<s>
Magnetoencephalography	B-Algorithm
(	O
MEG	O
)	O
is	O
a	O
functional	B-Algorithm
neuroimaging	I-Algorithm
technique	O
for	O
mapping	O
brain	B-Application
activity	I-Application
by	O
recording	O
magnetic	O
fields	O
produced	O
by	O
electrical	O
currents	O
occurring	O
naturally	O
in	O
the	O
brain	O
,	O
using	O
very	O
sensitive	O
magnetometers	O
.	O
</s>
<s>
Arrays	O
of	O
SQUIDs	B-Protocol
(	O
superconducting	O
quantum	O
interference	O
devices	O
)	O
are	O
currently	O
the	O
most	O
common	O
magnetometer	O
,	O
while	O
the	O
SERF	O
(	O
spin	O
exchange	O
relaxation-free	O
)	O
magnetometer	O
is	O
being	O
investigated	O
for	O
future	O
machines	O
.	O
</s>
<s>
Applications	O
of	O
MEG	O
include	O
basic	O
research	O
into	O
perceptual	O
and	O
cognitive	O
brain	O
processes	O
,	O
localizing	O
regions	O
affected	O
by	O
pathology	O
before	O
surgical	O
removal	O
,	O
determining	O
the	O
function	O
of	O
various	O
parts	O
of	O
the	O
brain	O
,	O
and	O
neurofeedback	B-Application
.	O
</s>
<s>
This	O
can	O
be	O
applied	O
in	O
a	O
clinical	O
setting	O
to	O
find	O
locations	O
of	O
abnormalities	O
as	O
well	O
as	O
in	O
an	O
experimental	O
setting	O
to	O
simply	O
measure	O
brain	B-Application
activity	I-Application
.	O
</s>
<s>
MEG	O
signals	O
were	O
first	O
measured	O
by	O
University	O
of	O
Illinois	O
physicist	O
David	O
Cohen	O
in	O
1968	O
,	O
before	O
the	O
availability	O
of	O
the	O
SQUID	B-Protocol
,	O
using	O
a	O
copper	O
induction	O
coil	O
as	O
the	O
detector	O
.	O
</s>
<s>
Later	O
,	O
Cohen	O
built	O
a	O
much	O
better	O
shielded	O
room	O
at	O
MIT	O
,	O
and	O
used	O
one	O
of	O
the	O
first	O
SQUID	B-Protocol
detectors	O
,	O
just	O
developed	O
by	O
James	O
E	O
.	O
Zimmerman	O
,	O
a	O
researcher	O
at	O
Ford	O
Motor	O
Company	O
,	O
to	O
again	O
measure	O
MEG	O
signals	O
.	O
</s>
<s>
This	O
time	O
the	O
signals	O
were	O
almost	O
as	O
clear	O
as	O
those	O
of	O
EEG	B-Application
.	O
</s>
<s>
This	O
stimulated	O
the	O
interest	O
of	O
physicists	O
who	O
had	O
been	O
looking	O
for	O
uses	O
of	O
SQUIDs	B-Protocol
.	O
</s>
<s>
At	O
first	O
,	O
a	O
single	O
SQUID	B-Protocol
detector	O
was	O
used	O
to	O
successively	O
measure	O
the	O
magnetic	O
field	O
at	O
a	O
number	O
of	O
points	O
around	O
the	O
subject	O
's	O
head	O
.	O
</s>
<s>
At	O
the	O
same	O
time	O
,	O
they	O
feature	O
sensitivity	O
equivalent	O
to	O
that	O
of	O
SQUIDs	B-Protocol
.	O
</s>
<s>
Synchronized	B-Algorithm
neuronal	I-Algorithm
currents	I-Algorithm
induce	O
weak	O
magnetic	O
fields	O
.	O
</s>
<s>
The	O
MEG	O
(	O
and	O
EEG	B-Application
)	O
signals	O
derive	O
from	O
the	O
net	O
effect	O
of	O
ionic	O
currents	O
flowing	O
in	O
the	O
dendrites	O
of	O
neurons	O
during	O
synaptic	B-Application
transmission	O
.	O
