See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/275274683

Body MRI sequences: A conceptual framework

Article in Applied Radiology · January 2012

CITATION READS

1 63

1 author:

Eugene C Lin
Virginia Mason Medical Center
56 PUBLICATIONS 539 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Eugene C Lin
letting you access and read them immediately. Retrieved on: 23 November 2016
Body MRI sequences:
A conceptual framework
Eugene Lin, MD

B
ody magnetic resonance imag- potential confusion by thinking of the which greatly decreases imaging time.
ing (MRI) typically presents sequences generically.1 For example, However, for T2-weighted imaging,
more challenges, from the per- the same 3-dimensional spoiled gradi- gradient-echo imaging presents several
spective of imaging sequences, than ent-echo sequence will have numerous problems. Without the refocusing pulse,
MR imaging of the neurological and trade names attached to it by different the images are now T2* rather than
musculoskeletal systems. This is be- manufacturers. However, referring to T2-weighted, and we no longer have
cause to adequately image many of the sequence as a 3-dimensional spoiled the proper weighting. In addition, as
the organs, the patient must hold his gradient-echo sequence avoids potential relatively high TE values are needed for
breath, which limits the amount of time confusion and is usually much more de- T2-weighted imaging, the T2* weight-
available for imaging, or the organ it- scriptive. That is, referring to a sequence ing will often result in substantial sus-
self (the heart) is moving. Many of the generically tells us almost everything ceptibility artifacts and the sequence
sequences used in body MRI may be one needs to know about the sequence, may no longer be robust. So it appears
less familiar to radiologists than the se- while the trade names often give little that in order to generate a robust, truly
quences used in the neurological and information about the sequence other T2-weighted image, the 180-degree re-
musculoskeletal systems. The goal of than a memorable catchphrase. Table focusing pulse should be retained.
this article is to present a broad concep- 1 lists generic sequences and common  The other basic modification to the
tual and nontechnical framework that trade names. The sequences we want to spin-echo sequence that can be made
can serve as a basis for understanding use should be fast, robust (meaning they is to acquire multiple lines of k-space
these sequences. For the purposes of generate diagnostic-quality images the during the TE interval. In a spin-echo
brevity, this article will assume a basic majority of the time), and have the sequence with a TE of milliseconds
knowledge of the physics of and se- proper weighting.  (msec), the time between 0 and 100
quences employed in MRI. msec, when one line of k-space is
    In the author’s experience, a major Fast T2-weighted sequences  filled, is “wasted” time in which noth-
barrier for radiologists learning body  Let’s start with finding a T2- ing is being acquired. If 10 lines of
MRI sequences is the proliferation of weighted sequence, which is fast, ro- k-space (an “echo train” of 10) are
trade names used by different manu- bust, and has the proper weighting. filled during this time (eg, at 10, 20,
facturers. The use of these trade names While there are many ways to increase and up to 100 msec), the sequence can
in the published literature may con- the speed of a sequence once we have be completed 10 times as quickly. This
tribute to the confusion. However, the decided upon the sequence—eg, in- is called fast spin-echo imaging. I will
number of ways to generate an image creasing bandwidth or decreasing rep- use this more common generic term,
in body MRI is actually limited, and etition time (TR)—starting from the although it is also a trade name; a more
the radiologist can avoid much of the basic spin-echo sequence, there are 2 generic but less familiar term for the
fundamental modifications that can be same sequence is rapid relaxation with
Dr. Lin is a Clinical Assistant Professor made. One of these is to eliminate the relaxation enhancement (RARE). How
of Radiology, Department of Radiology
180-degree refocusing pulse — that is, well does this work for T2-weighted
at the University of Washington Medical
Center and an Attending Physician at the gradient-echo imaging. Without the re- imaging? Very well, as the reten-
Virginia Mason Medical Center, Seattle, focusing pulse, the TR and echo time tion of the 180-degree pulse gives us
WA. (TE) can be substantially decreased, both T2-weighting and decreases

