Magnetoencephalography (MEG) and Magnetic Resonance Imaging (MRI) are two of the most powerful tools in modern neuroscience, but they operate on fundamentally different physical principles. While MRI creates static anatomical images by manipulating the magnetic properties of hydrogen nuclei, MEG records the dynamic magnetic fields produced by the brain’s own electrical activity. Understanding the physics behind each technique—and how they complement one another—reveals their unique roles in both research and clinical practice. This article explores the core physical concepts of MEG, the basics of MRI physics, and the synergistic relationship between these two modalities.

What is Magnetoencephalography (MEG)?

Magnetoencephalography is a completely non‑invasive functional neuroimaging technique that measures the magnetic fields generated by neuronal currents. When large groups of pyramidal neurons in the cerebral cortex fire synchronously, the resulting electrical currents produce tiny magnetic fields (on the order of femtoTesla to picoTesla—about one billionth of the magnetic field of the Earth). MEG sensors, typically superconducting quantum interference devices (SQUIDs), detect these fields with exquisite sensitivity, providing millisecond‑temporal resolution that electroencephalography (EEG) cannot match for spatial accuracy.

The ability to track neural activity in real time makes MEG indispensable for mapping sensory, motor, and cognitive functions. It is used extensively in presurgical planning for epilepsy and brain tumor resections, in studying the dynamics of language processing, and in characterizing the oscillatory rhythms of the healthy and diseased brain. The technique is silent, painless, and does not expose the subject to any ionizing radiation.

Physical Principles of MEG

The magnetic fields measured by MEG arise from the intracellular currents flowing along the dendrites of synchronised neurons. According to the Biot–Savart law, a moving charge creates a magnetic field perpendicular to the direction of current flow. Because the head is largely a volume conductor with different conductivity layers (scalp, skull, cerebrospinal fluid, brain), the magnetic fields are less distorted by tissue boundaries than are the electric potentials measured by EEG. This gives MEG better spatial resolution for sources located in cortical sulci.

To detect such weak signals, MEG systems require magnetically shielded rooms to block environmental noise (e.g., power‑line hum, moving metal objects). The sensors are cooled to liquid helium temperatures (4.2 K) to maintain superconductivity in the SQUIDs. Advanced processing algorithms, including beamforming and dipole fitting, are then applied to estimate the location and orientation of the underlying neural sources—a process called source localisation.

Basics of MRI Physics

Magnetic Resonance Imaging is based on the principles of nuclear magnetic resonance (NMR). Hydrogen nuclei (protons) in water and fat molecules possess a property called spin, which endows them with a small magnetic moment. When placed in a strong static magnetic field (typically 1.5 T to 7 T for human scanners), a fraction of these proton spins align with the field, creating a net magnetisation vector within the tissue.

A radiofrequency (RF) pulse is then applied at the Larmor frequency—the precessional frequency specific to the magnetic field strength (e.g., about 64 MHz at 1.5 T). This pulse tips the net magnetisation away from its equilibrium position. After the pulse ends, the protons relax back to their aligned state, emitting a faint RF signal that is captured by receiver coils. The time constants of relaxation—T1 (longitudinal) and T2 (transverse)—differ among tissues and form the basis of image contrast.

By applying magnetic field gradients in three orthogonal directions, the spatial origin of the relaxation signals is encoded. A mathematical reconstruction (Fourier transform) converts the raw data into a detailed, three‑dimensional anatomical image. Unlike X‑ray methods, MRI provides excellent soft‑tissue contrast without ionising radiation, making it the gold standard for structural brain imaging.

Relationship Between MEG and MRI Physics

At first glance, MEG and MRI share the word “magnetic” but employ opposite aspects of magnetism: MRI uses a strong, homogeneous, static field to create an image; MEG passively measures the exceedingly weak, dynamic fields produced by the brain. Despite this difference, the two techniques are deeply complementary in both physics and clinical utility.

Complementary Underlying Physics

The static B₀ field in an MRI scanner (~1.5 T) is more than a billion times stronger than the magnetic fields detected by MEG (femtoTesla). This means that MRI systems are designed to generate a uniform field, while MEG systems are designed to detect infinitesimal field fluctuations with minimal interference. The noise floor of an MEG system must be extremely low; thus, MEG is never performed inside an MRI scanner simultaneously (though combined MEG‑MRI systems exist for sequential acquisition).

However, MRI physics provides the structural framework for MEG analysis. Coregistration of MEG source localisation results onto an individual’s high‑resolution MRI scan is essential for accurate mapping. The electromagnetic forward model—which predicts what magnetic fields a given source would produce—requires knowledge of the head geometry, including the scalp, skull, and brain surfaces. This information is derived from T1‑weighted MRI images. Thus, while the physical phenomena differ, the mathematical models that connect them rely on MRI data.

Combined Use in Research and Clinical Practice

In research, MEG and MRI together enable the study of brain function with both high temporal (MEG) and high spatial (MRI) resolution. For example, in a language mapping experiment, MEG can show that the left inferior frontal gyrus becomes active 200 ms after a visual stimulus, while the structural MRI confirms the exact gyral location and any pathology. This synergy is even more powerful in diffusion MRI, which maps white‑matter tracts; combining MEG functional data with diffusion tractography reveals how neural signals travel along structural pathways.

Clinically, the combination is vital for presurgical evaluation of epilepsy. MEG identifies the location of interictal epileptiform discharges (spikes), which are then projected onto the patient’s MRI to guide electrode placement or surgical resection. In brain tumor surgery, MEG maps eloquent cortex (e.g., motor, language) to minimise postoperative deficits, while MRI delineates the tumor margins. The integration is often performed using specialised software that fuses functional MEG maps with structural MRI volumes.

