Both MEG and EEG are passive technologies; they use sensitive instrumentation to detect the tiny electrical and magnetic perturbations associated with neural activity. Although some aspects of biological electricity can be measured directly even with primitive galvanometers, the practical application of EEG required the development of sensitive amplifiers. In particular, amplifiers with high input impedance [such as devices incorporating field effect transistors (FETs) on the front end] allow precise potential measurements without drawing significant current. This strategy has a number of advantages, including reduced sensitivity to high impedance electrodes or connections to the scalp.
EEG techniques traditionally employ a modest number of electrodes that are applied individually by hand. The most commonly used system (particularly for clinical practice) is the international 10-20 system —a montage of around 20 electrodes placed over the surface of the scalp with reference to anatomical landmarks. Electrodes are affixed to the scalp with conductive gel or other adhesives after preparation of the surface, typically by mild abrasion. This procedure produces lower contact resistance but is labor-intensive. Advances in source localization techniques, to some extent driven by MEG, have provided motivation to increase the density of potential sampling across the scalp surface. Over the past decade typical research systems have grown from 32 channels to 128 channels or more. The development of electrode arrays based on caps or mechanical tension structures (see Fig. 3) has made the application of such arrays considerably more practical, requiring minutes instead of the hours required to apply dense arrays using conventional techniques. Although it is not yet clear how many electrodes are required to capture the nuances of the scalp potential map, studies of the spatial Nyquist frequency for EEG suggest that useful new details are available using 128-channel electrode arrays and perhaps even larger arrays.
MEG is based on ultrasensitive magnetic field measurement technology, incorporating superconducting quantum interference devices (SQUIDs). The field sensors are typically superconducting pickup coils consisting of multiple loops configured for sensitivity to field gradients. Both SQUIDs and gradiometers can be constructed from high-temperature superconductors. In designs to date, the increase in noise associated with higher temperatures has proven unacceptable for brain measurements, although such systems are adequate for cardiac applications. From early in the history of MEG, it has been clear that sensor arrays of
256 channels or more would be required to adequately sample field distributions associated with superficial cortical sources. However, the first systems consisted of detector arrays of limited extent. Mapping studies involved multiple placements of the sensor array in the context of event-related or evoked response paradigms. Most current MEG and EEG systems still sacrifice spatial sampling in order to provide whole-head coverage with a practical and more affordable number of sensor channels.
MEG sensor arrays are housed in a dewar, essentially a double-walled, vacuum-insulated flask, designed to hold cryogenic fluids such as helium. The dewar limits the proximity of sensors to the head surface; most superconducting sensor arrays are separated from the outside world by 1 cm or more. With advanced designs and considerable care, this distance can be reduced to closer to 1 mm for some applications. The standoff distance produces a significant loss of sensitivity, but this is not the limiting factor for most human brain measurements. Sensor arrays must be constructed with a rigid housing designed to accommodate the majority of heads, and smaller heads will involve greater separations over at least portions of the sensor array. Even the most superficial areas of cortex are on the order of 1 cm below the head surface, and some subcortical structures of interest may be
5-10 cm below the scalp. The fixed geometry of the MEG sensor array, the noncontact nature (and associated efficiency of subject preparation), and the relative insensitivity of MEG to head conductivity properties are all practical advantages of MEG for brain mapping applications. However, MEG sensor arrays providing full head coverage together with the magnetically shielded room used by most existing systems can cost as much as or more than a typical clinical MRI system with capabilities for functional neuroimaging.
Was this article helpful?