What Is Magnetoencephalography?

Magnetoencephalography (MEG) is a non-invasive neuroimaging technique that records the weak magnetic fields produced by synchronous electrical activity in populations of cortical neurons. Where electroencephalography measures the electric potential on the scalp, MEG measures the magnetic field component of the same underlying neural currents, and because magnetic fields are not distorted by the skull and scalp the way electric potentials are, MEG preserves both spatial precision and sub-millisecond temporal resolution. The technique was first demonstrated by David Cohen at MIT in 1968 and became clinically and scientifically viable during the 1980s with the development of superconducting quantum interference device (SQUID) sensor arrays. MEG sits at the intersection of neuroscience, biomedical image processing, and cryogenic instrumentation.

Signal Origins and Detection

Measurable MEG signals arise primarily from the synchronized postsynaptic currents of populations of tens of thousands of pyramidal neurons in the cerebral cortex. These cells are oriented perpendicular to the cortical surface, and when located in sulcal folds their primary current components are roughly tangential to the skull, producing magnetic fields that exit and re-enter the head in a pattern detectable by sensors placed outside. Radially oriented sources are largely invisible to MEG because their magnetic fields close entirely inside the head. This geometry contrasts with EEG, which is equally sensitive to both tangential and radial sources. The magnetic flux produced by a single active cortical patch is on the order of 10 to 100 femtotesla, a hundred million times weaker than the Earth's static magnetic field, requiring extreme sensitivity in the detection system.

SQUID Sensor Technology

MEG systems rely on superconducting quantum interference devices, cooled to approximately 4 K by liquid helium, to achieve the sensitivity needed to detect femtotesla-level fields. A SQUID operates by exploiting the Josephson effect: magnetic flux threading the sensor loop induces a measurable voltage change with a resolution limited by quantum noise. Modern whole-head MEG systems house 200 to 300 SQUID gradiometer channels in a helmet-shaped dewar that encloses the head without contact. The systems are operated inside magnetically shielded rooms that attenuate environmental noise by five to six orders of magnitude, because ambient urban magnetic noise from traffic, power lines, and elevators would otherwise overwhelm the neural signal. A comprehensive review of MEG physics, techniques, and neuroscience applications covers both conventional helium-cooled arrays and emerging optically pumped magnetometer designs that operate at room temperature.

Source Localization and Image Reconstruction

Because MEG records a magnetic field distribution outside the head rather than an image directly, the neural source configuration must be inferred computationally from the observed field pattern. This is the electromagnetic inverse problem, which is mathematically underdetermined: infinitely many source configurations can produce the same external field. Practical source localization methods include equivalent current dipole fitting, which assumes a small number of focal sources; beamforming spatial filters, which estimate source activity at each location independently; and minimum-norm estimates, which distribute source power across the cortical surface to minimize total energy. Combining MEG with structural MRI allows sources to be projected onto a realistic model of the individual's cortical geometry, producing functional maps with spatial accuracy of 2 to 5 mm and temporal resolution under 1 ms. The history of MEG development from SQUIDs to modern neuroscience documents the progression of inverse-problem algorithms alongside hardware improvements.

Applications

Magnetoencephalography has applications in a range of fields, including:

• Pre-surgical mapping of eloquent cortex in patients with brain tumors or epilepsy, localizing motor, language, and sensory areas before resection
• Epilepsy evaluation, where interictal spike localization from MEG helps identify seizure onset zones not visible on MRI
• Cognitive neuroscience research on attention, working memory, perception, and language, exploiting MEG's combination of temporal and spatial resolution
• Brain-computer interface development, using real-time MEG decoding of motor intentions
• Research on autism spectrum disorder, concussion, and Alzheimer's disease, where MEG functional connectivity patterns serve as potential diagnostic biomarkers