16 January–February 2012
©

n APPLIED RADIOLOGY www.appliedradiology.com

 BODY MRI SEQUENCES

Table 1. Common Body MRI Sequences

Siemens GE Philips Toshiba
Fast spin-echo Turbo SE Fast SE Turbo SE Fast SE

Single-shot fast spin-echo with half-Fourier acquisition HASTE SS-FSE SS-FSE FASE

2D spoiled gradient-echo FLASH SPGR T1-FFE T1-FFE

3D spoiled gradient-echo* VIBE FAME THRIVE 3D QUICK

Balanced steady state gradient-echo TrueFISP FIESTA Balanced FFE True-SSFP

GE = GE Healthcare, a division of General Electric Company.

* 3-dimensional spoiled gradient-echo sequences can be used for a wide variety of applications. The sequences listed here are optimized for
nonvascular body MRI.

previous example of a sequence with a suggests that the use of very long echo
TE of 100 msec and echo train of 10, trains to image fluid is optimal. On the
each of the 10 lines of k-space filled other hand, solid tissue has a shorter T2-
actually has a different TE, ranging relaxation time, and the difference in
from 10 to 100 msec and thus a dif- signal intensity of solid tissue between
ferent signal intensity, and there is no 10 and 100 msec would be substantially
actual TE for the sequence as a whole. greater than that of fluid. So image blur-
In practice, the actual weighting of a ring in solid tissues might increase sub-
fast-spin echo sequence is determined stantially at higher echo trains. During
by the effective TE. As the center of k- magnetic resonance cholangiopancrea-
FIGURE 1. Single-shot fast-spin echo. space is important for signal-to-noise tography (MRCP) (Figure 1), a long
Coronal single-shot fast-spin echo with half- ratio and the periphery of k-space is echo train allows for very fast imag-
Fourier acquisition (TR 836 msec, TE 88 important for image detail, the TE at ing of fluid, which is not substantially
which the center of k-space is obtained blurred. Solid tissue is substantially
msec) image from an MRCP study demon-
strates the common bile-duct and multiple-
liver metastases. The common duct is well (the effective TE) will determine the blurred, but this is not what is being pri-
visualized as there is minimal decay of fluid weighting. However, there is still a marily imaged. For other applications,
signal during the echo time. However, the problem with the 9 other lines of k- shorter echo trains at the expense of im-
T2-relaxation time of solid tissue is substan- space, which contributes to image de- aging time may be desirable, if image
tail. As these are obtained at different detail of solid tissues is important.
tially shorter, which results in greater signal
loss during the echo time interval as well
as a greater difference in signal intensity TEs and will have different signal in- A supplemental way of decreasing
between each line of k-space. This results tensities, the image will be blurred. If imaging time is the variety of tech-
in solid tissue that is low in signal intensity the echo train is longer, the image blur- niques that rely on alternate ways of
and blurred. The liver metastases are poorly ring will increase as the number of dif- filling k-space.2 These are essentially
ferent signal intensities increases.  shortcuts in k-space and decrease
visualized and this sequence would not be
optimal to detect metastases.
Thus, increasing the echo train to image time at the expense of signal-to-
susceptibility artifact, and the se- decrease imaging time will typically noise ratio (SNR). These techniques
quence can be very fast with a long- increase image blurring. How impor- can be used to decrease imaging time
echo train. This technique can also be tant is this blurring? It depends on what in both fast spin-echo and gradient-
used for gradient-echo imaging. In a is being imaged. Let’s consider fluid. echo imaging, and are best thought of
single shot, echoplanar gradient-echo Fluid has a long T2-relaxation time, so in the framework of those sequences
sequence, all of the k-space is filled in during a fast-spin echo sequence with rather than in isolation. One class of
one excitation. Echoplanar sequences a TE of 100 msec and an echo train of techniques uses nontraditional tra-
tend not to be very robust. 10, there is actually little difference jectories for filling k-space, eg, spiral
Now it might be asked that if the in the signal intensity of fluid between or radial. In another class of tech-
length of the echo train determines the 10 and 100 msec, as there has not been niques, not all of the k-space lines
speed of the sequence, why not always much T2 relaxation in this time inter- are acquired and the remainder of
use the longest echo train possible? val. Therefore, the effects of image k-space is filled with a partial Fourier
One problem is that if we consider the blurring on fluid are minimal. This method. Half-Fourier acquisition is a

