUID systems require continuous supplies of liquid helium for cryogenic boil-off replenishment or dedicated closed-loop helium re-liquefiers. Rising global helium costs and supply volatility have driven increased interest in OPM installations that avoid cryogen logistics altogether.

Common misunderstandings

Several methodological misconceptions exist regarding MEG data acquisition and interpretation:

• Misconception: MEG is simply a more expensive version of EEG. While both record post-synaptic neural activity at sub-millisecond rates, they measure orthogonal physical phenomena. As detailed in comparative studies of electroencephalography (EEG), volume conduction through skull, cerebrospinal fluid, and scalp significantly blurs electrical potentials. In contrast, biological tissues have magnetic permeability virtually identical to free space (μ₀), meaning MEG fields pass through cranial boundaries undistorted, yielding sharper spatial localization.
• Misconception: MEG cannot record deep brain structures. Although sensitivity decreases sharply with source depth due to inverse-square and dipole-decay laws, advanced beamforming algorithms and high-density sensor arrays can reliably reconstruct activity from hippocampal and amygdalar structures when sufficient trial averaging is applied.
• Misconception: MEG provides anatomical brain pictures directly. MEG records only time-varying magnetic flux densities at sensor pickup loops. It generates no anatomical tissue images. Anatomical structural visualization requires co-registration with a high-resolution T1 structural MRI scan acquired in a separate session.
• Misconception: More sensors always guarantee better spatial resolution. While sensor density is critical—comparable to selecting high-density sensor channel montages in surface electrophysiology—spatial resolution in MEG is ultimately bounded by signal-to-noise ratio, coregistration accuracy, forward model fidelity, and biological source extent rather than sensor count alone.

FAQ

What does magnetoencephalography (MEG) measure?

Magnetoencephalography measures the minute magnetic fields produced by net intracellular postsynaptic currents flowing through aligned cortical pyramidal neurons. These fields, measured in femtoteslas (fT), reflect direct electrical signaling in the brain at millisecond resolution.

What is fMRI a technique used for?

Functional magnetic resonance imaging (fMRI) is a non-invasive imaging technique used to localize brain activity by measuring blood-oxygen-level-dependent (BOLD) hemodynamic changes. While fMRI provides millimeter-scale spatial resolution of metabolic vascular responses, its temporal resolution is bounded by the slow hemodynamic response (several seconds), in contrast to the instantaneous electrophysiological tracking of MEG.

What is the newest brain scanning technique?

Among non-invasive human electrophysiology techniques, the newest major scanning advance is Optically Pumped Magnetometer MEG (OPM-MEG). OPM-MEG uses wearable, room-temperature quantum sensors placed directly against the scalp, eliminating the rigid, liquid-helium-cooled dewars of traditional SQUID systems and allowing participants to move naturally during recordings.

Is EEG a brain scanning technique?

Yes. Electroencephalography (EEG) is a non-invasive electrophysiological monitoring method that records electrical potential fluctuations generated by cerebral neuronal activity using electrodes placed along the scalp. It shares millisecond-level temporal resolution with MEG but is more heavily influenced by volume conduction through cranial tissues.

Sources

• Gross, J., Baillet, S., Barnes, G. R., Henson, R. N., Hämäläinen, M. S., Jensen, O., Jerbi, K., Litvak, V., Leahy, R. M., Mosher, J. C., Salmelin, R., Schnitzler, A., Taulu, S., & Schoffelen, J. M. (2013). Good practice for conducting and reporting MEG research. NeuroImage, 65, 349–363. https://doi.org/10.1016/j.neuroimage.2012.10.001 (Read 2026-11-08).
• Boto, E., Holmes, N., Leggett, J., Roberts, G., Shah, V., Meyer, S. S., Muñoz, L. D., Mullinger, K. J., Tierney, T. M., Bestmann, S., Barnes, G. R., Bowtell, R., & Brookes, M. J. (2018). Moving magnetoencephalography towards real-world applications with a wearable system. Nature, 555(7698), 657–661. https://doi.org/10.1038/nature26147 (Read 2026-11-08).
• Niso, G., Gorgolewski, K. J., Bock, E., Brooks, T. L., Flandin, G., Gramfort, A., Henson, R. N., Jas, M., Litvak, V., Moreau, J. T., Oostenveld, R., Schoffelen, J. M., Tadel, F., Wexler, J., & Baillet, S. (2018). MEG-BIDS, the brain imaging data structure extended to magnetoencephalography. Scientific Data, 5, 180110. https://doi.org/10.1038/sdata.2018.110 (Read 2026-11-08).