Psy 5065 - fMRI: Hands-on Training
FINAL EXAM
Date assigned: May 4, 2015
Date due: May 13, 2015
Each problem: 5 points
Total exam: 15% of course grade
Problem 1: Steady-state magnetization, SNR & Bandwidth
A) 1 point. What is the Ernst angle? (Describe what the term means.)
B) 1 point. You develop a sudden interest in functional activation in the basal ganglia,
and a colleague mentions that T1 is different in the caudate nucleus than it is in most
gray matter. A PubMed search on "high field T1" turns up a paper describing a pulse
sequence, DESPOT1-HIFI [Deoni, S. (2007), J. Mag Res Imaging 26:1106], which
reports that T1 in the caudate and putamen -- the 2 subregions of the basal ganglia that
you're most interested in -- is 1600ms. What flip angle will you use for an fMRI
experiment with a gradient echo EPI pulse sequence that has a TR of 1.5 sec?
What flip angle would you have used for cortex (T 1=1100ms)?
C) 0.5 point. Why does the scanner automatically throw away the first 2 or 3 volumes?
D) 0.5 point. Why is the gray/white contrast in the first volume of an EPI series often
better than in subsequent volumes?
Page 1 of 6
Psy 5065 - fMRI: Hands-on Training
E) 1 point. You are comparing the SNR between high-resolution and low-resolution
acquisitions. If the sampling bandwidth, matrix size and slice thickness are not different,
but the field of view was decreased for the high-res acquisition to increase the
resolution (decrease the voxel size) by a factor of 2 in both the phase-encode and readout directions, what is the thermal signal-to-noise ratio (SNR) in the new (small) voxels if
the SNR in the big voxels was 100. (Hint: you can assume, since the matrix size and
bandwidth are unchanged, that the thermal noise is the same in both acquisitions. So
the denominator is the same, and the SNR will be determined by the signal strength in
each voxel … which you can assume is proportional to volume.)
F) 1 point. If you increase the bandwidth …
… is image acquisition faster or slower?
… will distortion get better or worse?
… on an EPI acquisition, will the pitch (frequency) of the noise that the scanner makes
increase or decrease?
… will SNR get better or worse?
Problem 2: Distortion & Drop-out
A) 2 points. You run an fMRI experiment and acquire a fieldmap at the end of the hourlong scanning session. The subject has, however, moved her head since the beginning
of the session - her chin has rotated down a bit. Will the fieldmap be useful for
predicting/correcting distortions in all of the EPI images acquired since the beginning of
the scan? Explain/justify your answer.
Page 2 of 6
Psy 5065 - fMRI: Hands-on Training
B) 0.75 points. To include fieldmap-based distortion compensation in your data analysis
stream (usually considered a pre-processing step), what 3 pieces of information do you
need, in addition to your EPI images and your fieldmap? Describe briefly what each is
used for. (Hint: you can jog your memory by looking at commands you used in HW9 or
the lecture from Apr 13.)
C) 0.25 points. If you want to use TOPUP to do distortion compensation, instead of
doing fieldmap-based distortion compensation, what kind of image acquisition do you
need to include in your scanning session?
D) 0.5 points. Label the phase encode direction on each image.
Superior
Right
Left
Inferior
E) 1.5 points. List 3 strategies for avoiding signal loss due to intra-voxel (usually
through-slice) dephasing. Describe a strength and limitation for each approach.
Page 3 of 6
Psy 5065 - fMRI: Hands-on Training
Problem 3: Experiment design
A) 1 point. You are investigating a hypothesis about how the brain incorporates spatial
information in a decision-making process -- for this you need to cover the entire brain
(excluding cerebellum, but not missing bits of temporal or parietal cortex) with 1.5 s
temporal resolution. With current technology and techniques, what kind of spatial
resolution can you expect?
B) 1.5 points. A colleague of yours is interested in the visual representation of
information in inferior temporal cortex. She needs both good spatial accuracy (low
distortion) and uniform sensitivity throughout inferior temporal cortex. She is, however,
working with a subject population that will not tolerate tricks like filling the auditory
canals with saline solution (imagine that …), and she wants high contrast-to-noise ratio
(so she doesn't want to use a spin echo pulse sequence at 3T).