</s>
<s>
It	O
is	O
worth	O
noting	O
that	O
action	B-Algorithm
potentials	I-Algorithm
do	O
not	O
usually	O
produce	O
an	O
observable	O
field	O
,	O
mainly	O
because	O
the	O
currents	O
associated	O
with	O
action	B-Algorithm
potentials	I-Algorithm
flow	O
in	O
opposite	O
directions	O
and	O
the	O
magnetic	O
fields	O
cancel	O
out	O
.	O
</s>
<s>
This	O
helps	O
eliminate	O
radio	O
frequency	O
radiation	O
,	O
which	O
would	O
degrade	O
SQUID	B-Protocol
performance	O
.	O
</s>
<s>
Problems	O
such	O
as	O
this	O
,	O
where	O
model	O
parameters	O
(	O
the	O
location	O
of	O
the	O
activity	O
)	O
have	O
to	O
be	O
estimated	O
from	O
measured	O
data	O
(	O
the	O
SQUID	B-Protocol
signals	O
)	O
are	O
referred	O
to	O
as	O
inverse	O
problems	O
(	O
in	O
contrast	O
to	O
forward	O
problems	O
where	O
the	O
model	O
parameters	O
(	O
e.g.	O
</s>
<s>
Possible	O
solutions	O
can	O
be	O
derived	O
using	O
models	O
involving	O
prior	O
knowledge	O
of	O
brain	B-Application
activity	I-Application
.	O
</s>
<s>
One	O
type	O
of	O
localization	O
algorithm	O
for	O
overdetermined	O
models	O
operates	O
by	O
expectation-maximization	B-Algorithm
:	O
the	O
system	O
is	O
initialized	O
with	O
a	O
first	O
guess	O
.	O
</s>
<s>
Another	O
common	O
technique	O
is	O
beamforming	B-Algorithm
,	O
wherein	O
a	O
theoretical	O
model	O
of	O
the	O
magnetic	O
field	O
produced	O
by	O
a	O
given	O
current	O
dipole	O
is	O
used	O
as	O
a	O
prior	O
,	O
along	O
with	O
second-order	O
statistics	O
of	O
the	O
data	O
in	O
the	O
form	O
of	O
a	O
covariance	O
matrix	O
,	O
to	O
calculate	O
a	O
linear	O
weighting	O
of	O
the	O
sensor	O
array	O
(	O
the	O
beamformer	B-Algorithm
)	O
via	O
the	O
Backus-Gilbert	O
inverse	O
.	O
</s>
<s>
This	O
is	O
also	O
known	O
as	O
a	O
linearly	O
constrained	O
minimum	O
variance	O
(	O
LCMV	O
)	O
beamformer	B-Algorithm
.	O
</s>
<s>
When	O
the	O
beamformer	B-Algorithm
is	O
applied	O
to	O
the	O
data	O
,	O
it	O
produces	O
an	O
estimate	O
of	O
the	O
power	O
in	O
a	O
"	O
virtual	O
channel	O
"	O
at	O
the	O
source	O
location	O
.	O
</s>
<s>
The	O
two	O
sets	O
of	O
data	O
are	O
combined	O
by	O
measuring	O
the	O
location	O
of	O
a	O
common	O
set	O
of	O
fiducial	B-Architecture
points	I-Architecture
marked	O
during	O
MRI	O
with	O
lipid	O
markers	O
and	O
marked	O
during	O
MEG	O
with	O
electrified	O
coils	O
of	O
wire	O
that	O
give	O
off	O
magnetic	O
fields	O
.	O
</s>
<s>
The	O
locations	O
of	O
the	O
fiducial	B-Architecture
points	I-Architecture
in	O
each	O
data	O
set	O
are	O
then	O
used	O
to	O
define	O
a	O
common	O
coordinate	O
system	O
so	O
that	O
superimposing	O
the	O
functional	O
MEG	O
data	O
onto	O
the	O
structural	O
MRI	O
data	O
(	O
"	O
coregistration	B-Algorithm
"	O
)	O
is	O
possible	O
.	O
</s>
<s>
Independent	B-Algorithm