  • Presurgical Planning: MEG localises functional areas (sensorimotor, language, visual); MRI provides the anatomical road map.
  • Epilepsy: MEG source imaging pinpoints epileptic foci; high‑resolution MRI reveals underlying structural lesions like cortical dysplasia or hippocampal sclerosis.
  • Stroke and Trauma: MEG can detect slow‑wave activity indicating tissue dysfunction, while MRI shows infarct location and extent.
  • Connectomics: MEG measures resting‑state network dynamics; diffusion MRI reconstructs the white‑matter connections between those network nodes.

Technical Integration Challenges

From a physics standpoint, combining MEG and MRI data requires careful alignment. The MEG sensor array is fixed in a room coordinate system, while the subject’s head is positioned inside it. Head position indicator coils or optical tracking systems continuously monitor head movement during MEG acquisition. After the MEG session, an MRI is obtained, and the two data sets are aligned using fiducial markers (typically vitamin E capsules placed on the nasion and preauricular points) or surface matching algorithms (e.g., iterative closest point).

Another challenge is the magnetic susceptibility artifact that arises when MEG‑generated field maps are overlaid on MRI. For example, blood‑oxygen‑level‑dependent (BOLD) functional MRI captures a haemodynamic response that lags neural activity by several seconds, whereas MEG captures the electrical event itself. Dozens of studies have attempted to cross‑validate BOLD and MEG signals; the results generally show that MEG gamma‑band power correlates positively with BOLD, but the relationship is complex because MEG is sensitive to both the magnitude and orientation of currents.

Despite these challenges, the integration of MEG with structural and functional MRI has become standard practice. Many institutions now have combined MEG‑MRI suites where the patient undergoes both scans in a single visit, reducing coregistration errors and improving clinical workflow.

Applications That Depend on the MEG–MRI Relationship

Functional Mapping in the Operating Room

One of the most impactful applications is in “awake craniotomy” for brain tumours near eloquent areas. Preoperatively, a patient undergoes MEG to map motor and language cortex. The MEG source localisation is overlaid on the MRI, which the neurosurgeon imports into a neuronavigation system. During surgery, the navigation system shows the surgeon where the functional zones are relative to the tumour, even if the brain shifts (the system updates based on intraoperative MRI or ultrasound). This physical fusion of the MEG‑derived functional map with the MRI‑derived anatomy directly reduces the risk of permanent deficits.

Understanding Brain Development and Aging

MEG and MRI are increasingly used together in developmental studies. For example, to study how the brain’s resting‑state networks mature, researchers use MEG to measure oscillatory power (e.g., alpha rhythm) and diffusion MRI to assess white‑matter integrity. The combination reveals that network connectivity (measured by MEG) correlates with the microstructural properties of the underlying tracts (measured by MRI). This interplay would be impossible to deduce from either modality alone.

Investigation of Epileptic Networks

In epilepsy, MEG can identify not just the irritative zone (where spikes originate) but also the propagation pathway of seizures. When these MEG‑derived maps are co‑registered with MRI‑based morphometry (e.g., cortical thickness, grey‑matter volume), clinicians can often detect subtle cortical abnormalities that are missed on standard MRI. In one study, combining MEG source imaging with high‑resolution 7 T MRI increased the detection rate of focal cortical dysplasia by 30% compared with 3 T MRI alone.

Research on Cognitive Processes

Cognitive neuroscientists routinely exploit the synergy. A typical experiment might examine the neural correlates of visual working memory. MEG provides the time course of activation in prefrontal and parietal cortices (e.g., sustained gamma activity during the delay period), while MRI gives the precise sulcal geometry of each participant. This allows the researcher to model how the shape of a specific sulcus influences the MEG signal (because the orientation of pyramidal neurons relative to the skull affects the measured field). Such integrative analyses are only possible by merging MEG and MRI physics.

Future Directions

The relationship between MEG and MRI physics continues to evolve. Hybrid devices that can simultaneously acquire MEG and MRI data are being developed. These systems use optically pumped magnetometers (OPMs) instead of cryogenic SQUIDs, allowing placement closer to the scalp and potentially inside a low‑field MRI scanner (< 0.1 T). If successful, such hybrid systems would enable truly simultaneous recordings of neural magnetic fields and MR images, eliminating coregistration errors and providing a seamless view of structure and function.

Another frontier is the use of MEG to validate and inform MRI‑based connectivity. Whole‑brain modelling efforts (e.g., the Human Connectome Project) increasingly require dynamic data to constrain the static structural connectome. MEG’s ability to measure inter‑regional coherence in multiple frequency bands provides the dynamic counterpart needed to build realistic models of brain function.

Finally, advances in machine learning are being applied to fuse MEG and MRI data. Deep learning algorithms can learn to predict MEG source localisation directly from MRI anatomy, or to use MEG oscillatory patterns to infer structural connectivity. These tools promise to extract even more information from the physical link between the two modalities.

Conclusion

Magnetoencephalography and magnetic resonance imaging, while based on distinct physical phenomena—measurement of neural magnetic fields versus manipulation of nuclear spins—are inseparable partners in modern neuroscience and clinical neurology. The static, high‑resolution anatomy offered by MRI physics provides the essential scaffold for interpreting the dynamic, millisecond‑scale brain activity captured by MEG. Their relationship is not one of competition but of deep complementarity, enabling unparalleled insights into the normal and diseased human brain. As instrumentation and analysis methods continue to advance, the integration of MEG and MRI will only grow tighter, driving progress in diagnostics, treatment planning, and fundamental brain science.

For further reading: see the Nature subject page on Magnetoencephalography, the NIH primer on MEG, and a detailed technical review in Human Brain Mapping.