Thomas R. Gregg
Department of Neurosciences
MSB/H506, Newark campus, UMDNJ
Final paper for Methods in Neuroscience course
Dec. 23, 1996
Use of functional magnetic resonance imaging to investigate
the involvement of telencephalic brain areas in the
emotion of fear
Functional magnetic resonance imaging (fMRI) is a tool used
to visualize brain function, by visualizing changes in chemical
composition of brain areas or changes in the flow of fluids that
occur over timespans of seconds to minutes. This method can also
be used to study the physiology of other organs, including
studying blood flow to pathological organs, thus helping us to
understand the disease process. In the brain, blood perfusion is
presumably related to neural activity, so fMRI, like other
imaging techniques such as PET, can be used to find out what the
brain is doing when subjects perform specific tasks or are
exposed to specific stimuli. However, fMRI has a better temporal
and spatial resolution than PET (Cohen & Bookheimer, 1994).
In what follows, I will first describe the general effect
exploited in nuclear magnetic resonance studies, and describe how
2 major components of MRI technology (the main magnetic field and
the radiofrequency pulse) relate to this effect. Then I will
describe how magnetic field gradients (which are superimposed on
the main magnetic field) and Fourier analysis allow imaging of
tissues in 3 dimensions. Last, I will describe fMRI and an
experiment using fMRI to investigate the brain regions involved
in fear and phobias in humans.
MAGNETIC RESONANCE OF ATOMIC NUCLEI
Functional MRI is a new use of existing MRI technology. The
basic phenomenon of nuclear magnetic resonance has been known
since the 1940s (Le Bihan, 1995), and MRI has been developed over
the last 30 years (Cohen & Bookheimer, 1994).
Magnetic resonance can be adequately understood in terms of
electromagnetic theory, as follows. All atomic nuclei spin on
their axes; nuclei have a positive electronic charge; and any
spinning charged particle will act as a magnet with north and
south poles located on the axis of spin. In magnetic resonance
studies, an object is put in a strong, externally-imposed
magnetic field ("main magnetic field"); the spin-axes of all the
nuclei in the object line up with the field, with the north poles
of the nuclei pointing in the "southward" direction of the field.
This creates an average vector of magnetization of the object
that points parallel to the magnetic field (the main magnetic
field is conventionally referred to as pointing along the z-axis)
(Horowitz, 1995).
Then a radiofrequency (RF) pulse is broadcast toward the
object in a line perpendicular to the magnetization vector. The
RF pulse causes the axes of the nuclei to tilt with respect to
the main magnetic field, thus causing the net magnetization
vector to deviate from the main magnetic field by a certain
angle. However, only those nuclei which precess about their axes
at the RF pulse frequency will be affected by the pulse; in other
words, the nuclei that "resonate" to that frequency will be
affected (Horowitz, 1995).
The net magnetization vector gradually (over 20-300 msec)
returns to the state of being parallel with the external magnetic
field, and the time that this takes is called the T2 relaxation
time or "spin-spin relaxation time" after deactivation of the RF
pulse. The amount by which the magnetization vector tilts away
from the z-axis is controlled by the intensity and duration of
the RF pulse; for example, if a 5 msec pulse at a certain
intensity caused it to deviate 90 degrees from the z-axis, then a
10 msec pulse would cause a 180 degree deviation. In MRI studies
on biological tissue, hydrogen nuclei are examined; T2 relaxation
time of these nuclei differs from tissue to tissue (Horowitz,
1995).
The RF pulse also increases the angle of precession of the
nuclei about their axes. This increase of precession angle does
not affect the direction of the net magnetization vector (which
is in the z direction), but it does affect the strength of the
magnetization vector, since the nuclei's spin axes now deviate
more from the z-axis than they did before the pulse, therefore
the standard deviation of this angle is high, although the mean
of the angle is zero. Over time, as the standard deviation of
the precession angle decreases, the strength of the net
magnetization vector increases, since all the nuclei are
gradually coming to point in the z-direction again. In other
words, over time, the angle of precession of the axes declines
back to its resting state (so that the average angle of
precession is zero and the standard deviation of the angle of
precession is low). Another way to say this is that over time,
the average angle at which the nuclei's spin axes deviate from
the z direction returns to its normal low value. The time that
this takes is called "T1 relaxation time" or "spin-lattice
relaxation time", and usually lasts between 200 and 2,000 msec
(Horowitz, 1995).
