Biomedical Engineering Reference
In-Depth Information
gradiometers, although someMEG systems do not use gradiometers relying on clever
noise cancellation methods. The two commonly used gradiometer configurations are
axial and planar gradiometers. Axial (first order) gradiometers consist of two coils
that share the same axis, whereas planar (first order) gradiometers consist of two
coils that share the same plane. The sensitivity profile of a planar gradiometer is
somewhat similar to EEG, whereby a sensor is maximally sensitive to a source clos-
est to it. In contrast, the sensitivity profile of an axial gradiometer can be somewhat
counterintuitive because it is not maximally sensitive to sources closest to the sensors.
ModernMEG systems consist of simultaneous recordings frommany sensors that
providewhole head coverage. The total number of sensors varies from100 to 300. The
advent of such large sensor-array systems has significantly advanced MEG studies.
Although the maximum sampling rate for many MEG systems reaches more than
10kHz, most MEG data is usually recorded at a sampling rate of around 1,000Hz,
which still provides excellent temporal resolution for measuring the dynamics of
cortical neuronal activity at millisecond order.
1.3 Electromagnetic Brain Imaging
MEG sensor data only provides qualitative information about underlying brain activ-
ities. The analysis of the sensor data is typically performed based on the intuitions
of experienced users regarding the sensitivity profile of the sensors. To extract more
precise information from the observed sensor data, it is essential to apply imaging-
type analysis involving the reconstruction of brain activities from the sensor data.
Major components for the electromagnetic brain imaging are the forward model and
the inverse algorithms.
1.3.1 Forward Model
The forward model consists of three subcomponents: the source model, the volume
conductor model, and the measurement model. Typical source models assume that
the brainmagnetic fields are generated by equivalent current dipoles in the brain. This
model is consistent with available measurements of coherent synaptic and intracellu-
lar currents in cortical columns that are thought to be major contributors to MEG and
EEG signals. Although several more complex source models have been proposed,
the equivalent current dipole is the dominant source model. This is because, given the
distance between the sources and sensors, the dipole is a reasonable approximation
of brain sources.
The volume conductor model refers to the equations that govern the relation
between the source model and the sensor measurements, namely the electric poten-
tials or the magnetic fields. These surface integral equations, obtained by solving
Maxwell's equations under quasi-static conditions, can be solved analytically for