January–February 2012 17
©

www.appliedradiology.com APPLIED RADIOLOGY n

BODY MRI SEQUENCES

FIGURE 2. Steady-state transverse magnetization. The initial excitation (A) flips longitudinally
into transverse magnetization. If TR is shorter than the T2-relaxation time, as it usually is for
fast gradient-echo imaging, there will be residual-transverse magnetization at the next excita- FIGURE 3. Steady-state gradient-echo. Cor-
tion (B). After the next excitation pulse (C), the residual-transverse magnetization is converted onal steady-state gradient-echo image (TR
into longitudinal magnetization, and the residual-longitudinal magnetization is converted into 4.7 msec, TE 2.3 msec, flip angle 70°) used
transverse magnetization. After multiple excitation pulses, there will be constant “steady- as a scout image as part of the same study
state” transverse and longitudinal magnetization. As the amount of residual-transverse mag- as Figure 6. Note that the flip angle is much
netization depends upon the T2-relaxation time, there is additional T2* weighting, which is higher than the spoiled gradient-echo image
undesirable for fast T1-weighted imaging. Steady state typically refers to the steady-state in Figure 6, even though the TR is identical.
transverse magnetization, as both steady-state and spoiled sequences have steady-state lon- Unlike spoiled gradient-echo sequences, the
gitudinal magnetization. optimal SNR on a steady-state sequence
does not depend upon the flip angle. Instead,
the T2/T1 ratio will increase with higher
flip angles. Therefore, high flip angles are
typically used in steady-state gradient-echo
imaging. Due to the T2/T1 weighting, only
fat and fluid are bright. The gallbladder and
common bile duct are well visualized, with
a lesser degree of visualization of the vas-
cular structures. However, the liver lesion
is not discretely visualized and can only be
detected by its mass effect (arrowheads).

In summary, T2-weighted sequences

in body MRI are typically performed
with a fast spin-echo sequence. In these
sequences, there is a trade-off between
imaging time and image blurring, which
is determined by the echo train. Very
fast T2-weighted images can be ob-
tained using a single-shot technique.
Additional k-space shortcuts can be
FIGURE 4. Spoiled gradient-echo sequence. T1-weighting is achieved through “spoiling” the used with fast spin-echo sequences,
residual-transverse magnetization. This can be done through a spoiler gradient (as in this which further decrease imaging time at
example) or radiofrequency pulse. the expense of SNR. 

commonly used technique that relies sequence. In a fast spin-echo technique, Fast T1-weighted sequences 
on acquiring slightly more than half of multiple lines of k-space are acquired in If we want to acquire a fast, robust
the k-space lines and filling in the miss- one excitation. A single-shot sequence study with T1-weighted image contrast,
ing data using the conjugate symmetry is one in which all of the k-space is ac- one option is a fast spin-echo sequence,
of k-space. Fluid-imaging techniques, quired in one excitation to maximize as used for T2-weighted imaging. How-
such as MRCP, typically will use this imaging speed and thus a large number ever, this does not work nearly as well
sequence: single-shot fast spin-echo of k-space lines must be filled. In order as with T2-weighted imaging. The
with half-Fourier acquisition. If we to accomplish this, a technique, half- substantially shorter TEs needed for
learn the generic name of the sequence, Fourier acquisition, where only half of T1-weighted image contrast limit the
it will tell us the essentials of the k-space is acquired is employed. increase in acquisition time possible.