As the nuclei relax, each becomes a miniature radio
transmitter, giving out a characteristic pulse that changes over
time, depending on the local microenvironment surrounding the
proton. For example, hydrogen nuclei in fats have a different
microenvironment than do those in water, and thus transmit
different pulses. Due to these differences, in conjunction with
the different water-to-fat ratios of different tissues, different
tissues transmit different radio signals. These miniature radio
transmissions can be used to form MRI images (Horowitz, 1995).
Such is the general magnetic resonance procedure, by which
the chemical composition of objects can be determined. MRI, as
usually used in biomedical science, has two additional
characteristics: 1) a 3-dimensional picture of the object is
obtained, in addition to the chemical composition of the object;
2) the nuclei of hydrogen atoms are usually imaged (Horowitz,
1995).
SPIN-ECHO MAGNETIC RESONANCE IMAGING
In spin-echo MRI, gradients and Fourier analysis are used to
perform three-dimensional imaging. Other techniques of MRI, such
as gradient-echo, are slight variations of spin-echo imaging, so
I will only describe spin-echo imaging in detail. The component
of the imaging system which allows the spatial localization of
the protons is a set of magnetic field gradients, set up by
magnetic coils which are turned on and off at appropriate times
(Horowitz, 1995).
When hydrogen nuclei relax, the frequency that they transmit
is positively correlated with the strength of the magnetic field
surrounding them. A magnetic field gradient along the z-axis,
called the "slice select gradient," is set up when the RF pulse
is applied, and is shut off when the RF pulse is turned off.
This gradient causes the hydrogen nuclei at the high end of the
gradient (where the magnetic field is strong) to precess at a
high frequency (e.g., 65 MHz), and those at the low end (weak
field) to precess at a lower frequency (e.g., 63 MHz). When the
RF pulse, of a single frequency, is applied, only those nuclei
which precess at that frequency will be tilted, to later relax
and emit a radio transmission (i.e., the nuclei "resonate" to
that frequency). For example, if the magnetic gradient caused
hydrogen nuclei to precess at rates from 63 MHz at the low end of
the gradient to 65 MHz at the high end, and the gradient were set
up such that the high end was located at the patient's head and
the bottom part at the patient's feet, then a 63 MHz RF pulse
would excite the hydrogen nuclei in a slice near the feet, and a
65 MHz pulse would excite them in a slice near the head. Thus a
single "slice" along the z-axis is selected; only the protons in
this slice are excited to a higher energy level, to later relax
to a lower energy level and emit a radio transmission (Horowitz,
1995).
The second dimension of the image is extracted with the help
of a phase encoding gradient. Immediately after the RF pulse
ceases, all of the nuclei in the activated slice are "in phase,"
that is, their magnetic vectors all point in the same direction.
Left to their own devices, these vectors would relax. In MRI,
however, the phase encoding gradient (in the y-dimension) is
briefly applied, in order to cause the magnetic vectors of nuclei
along different portions of the gradient to point in different
directions (Horowitz, 1995).
After the RF pulse, slice select gradient, and phase
encoding gradient have been turned off, the MRI instrument sets
up a third magnetic field gradient, along the x axis, called the
"frequency gradient" or "read-out gradient". This gradient
causes the relaxing protons to be differentially re-excited, so
that the nuclei near the low end of the gradient begin to precess
at a faster rate, and those at the high end pick up even more
speed. When these nuclei relax again, the fastest ones (those
which were at the high end of the gradient) will emit the highest
frequency of radio waves. The frequency gradient is applied
"only when the signal is measured" (Horowitz, 1995).
The second and third dimensions of the image are extracted
by means of Fourier analysis. The entire procedure must be
repeated multiple times in order to form an image with a good
signal-to-noise ratio.
Finally, in spin-echo imaging, there is the problem that the
inhomogeneity of the main magnetic field induces variations in
the rate of precession of nuclei. To fix this problem, a
180-degree RF pulse is inserted into the cycle, at a time point
halfway between the 90-degree pulse and the measurement of the
radio transmission signal given off by the relaxing nuclei
(Horowitz, 1995).
