Flashcards in MEG Deck (20):
-MEG measures neural activity by measuring the changing
magnetic field produced by the brain.
-The same neural activity that produces the scalp
-potentials that are detected by EEG also produce magnetic fields that can be detected in MEG.
-Thus, MEG and EEG are highly related techniques, differing only in whether the magnetic or electric field is
-Like EEG, MEG is completely non-invasive.
Like EEG, MEG primarily detects the post
-synaptic currents of the pyramidal cells of the cortex.
-Electrical currents produce magnetic fields
-For an MEG recording session, the observer places his/her head inside the scanner.
This can be done either in a supine or in a sitting position
magnetic field strength
The magnetic field produced by the brain is very weak
Much weaker than the earth’s magnetic field of 10-4tesla.
Thus, for MEG you need very sensitive detectors and a
way of suppressing the background magnetic field.
-The detectors are called
-The acronym stands for Superconducting Quantum
-Each one is cooled to 4 Kelvin by liquid helium!
-A typical scanner will have about 300 of them
Suppressing The Background Magnetic
-The scanner is placed inside
a magnetically shielded
-The walls, floor and ceiling
of the room are made of
metal, greatly reducing the
magnetic field that enters
-The closer a magnetic source is to the
SQUIDs the more rapidly its magnetic field will decrease with
-Thus, by recording only those magnetic fields that vary rapidly with distance, you can isolate the magnetic field originating from the brain and distinguish it from that originating from more distant sources
-You can also sometimes remove external magnetic field using software tricks.
-For example, you can filter out the 50 Hz
associated with the mains voltage this way.
Two ways of using it
-Temporal analysis (e.g. evoked recordings)
-Spatial analysis (e.g. dipole modeling
-Recall that in EEG a popular temporal analysis method was the evoked recording technique.
-According to this technique, a stimulus would be presented to an observer and the resultant EEGs recorded.
-This process would be repeated many times so the average response to a stimulus could be
-The evoked recording is thus the
to the stimulus
Evoked Recordings continued
Evoked recordings in MEG are just the same as EEG
-You present the same stimulus multiple times and
record the average response.
-The only difference is that you have two types of
-Gradiometers measure the gradient of the magnetic
-Magnetometers measure the absolute magnetic field strength
-Sometimes the recordings are
looking down at the top of
-Notice that similar waveform
shapes appear in similar
regions (e.g. SI)
A computer analyses the MEG
signals to try to figure out how
many separate sources there
-In the previous example, there
seemed to be four different
waveforms types, suggesting
that there are four sources (SI,
SIIc, SIIi, MI).
dipole modelling continued
-For each source, the computer calculates the location of a magnetic dipole that would give rise to recorded magnetic fields. (See Piers’ EEG lecture
for more description).
-These dipoles are then superimposed on an image
of the brain.
-MEG can not be used to obtain an anatomical image of the brain so the anatomical image is obtained by MRI.
-Different dipole arrangements can give rise to the same magnetic fields at the surface of the head.
-Thus, trying to determine the dipole arrangements
from the magnetic fields is an “ill
-posed inverse problem”
-“Inverse problem” refers to the difficulty in working backwards from the magnetic fields to determine the dipole arrangement that caused them.
-The problem is “ill posed” because multiple dipole arrangements could give rise to the same magnetic fields
-In addition, sometimes the neural activity is
extended rather localized to just a single cortical area, making it wrong to assume that the magnetic fields can be well described by dipoles (i.e. the
dipole assumption underlying dipole modeling is probably incorrect in this case).
differences between EEG and MEG
-MEG can only detect those magnetic dipoles that are not oriented
-Thus, it typically cannot detect neural activity at the surface of brain as this activity typically gives rise to
-It also has difficulty detecting neural activity near the center of the brain as this activity also typically gives rise to radial dipoles.
-Although EEG can (in principle) detect sources anywhere in the brain, sources nearer the surface are easier to detect, so mask those deeper in the
-Thus, EEG tends to be dominated by sources on the surface of the brain (i.e. on the gyri).
-Conversely, MEG records primarily from the sulci
-Thus, while MEG and EEG recordings tend to be similar, they are usually (at least slightly) different.
-In addition, because MEG tends to detect fewer sources it is often able to localise them more precisely.
-(The fact that the skull and meninges
tend to distort magnetic fields less than electric ones also
-However, the improvement in spatial resolution
compared to EEG is typically quite small.
-So while MEG sources can usually be localised slightly better than EEG sources...
...in both cases you still have to make assumptions (e.g. assume a dipole model)...
...so in both cases the localisations are uncertain.
-fMRI is much better at localising neural activity as you do not have to make as many assumptions.
-The temporal resolution of MEG and EEG is about the same.
Example MEG experiment
-MRI studies have shown that fusiform
face area is specialisedfor processing faces.
-Suppose that you want to know how long it takes for this areas to process a face.
-The temporal resolution of
fMRIis too poor to answer this question.
Halgren et al. (2000) addressed this issue using MEG.
-On some trials they presented pictures of faces to the observers (the “face stimuli”) and on other trials they presented randomized (i.e. scrambled) face stimuli (the “randomized stimuli”)
-They isolated a magnetic source that was more active for face stimuli than randomized stimuli.
-They found that this source was well represented by a dipole located in the right
-They assumed that this dipole was therefore located in the fusiform face area (as this area is a)
located in the fusiform gyrus
and b) known to be
response to face stimuli)
This source responded
more strongly to faces
than randomised faces
or animal faces or
-This selectivity is
reliable at 165 ms after
the stimulus onset