Magnetoencephalography¶

Recording brain-generated magnetic fields over time and inferring neural activity under source-model limits.

Version: v1 · 2026-09-28 · History
Domain-specific #: 10530
Domain group: Natural Sciences
Origin domain: Neuroscience
Subdomain: Functional Neuroimaging → Neuroscience
Aliases: MEG, Neuromagnetic recording

Core Idea¶

Magnetoencephalography, or MEG, is a noninvasive measurement technique for weak magnetic fields produced by brain electrical activity. Sensor arrays outside the head register field or gradient changes over time; the carrier is magnetic rather than the scalp voltage read by EEG or the blood-oxygen contrast of fMRI. Synchronized neuronal currents can create fields large enough to detect, but sensor type, positioning, source orientation, depth, and external magnetic noise affect the observed trace. SQUIDs and optically pumped magnetometers are different ways to fill the sensor role, not different abstractions.

MEG data are not direct pictures of particular neurons. To infer a location from extracranial fields, an analyst must solve an inverse problem with assumptions about the head and possible current sources; several sources may explain the same trace. That is why measured signal, modeled source, and any clinical interpretation must remain distinct. Published presurgical mapping studies show real use, including a pediatric motor-cortex series, but no one such study warrants a universal localization guarantee or a standalone treatment recommendation.

Structural Signature¶

Sig role-phrases:

neural current source — Supplies coordinated brain electrical activity capable of generating an extracranial magnetic field. It is constitutive. Counterfactual: An externally generated magnetic fluctuation without a brain-current source is interference, not the neural signal.
extracranial magnetic field — Carries a weak time-varying physical signal from currents to sensors outside the head. It is constitutive. Counterfactual: A blood-oxygen image or scalp voltage alone is not an MEG field recording.
sensitive sensor and reference frame — Records field or gradient traces with known timing, sensor position, and noise context. It is constitutive. Counterfactual: A field trace without a sensor/time frame cannot support MEG source interpretation.
source-inference model — Relates measured sensor patterns to possible current locations/orientations under anatomical and mathematical assumptions. It is central. Counterfactual: Sensor fields alone do not uniquely identify an intracranial source.
resolution and clinical limit — Qualifies inferred location, timing, depth sensitivity, noise, and permitted decision use. It is boundary. Counterfactual: A source estimate must not be treated as a direct image or standalone surgical instruction.

What It Is Not¶

EEG. EEG measures scalp electrical potentials rather than the magnetic carrier.
Functional MRI. Hemodynamic contrast is not a direct neuromagnetic field trace.
A direct neuron image. Source maps solve a nonunique inverse problem under model assumptions.
A standalone clinical verdict. One MEG-derived localization requires context and does not itself prescribe treatment.
Closest near-miss. EEG is the closest electrophysiologic neighbor: it records voltages from related neural activity, but the measured physical carrier is electric potential rather than extracranial magnetic field.

Scope of Application¶

Neuroscience research. Study time-resolved field patterns related to perceptual, motor, or cognitive activity.
Source-model comparison. Test how head and current assumptions alter possible field origins.
Presurgical mapping evidence. Use qualified field-derived maps as one information source in published evaluations.
Instrument comparison. Separate sensor technology changes from the shared neuromagnetic measurement identity.

Clarity¶

First identify the brain-generated field and sensor/time frame; then distinguish the recorded trace from a source estimate. EEG is the closest neighbor because it concerns related neural currents but measures scalp voltage. fMRI uses a hemodynamic carrier. A colorful MEG map is not a direct image of neurons: localization depends on a model and is not unique from sensor data alone. Neither SQUID cooling nor one magnetometer type defines every MEG implementation.

Manages Complexity¶

MEG compresses distributed neuronal currents into a manageable array of magnetic time series and then into source maps. This permits temporal comparison, but the second compression hides noise, sensor geometry, and inverse assumptions. Restating the physical carrier and inference chain prevents an attractive spatial visualization from being read as unmediated brain anatomy.

Abstract Reasoning¶

Identify the target neural activity and magnetic—not voltage or blood-oxygen—signal.
Record what sensor array, temporal frame, and interference context give meaning to the trace.
Separate detected field patterns from proposed neural source positions.
State the head/current model and source ambiguity behind each localization.
Bound any research or clinical interpretation to the evidence and corroboration available.

Knowledge Transfer¶

The sensor-to-field-to-source-inference audit transfers among MEG systems and study settings. SQUID-specific conditions do not transfer automatically to optically pumped sensors; a mapping result for one cohort or cortical orientation cannot certify another. Prime Measurement supplies the broader instrument/procedure/value/uncertainty chain, while this specialist technique is distinguished by brain-generated magnetic fields.

Examples¶

Canonical¶

In a conceptual evoked-field case, a group of synchronized cortical currents produces a weak time-varying magnetic pattern registered by multiple sensors outside the head. Plotting the field over time is direct measurement; placing a source estimate in cortex requires a declared head/current model and can admit multiple compatible solutions. A spatial map is therefore a qualified inference, not a photograph of individual neurons.

Mapped back: neural current source → coordinated cortical electrical activity; extracranial magnetic field → time-varying field outside the head; sensitive sensor and reference frame → multi-sensor trace and timing frame; source-inference model → explicit head/current assumptions; resolution and clinical limit → nonunique localization and no direct neuronal image.