OTHER MRI TECHNIQUES
When 90-degree and 180-degree RF pulses are used, as
described above, the technique is called "spin echo imaging", and
takes several minutes to create a single image. According to
Horowitz (1995), "Nearly every MR scan makes use of at least one
spin-echo pulse sequence, and more specifically, the double echo
spin-echo is probably the workhorse of the clinical MR imaging
world" (p. 44.) The techniques which can be used in fMRI include
gradient echo imaging (in which spin-echo's 180-degree pulse is
replaced by a "reversal of magnetic field gradients"; Heiken &
Brown, 1991), echo planar imaging (a variation of spin-echo, also
called "fast spin echo imaging"), and spin-echo inversion
recovery imaging.
FUNCTIONAL MRI
The term "functional MRI" can include the technique of
co-registering PET and MRI scans, but it is usually used to
denote techniques involving fast MRI scans, which can allow
imaging of a complete brain slice in 20 ms (Cohen & Bookheimer,
1994). The first fMRI of the brain (a perfusion MRI) was done in
1991 by Belliveau and co-workers, who injected a chemical that
increases MRI contrast into a patient and imaged the brain using
echo-planar techniques, to show that the perception of visual
stimuli increases blood volume in primary visual cortex
(Belliveau, et al., 1991; Rosen, et al., 1991). The same group
later used gradient echo and spin-echo inversion recovery fMRI to
examine blood oxygenation levels and blood flow rates,
respectively, in brain (Kwong, et al., 1992).
The principle of fMRI imaging is to take a series of images
of the brain in quick succession and to statistically analyze the
images for differences among them. For example, in the
blood-oxygen level dependent method (BOLD; Ogawa, et al., 1990),
the fact that hemoglobin and deoxyhemoglobin are magnetically
different is exploited. Hemoglobin shows up better on MRI images
than deoxyhemoglobin; thus, oxygenated blood shows up better.
Brain areas with more blood flow have been shown to have better
visibility on MRI images (Cohen & Bookheimer, 1994). Therefore,
better visibility is thought to be correlated with brain
activation. This has been exploited in the following type of
procedure: a series of baseline images are taken of the brain
region of interest when the subject is at rest; the subject
performs a task and a second series is taken; then the first set
of images is subtracted from the second, and the areas that are
most visible in the resulting image are presumed to have been
activated by the task. Other fMRI methods exploit the fact that
the bulk movement of hydrogen nuclei causes changes in the MRI
signal. Such methods could image CSF flow, blood flow, or
diffusion of water through tissue. Care must be taken not to
move the head, since spurious results could occur due to movement
artifacts (Cohen & Bookheimer, 1994).
PROPOSED EXPERIMENT: NEURAL BASIS OF FEAR
What happens in the brain when a person sees a
fear-provoking stimulus and becomes afraid? The functional
neurophysiology of fear could be investigated using fMRI, as
follows: First, an estimate of effect size would be made to
determine how many subjects would be needed for the experiment.
Next, many people would be screened by use of a questionnaire to
find 1) adults who have a snake phobia strong enough to cause
them to fear snake photographs, and 2) an equal number of
controls. The phobia should be long-standing and severe, to
ensure that the experimental procedure would not cause
desensitization, and to ensure that the brain areas of interest
would be activated enough to produce a change in blood flow. The
questionnaire would ask such questions as, "do you have severe
fear of snakes?," and "do you get nervous when you see pictures
of brown snakes?", to be rated by the subject on a Likert scale.
The questions would be based on the American Psychiatric
Association's DSM-IV criteria for phobias. The subjects could
then be evaluated by a neurologist and psychiatrist to ensure
that no other psychological or neurological problems were
present, and evaluated to ensure the absence of metal implants.
Subjects would be paid for their participation and would give
informed consent for all procedures.
In pre-scanning trials in a doctor's-office setting,
subjects would be shown various pictures of snakes, and their
physiological responses such as heart rate spectrum, galvanic
skin response, respiration, and muscle tone, as well as
subjective psychological verbal responses, would be recorded.
These pre-scanning measures would be taken to ensure that
subjects felt intense fear when snake photographs were presented;
to determine the minimum duration of exposure to the picture
needed to generate intense fear; to determine the time course of
the fear, so that the fMRI scan could be taken at the time at
which fear peaked; and to determine the time of cessation of
fear.