component	I-Algorithm
analysis	I-Algorithm
(	O
ICA	O
)	O
is	O
another	O
signal	O
processing	O
solution	O
that	O
separates	O
different	O
signals	O
that	O
are	O
statistically	O
independent	O
in	O
time	O
.	O
</s>
<s>
from	O
MEG	O
and	O
EEG	B-Application
signals	O
that	O
may	O
be	O
contaminated	O
with	O
outside	O
noise	O
.	O
</s>
<s>
MEG	O
can	O
resolve	O
events	O
with	O
a	O
precision	O
of	O
10	O
milliseconds	O
or	O
faster	O
,	O
while	O
functional	B-Algorithm
magnetic	I-Algorithm
resonance	I-Algorithm
imaging	I-Algorithm
(	O
fMRI	B-Algorithm
)	O
,	O
which	O
depends	O
on	O
changes	O
in	O
blood	O
flow	O
,	O
can	O
at	O
best	O
resolve	O
events	O
with	O
a	O
precision	O
of	O
several	O
hundred	O
milliseconds	O
.	O
</s>
<s>
For	O
creating	O
functional	O
maps	O
of	O
human	O
cortex	O
during	O
more	O
complex	O
cognitive	O
tasks	O
,	O
MEG	O
is	O
most	O
often	O
combined	O
with	O
fMRI	B-Algorithm
,	O
as	O
the	O
methods	O
complement	O
each	O
other	O
.	O
</s>
<s>
Neuronal	O
(	O
MEG	O
)	O
and	O
hemodynamic	O
fMRI	B-Algorithm
data	O
do	O
not	O
necessarily	O
agree	O
,	O
in	O
spite	O
of	O
the	O
tight	O
relationship	O
between	O
local	O
field	O
potentials	O
(	O
LFP	O
)	O
and	O
blood	O
oxygenation	O
level-dependent	O
(	O
BOLD	O
)	O
signals	O
.	O
</s>
<s>
The	O
responses	O
in	O
the	O
brain	O
before	O
,	O
during	O
,	O
and	O
after	O
the	O
introduction	O
of	O
such	O
stimuli/movement	O
can	O
then	O
be	O
mapped	O
with	O
greater	O
spatial	O
resolution	O
than	O
was	O
previously	O
possible	O
with	O
EEG	B-Application
.	O
</s>
<s>
There	O
has	O
been	O
great	O
success	O
isolating	O
unique	O
responses	O
in	O
patients	O
with	O
schizophrenia	B-Application
,	O
such	O
as	O
auditory	O
gating	O
deficits	O
to	O
human	O
voices	O
.	O
</s>
<s>
Recent	O
studies	O
have	O
reported	O
successful	O
classification	O
of	O
patients	O
with	O
multiple	O
sclerosis	O
,	O
Alzheimer	O
's	O
disease	O
,	O
schizophrenia	B-Application
,	O
Sjögren	O
's	O
syndrome	O
,	O
chronic	O
alcoholism	O
,	O
facial	O
pain	O
and	O
thalamocortical	O
dysrhythmias	O
.	O
</s>
<s>
Based	O
on	O
its	O
perfect	O
temporal	B-General_Concept
resolution	I-General_Concept
,	O
magnetoencephalography	B-Algorithm
(	O
MEG	O
)	O
is	O
now	O
heavily	O
used	O
to	O
study	O
oscillatory	O
activity	O
in	O
the	O
brain	O
,	O
both	O
in	O
terms	O
of	O
local	O
neural	O
synchrony	O
and	O
cross-area	O
synchronisation	O
.	O
</s>
<s>
Magnetoencephalography	B-Algorithm
can	O
also	O
be	O
used	O
to	O
study	O
changes	O
in	O
neural	B-Algorithm
oscillations	I-Algorithm
across	O
different	O
stages	O
of	O
consciousness	O
,	O
such	O
as	O
in	O
sleep	O
.	O
</s>
<s>
MEG	O
has	O
been	O
in	O
development	O
since	O
the	O
1960s	O
but	O
has	O
been	O
greatly	O
aided	O