Applied / In Practice¶

Gaetz and colleagues' published presurgical study recorded MEG in ten pediatric patients with focal brain lesions and analyzed motor-related fields to localize the hand area of primary motor cortex, comparing available results with intraoperative mapping in a subset. This is an actual clinical-research use of neuromagnetic recording and modeled localization, not a claim that MEG alone decides treatment or performs equally well in every patient.

Mapped back: neural current source → motor-related brain current activity in the studied cohort; extracranial magnetic field → recorded neuromagnetic activity; sensitive sensor and reference frame → study's temporally registered MEG sensor data; source-inference model → published spatial filtering and localization analysis; resolution and clinical limit → small study and corroboration in a subset, not universal decision sufficiency.

Structural Tensions¶

T1 — High Temporal Detail versus Inverse Spatial Ambiguity. Field timing can be precise even when several neural source configurations explain a sensor pattern.

Diagnostic: Which source assumptions support the displayed location?

T2 — Weak Neural Signal versus Ambient And Physiological Noise. Improved sensor sensitivity expands observation only when environmental and biological interference is bounded.

Diagnostic: What evidence separates neural fields from competing signals?

Structural–Framed Character¶

The approved DAG parent is Measurement: sensors map weak brain-generated magnetic fields into timed readings under calibration and uncertainty. Source inversion estimates possible neural origins but is nonunique.

Evaluative weight: A source estimate is model-dependent, not direct proof of one cortical location. Human-practice-bound: Moderate, because acquisition and inverse assumptions are selected while fields are physical. Institutional origin: Neuroimaging practice maintains systems; sensor technologies vary. Vocabulary travels: Different MEG arrays can qualify after rechecking sensitivity and geometry. Import versus recognize: Recognize MEG by extracranial neuromagnetic measurement; a generic magnetic image or transferred cohort result imports unsupported specificity.

Its character: A neuromagnetic measurement subtype with portable sensor-to-value logic and inverse-model limits.

Structural Core vs. Domain Accent¶

Skeletal core. An instrument interacts with a target attribute and produces framed, uncertain values.

Domain-bound accent. MEG records brain-current-related magnetic fields outside the head and estimates sources under a nonunique inverse model.

Why not prime. Measurement is broader; other magnetic readings lack this brain-origin and timing carrier.

This entry is a kind of Measurement.

Strict parent — measurement. MEG maps a brain-current-related field through an instrument and procedure onto time-stamped physical readings whose uncertainty and frame constrain interpretation.
Related — EEG. Both concern neural activity, but one measures magnetic field and the other scalp potential.
Related — magnetic source imaging. MRI registration and inverse localization extend a sensor trace into an anatomical map; they are not identical to raw MEG measurement.

Relationships to Other Abstractions¶

Current abstraction Magnetoencephalography Domain-specific

Parents (1) — more general patterns this builds on

Magnetoencephalography is a kind of Measurement Prime

MEG measures a brain-generated magnetic field through timed sensors and interprets the resulting readings under uncertainty and source-frame limits.

Hierarchy path (1) — routes to 1 parentless root

Magnetoencephalography → Measurement

Neighborhood in Abstraction Space¶

Magnetoencephalography sits in a moderately populated region (50^th percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Biomedical Signal Sensing & Recording (20 abstractions)

Nearest neighbors

Magnetogastrography — 0.88
Electrocochleography — 0.87
Correlated Double Sampling — 0.86
Periodic lateralized epileptiform discharges — 0.86
Electrocardiography — 0.86

Computed from structural-signature embeddings · 2026-10-08

Not to Be Confused With¶

EEG. Tell: Was electric potential recorded instead of magnetic field?
Functional MRI. Tell: Is the carrier blood-oxygen response rather than brain-generated magnetic flux?
External magnetic artifact. Tell: Can the measured pattern be attributed to neural current under the sensor/noise frame?
Source map. Tell: Which inverse assumptions convert the trace into a brain location?

References¶

Hämäläinen et al., Magnetoencephalography: theory, instrumentation, and applications (1993): https://megcore.nih.gov/images/8/88/1993-Hamalainen-RMP.pdf
Gaetz et al., pediatric motor-cortex MEG mapping (2009), PMID 19240567: https://pubmed.ncbi.nlm.nih.gov/19240567/
Frozen Wikipedia discovery revision: https://en.wikipedia.org/wiki/Magnetoencephalography (revision 1369226767).
Preserved source candidate: https://aaltodoc.aalto.fi/bitstream/123456789/18757/1/A1_h%c3%a4m%c3%a4l%c3%a4inen_matti_1993.pdf
Preserved source candidate: https://archive.org/details/physiologybehavi00carl_811
Preserved source candidate: https://archive.org/details/physiologybehavi00carl_811/page/n172
Preserved source candidate: https://www.ovid.com/jnls/annalsofian/fulltext/10.4103/0972-2327.128676~magnetoencephalography-basic-principles
Preserved source candidate: http://davidcohen.mit.edu/sites/default/files/documents/1972ScienceV175(SquidMEG).pdf
Preserved source candidate: https://books.google.com/books?id=7x3aBwAAQBAJ&pg=PA399
Preserved source candidate: https://pdfs.semanticscholar.org/92c5/d54611a8c35f262f82bd81b1c6223d71d29f.pdf
Preserved source candidate: https://web.archive.org/web/20200803165556/https://pdfs.semanticscholar.org/92c5/d54611a8c35f262f82bd81b1c6223d71d29f.pdf