Magnetoencephalography (MEG) is a noninvasive method used to directly measure and localize brain activity. Since the first MEG signal was recorded by Bruce Cohen in 1968 using a single sensor, the technique has evolved to include whole-head arrays with over 300 sensors. This allows for greater detection and localization of brain activity due to an improved signal-to-noise ratio and established MEG as a clinically useful technique. As a result, there are now more than 25 MEG centers in the United States alone, with an estimated 750 MEG scans performed in 2016.1
PHYSICS AND TECHNICAL REQUIREMENTS
Neuronal Generation of Magnetic Fields and Their Propagation
Individual neurons produce an electrical current when firing, known as an action potential, which is a critical component in neuronal signaling. When neurons fire synchronously, the resulting activity is summated into large-scale electrical currents that can propagate beyond the site of origin and be detected by electrodes on the scalp. However, due to the electrical resistivity of tissues through which the electrical currents must pass (eg, cerebrospinal fluid, dura, skull), the signal intensity is significantly reduced once detected by scalp electrodes.2,3 This represents the basic concept of electroencephalography (EEG) and is important in understanding the foundational differences between EEG and MEG.
A fundamental component of any electrical current is that it also produces a magnetic field, and the changes in magnetic fields produced by active neurons can also be detected on the scalp, which represent the basic concept of MEG. Unlike electrical fields, however, the properties of magnetic fields are such that intervening tissues offer very little resistance, and this allows for signal detection with minimal distortion or attenuation relative to EEG.4-6 Another relevant difference between these modalities is that EEG is thought to be sensitive to activity generated by neurons oriented both perpendicular (ie, within the base of sulci and crowns of gyri) and parallel (ie, within gyri) to the scalp, whereas MEG is most sensitive to parallel-oriented neurons (within the walls of the gyri).6
Detection of Magnetic Fields by Magnetometers
The intrinsic magnetic flux generated by active neurons is measured by MEG via sensors referred to as superconducting quantum interference devices (SQUIDs), which consist of detection coils capable of sensing extremely small magnetic fluctuations. As a reference of scale, SQUIDs can detect field strengths as small as 10 to 250 femtotesla, or one-billionth the strength of the earth’s magnetic field.
SQUIDs can record 1 data point per millisecond and so are equal in temporal resolution to the high sampling rate of EEG. When displayed as a function of time, MEG recordings appear very similar to EEG waveforms (Figure 24–1). Functional magnetic resonance imaging (MRI), as a comparison, has a sampling rate of 1 to 2 seconds. The spatial resolution of MEG ...