What will you recommend she try in order to minimize or compensate for distortion?
What slice orientation will you recommend she use? Justify your answer (there's no
right answer).
What will you recommend she do to minimize trouble with signal loss due to throughslice dephasing?
C) 2.5 points. You and a colleague are collaborating on a project to measure
lateralization of spatial working memory on a number-sequence memory task. In
Page 4 of 6
Psy 5065 - fMRI: Hands-on Training
addition to running 2 10-minute functional scans, your collaborator also wants to acquire
DTI data to look for possible correlation between behavior and fractional anisotropy in
prefrontal white matter, as well as connectivity between DLPFC and inferior frontal
cortex, and performance on the task. The table below gives you the sequences to
which you have access to build a scanning protocol for your colleague.
Base Sequence
3D MP-RAGE
3D MP-RAGE
GE FLASH
DW SE EPI #1
nDir = 35,
bVal=800
DW SE EPI #2
nDir = 35,
bVal=1000
DW SE EPI #3
nDir = 18,
bVal=1200
GE FLASH, dual
echo time #1
GE FLASH, dual
echo time #2
GE EPI #1
TE=30 ms
GE EPI #2
TE=30 ms
GE EPI #3
TE=30 ms
GE EPI #4
TE=30 ms
FOV
(read x
phase)
256 x
224
256 x
224
256 x
256
Resolution
(mm)
Slice
thk
(mm)
1
BW
(Hz /
px)
700
PE
dir
N
Slices
Orient
Tacq
(m:s)
F/H
Echo
spacing
(ms)
2.3
176
sag
7:18
1x1
(iPAT=2)
2x2
1
700
F/H
2.3
176
sag
4:18
2
700
2.3
3
2
2604
0.47
72
1 sag,
1 cor,
1 ax.
ax
0:10
2x2
(iPAT=2)
F/H
F/H
A/P
P/A
256 x
256
256 x
256
2x2
(iPAT=2)
2
2604
P/A
0.47
72
ax
5:42
256 x
256
2x2
(iPAT=2)
2
2604
P/A
0.47
72
ax
3:25
256 x
256
256 x
256
256 x
256
256 x
192
256 x
256
256 x
256
2x2
2
700
A/P
2.3
72
ax
2:07
3x3
3
700
A/P
1.6
32
1:03
3x3
3
2604
A/P
0.47
32
T->C
30
ax
3x3
3
2604
R/L
0.47
32
3x3
3
2604
A/P
0.47
32
3x3
2, w/
dist=
50%
2604
A/P
0.47
32
1x1
T->C
30
T->C
30
T->C
30
5:42
TR =
1.5 s
TR =
1.5 s
TR =
1.5 s
TR =
1.5 s
In the table below, enter your selected sequences in their selected order. Just
underneath the table, there's space for you to calculate how long the subject will spend
in the scanner, and how much time you should book on the 3T calendar for each
experiment (allowing 15 min. set-up time and 10 min. clean-up time). Then there are 3
questions for you to answer about the pulse sequence options.
After the specified questions, you can make any notes you want to explain your
selections, or list any questions you want to ask your collaborator before starting to run
subjects on this protocol.
Scan type (specify 3-axis localizer,
anatomy, fMRI, DTI or fieldmap)
Selected sequence (name in left column of
table above)
Page 5 of 6
Psy 5065 - fMRI: Hands-on Training
Total scanning session duration (to the nearest 5 minutes, allowing for ~30 s pauses
between sequences):
Total time booked on scanner (round up to the next 15-min. increment):
Which of the MP-RAGE sequences has better signal-to-noise ratio?
Which of the GE EPI (fMRI) sequences has the most signal loss in orbitofrontal cortex?
Which of the DTI sequences uses the strongest diffusion gradients?
Extra credit (1 point)
Under what circumstances might you be motivated to use parallel imaging? What
hardware is required for parallel imaging?
Page 6 of 6