18 January–February 2012
©

n APPLIED RADIOLOGY www.appliedradiology.com

 BODY MRI SEQUENCES

A B

FIGURE 5. Two-dimensional spoiled gradient-echo. Axial 2-dimensional spoiled gradient- FIGURE 6. Three-dimensional spoiled
echo images (A) in-phase (TR 152 msec, TE 4.8 msec, flip angle 70°) and (B) out-of-phase gradient-echo. Axial 3-dimensional fat-sat-
(TR 152 msec, TE 2.4, flip angle 70°) demonstrate fatty infiltration of the left lobe of the liver. It urated spoiled gradient-echo image (TR 4.7
is important to keep all parameters besides the TE the same for accurate comparison between msec, TE 2.3 msec, flip angle 10°) demon-
the 2 sequences. Note that the flip angle is much higher than the flip angle of the 3-dimen- strates an enhancing liver lesion. Compared
sional spoiled gradient-echo sequence in Figure 6, as the flip angle should increase with to the 2-dimensional sequence in Figure
increasing TR to maintain optimal SNR.  5, the 3-dimensional sequence has thin-
ner sections and a greater signal-to-noise
ratio. This particular sequence is optimized
for abdominal imaging. The highest signal
intensity is at the Ernst angle. [cos = exp
(-TR/T1)]. The flip angle of this sequence
is much smaller than the flip angle of the
2-dimensional sequence as the TR is much
lower. If this sequence were modified for MR
angiography, where the T1-relaxation time
of intravascular contrast is shorter, the flip
angle would need to be increased. Note that
even though the TE is out-of-phase, the typi-
cal “India ink” artifact is not seen because
the sequence is also fat-saturated. The fat-
saturation effect precedes the chemical shift
effect. Fat-saturation is optimally used in
combination with a low TE, as the SNR and
number of slices per TR are higher, but the
“India ink” artifact is not seen.
T1- and T2-relaxation times. Other tis-
sues will have substantially lower sig-
nals, as T2-relaxation times are much
FIGURE 7. Balanced steady-state gradient-echo sequence. Note the lack of the spoiler gradi-
ent used in Figure 4. The balanced gradients minimize dephasing and maintain the steady
shorter than T1-relaxation times. So we
state. will obtain images where fat and fluid is
bright, and everything else is isointense
For example, with a T1-weighted se- fore the next radiofrequency (RF) pulse. (Figure 3). This isn’t the type of image
quence with a TE of 20 milliseconds, After multiple RF pulses, there will be a contrast needed for most body MRI
only 2 lines of k-space might be filled “steady-state” residual-transverse mag- applications.
during 20 milliseconds. This would netization,3 which is constant (Figure Steady-state contrast is thus the
halve the imaging time, but is still not 2). This residual-transverse magneti- default condition for fast gradient-echo
nearly enough for breath-hold imaging.     zation contributes T2-weighting to the imaging. To remove this contrast, the TR
Therefore, gradient-echo imaging image. The sequence is now fast and could be increased so the transverse
must be used to acquire fast, robust, T1- robust, but it no longer has strict T1- magnetization will relax fully after each
weighted images. However, the require- weighting. RF pulse. However, the sequence
ment for speed now presents a problem.  More specifically, we have a steady- would no longer be fast enough for
Fast gradient-echo sequences need to state, or coherent, sequence in which breathhold imaging. Instead, the resid-
use a short TR. If the TR is shorter than the image contrast depends upon the ual-transverse magnetization is re-
the T2-relaxation times of the imaged T2- to T1-ratio. What does this actually moved through “spoiling,”4 which can
tissues, which will typically be case look like, and is this a positive or nega- be done with a gradient or RF pulse
for breathhold imaging, the transverse tive factor? Fat and fluid will have high (Figure 4). Removal of the residual
magnetization will not fully decay be- signal, as fat and fluid have comparable transverse magnetization does result in a