by	O
recent	O
advances	O
in	O
computing	O
algorithms	O
and	O
hardware	O
,	O
and	O
promises	O
improved	O
spatial	O
resolution	O
coupled	O
with	O
extremely	O
high	O
temporal	B-General_Concept
resolution	I-General_Concept
(	O
better	O
than	O
1	O
ms	O
)	O
.	O
</s>
<s>
Since	O
the	O
MEG	O
signal	O
is	O
a	O
direct	O
measure	O
of	O
neuronal	O
activity	O
,	O
its	O
temporal	B-General_Concept
resolution	I-General_Concept
is	O
comparable	O
with	O
that	O
of	O
intracranial	O
electrodes	O
.	O
</s>
<s>
MEG	O
complements	O
other	O
brain	B-Application
activity	I-Application
measurement	O
techniques	O
such	O
as	O
electroencephalography	B-Application
(	O
EEG	B-Application
)	O
,	O
positron	B-Application
emission	I-Application
tomography	I-Application
(	O
PET	O
)	O
,	O
and	O
fMRI	B-Algorithm
.	O
</s>
<s>
Its	O
strengths	O
consist	O
in	O
independence	O
of	O
head	O
geometry	O
compared	O
to	O
EEG	B-Application
(	O
unless	O
ferromagnetic	O
implants	O
are	O
present	O
)	O
,	O
non-invasiveness	O
,	O
use	O
of	O
no	O
ionizing	O
radiation	O
,	O
as	O
opposed	O
to	O
PET	O
and	O
high	O
temporal	B-General_Concept
resolution	I-General_Concept
as	O
opposed	O
to	O
fMRI	B-Algorithm
.	O
</s>
<s>
Although	O
EEG	B-Application
and	O
MEG	O
signals	O
originate	O
from	O
the	O
same	O
neurophysiological	O
processes	O
,	O
there	O
are	O
important	O
differences	O
.	O
</s>
<s>
Whereas	O
scalp	O
EEG	B-Application
is	O
sensitive	O
to	O
both	O
tangential	O
and	O
radial	O
components	O
of	O
a	O
current	O
source	O
in	O
a	O
spherical	O
volume	O
conductor	O
,	O
MEG	O
detects	O
only	O
its	O
tangential	O
components	O
.	O
</s>
<s>
Scalp	O
EEG	B-Application
can	O
,	O
therefore	O
,	O
detect	O
activity	O
both	O
in	O
the	O
sulci	O
and	O
at	O
the	O
top	O
of	O
the	O
cortical	O
gyri	O
,	O
whereas	O
MEG	O
is	O
most	O
sensitive	O
to	O
activity	O
originating	O
in	O
sulci	O
.	O
</s>
<s>
EEG	B-Application
is	O
,	O
therefore	O
,	O
sensitive	O
to	O
activity	O
in	O
more	O
brain	O
areas	O
,	O
but	O
activity	O
that	O
is	O
visible	O
in	O
MEG	O
can	O
also	O
be	O
localized	O
with	O
more	O
accuracy	O
.	O
</s>
<s>
Scalp	O
EEG	B-Application
is	O
sensitive	O
to	O
extracellular	O
volume	O
currents	O
produced	O
by	O
postsynaptic	O
potentials	O
.	O
</s>
<s>
MEG	O
detects	O
intracellular	O
currents	O
associated	O
primarily	O
with	O
these	O
synaptic	B-Application
potentials	O
because	O
the	O
field	O
components	O
generated	O
by	O
volume	O
currents	O
tend	O
to	O
cancel	O
out	O
in	O
a	O
spherical	O
volume	O
conductor	O
.	O
</s>
<s>
Finally	O
,	O
MEG	O
is	O
reference-free	O
,	O
while	O
scalp	O
EEG	B-Application
relies	O
on	O
a	O
reference	O
that	O
,	O
when	O
active	O
,	O
makes	O
interpretation	O
of	O
the	O
data	O
difficult	O
.	O
</s>