The experiment would have the purpose of establishing which
brain areas are activated by fear. Accordingly, the amygdala,
lateral geniculate nucleus, visual cortex, cingulate cortex, and
auditory cortex would be imaged both in fear conditions (arousal
of intense fear after snake picture presentation), and in no-fear
conditions, in which non-snake phobics were presented with
pictures of snakes or other objects, or the phobics were
presented with pictures of non-fear-producing objects.
The fMRI scanning procedure would be as follows: a subject
would be habituated to the scanning procedure for the shortest
length of time needed to ensure that the procedure did not arouse
any unwanted emotions. The subject would be put in the scanner
and rest 60 seconds, after which a conventional MRI would be
taken to establish coordinates; then the subject would rest 10
minutes, during which time the locations of the target areas
would be found and 15 "slices" through those areas would be
decided upon. A video camera would record changes in the
subject's facial musculature during photo presentation, both to
provide a measure of fear levels and to ensure that the subject
was looking at the stimulus. Then a snake picture would be
projected on a screen in front of the subject, and several sets
of fMRI images would be acquired, before, during, and after the
period of most intense fear, using acquisition parameters
appropriate for the BOLD method; then the subject would rest for
long enough for the fear to subside (as previously determined)
and then would be shown a picture of a neutral animal (the order
of presentation would be counterbalanced, so that half the
subjects would get the neutral picture first). Heart rate would
be measured (through a long, non-metallic stethoscope leading out
of the MRI chamber), and respiration volume and rate would be
monitored through plastic tubes to show that the snake-phobic
subjects' heart spectra and respiration rates did change in the
proper ways when the snake pictures were presented, thus proving
that the pictures did evoke fear, but the no-fear conditions did
not evoke fear. After the final acquisition, subjects would be
paid, given their brain image, and released.
The fMRI images would be statistically analyzed, allowing us
to determine whether; by how much; and at what time the relevant
structures were activated. We would expect that the lateral
geniculate and visual cortex would be activated by presentation
of all pictures; auditory cortex would not be activated by any
pictures; and amygdala and cingulate cortex would be activated by
snake pictures in phobics but not controls. If so, further
experiments could determine the temporal order in which amygdala
and cingulate become activated. REFERENCES
Belliveau, J. W., Kennedy D. N. Jr., McKinstry, R. C.,Buchbinder,
B. R., Weisskopf, R. M., Cohen, M. S., et al. (1991). Functional
mapping of the human visual cortex by magnetic resonance imaging.
Science, 254:716-719.
Cohen, M. S. & Bookheimer, S. Y. (1994). Localization of
brainfunction using magnetic resonance imaging. Techniques in
Neuroscience, 17(7):268-277.
Heiken, J.P & Brown, J. J. (Eds.) (1991). Manual of
ClinicalMagnetic Resonance Imaging (2nd ed.). New York: Raven.
Horowitz, A. L. (1995). MRI Physics for Radiologists: A
VisualApproach (3rd ed.). New York: Springer-Verlag.
Le Bihan, D. (moderator) (1995). NIH conference:
Functionalmagnetic resonance imaging of the brain. Annals of
Internal Medicine, 122:296-303.
Kwong, K. K., Belliveau, J. W., Chesler, D. A., Goldberg, I.
E.,Weisskopf, R. M., Poncelet, B. P., et al. (1992). Dynamic
magnetic resonance imaging of human brain activity during primary
sensory stimulation. Proceedings of the National Academy of
Sciences USA, 89:5675-5679.
Ogawa, S., Lee, T. M., Kay, A. R., & Tank, D. W. (1990).
Brainmagnetic resonance imaging with contrast dependent on blood
oxygenation. Proceedings of the National Academy of Sciences
USA, 87:9868-9872.
Rosen, B. R., Belliveau, J. W., Aronen, H. J., Kennedy,
D.,Buchbinder, B. R., Fischman, A., Gruber, M., Glas, J,
Weisskopf, R. M., Cohen, M. S., et al., (1991). Susceptibility
contrast imaging of cerebral blood volume: human experience.
Magnetic Resonance in Medicine, 22(2):293-299.
(
geocities.com/soho)