January–February 2012 19
©

www.appliedradiology.com APPLIED RADIOLOGY n

BODY MRI SEQUENCES

As the T1-relaxation time of intra-

vascular contrast is shorter than that
of a parenchymal organ, the 3-dimen-
sional spoiled gradient-echo sequence
would use a larger flip angle for MR
angiography and a smaller flip angle
for parenchymal organ imaging. The
flip angle must also be adjusted for the
TR (the higher the TR, the greater the
flip angle).
Magnetization-prepared gradient-
echo sequences are another way to ob-
tain fast T1-weighted sequences that
will not be addressed here as these se-
quences are not commonly used for
FIGURE 8. Steady-state gradient echo. Steady-state gradient-echo images (TR 26 msec, TE body MRI. In these sequences, the T1
1.6 msec, flip angle 54°) from the same study as in Figure 9 demonstrate aortic regurgitation. weighting is provided by a prepara-
However, the turbulent flow is not as well visualized as in Figure 9 as the “bright blood” is not tion pulse. Typically these sequences
related to flow, but to the T2/T1 ratio. However, the SNR and contrast-to-noise ratio is higher
and the endocardial borders are better delineated. The TR can be lower than the spoiled gra-
are used to obtain multiple rapid T1-
dient-echo sequence due to the higher-intrinsic SNR of this sequence. weighted images at the same level; eg,
for myocardial perfusion.  
In summary, the spoiled gradient-
echo sequence is the workhorse se-
quence for fast T1-weighted imaging.
Fast gradient-echo sequences, with-
out additional modification, are in the
steady-state and have T2/T1 weight-
ing; they must be spoiled to obtain
T1-weighting. Spoiled gradient-echo
sequences can be modified for a vari-
ety of applications by adjusting the flip
angle. 

Steady-state sequences 
Let’s return to the steady-state gra-
dient-echo sequence. This sequence is
FIGURE 9. Spoiled gradient-echo. Spoiled gradient-echo images (TR 69 msec, TE 4.2 msec,
both fast and robust. It has higher SNR
flip angle 25°) from a cine “bright blood” cardiac study demonstrate aortic regurgitation. The tur- than a spoiled gradient-echo sequence,
bulent flow is better visualized on this sequence than on the steady-state sequence from Fig- as the unspoiled residual-transverse
ure 8. The “bright blood” on a spoiled gradient-echo sequence is related to through-plane flow, magnetization contributes to the sig-
and thus dephasing secondary to turbulence is well visualized. However, SNR and contrast- nal. The intrinsically high SNR can be
to-noise ratio are less and the endocardial borders are not as well delineated as in Figure 8. 
parlayed into a faster imaging time by
fast, robust sequence, which is T1- acquisition time is needed (eg, MR an- increasing the bandwidth and decreas-
weighted. giography) these techniques are usually ing the TR. So what’s not to like about
Thus the majority of fast T1- combined with a k-space shortcut. a fast sequence with high SNR? We also
weighted sequences in body MRI ap- The 3-dimensional spoiled gradient- need the correct image contrast, and
plications are spoiled gradient-echo echo sequence is a versatile sequence T2/T1-image contrast is not the con-
images. These can be acquired 2-di- that can be used both for MR angiogra- trast needed for most body imaging ap-
mensionally or 3-dimensionally (in phy and imaging of the parenchymal or- plications. In the past, sequences were
MRI, 3-dimensional means that there is gans (Figure 6), simply by changing the always designed to avoid T2/T1 con-
phase-encoding in 2 axes). Often chem- flip angle. The Ernest angle5 is the flip trast. However, with increased gradient
ical shift (in- and out-of-phase) imag- angle in spoiled gradient-echo imaging strengths and shorter imaging times,
ing is used in conjunction (Figure 5). where signal intensity is maximized: there were applications in which strict
For applications in which a very short cos α = exp (-TR/T1). T1 or T2 weighting was not needed, as

22 January–February 2012
©

n APPLIED RADIOLOGY www.appliedradiology.com

 BODY MRI SEQUENCES

long as imaging could be performed There are substantial effects of this typically preferred for most of these ap-
quickly. A good example is cine car- change in sequences. For example, plications, but one advantage of steady-
diac imaging, in which what is required with steady-state gradient-echo se- state sequences is substantially lower
for image contrast is not strict T1 or T2 quences, the myocardium can be better radiofrequency deposition.
weighting but simply that the blood ap- delineated from the cardiac chamber In summary, steady-state gradient-
pears much brighter than the myocar- (Figure 8). This affects determination echo sequences have become the stan-
dium. So once MRI scanners were fast of ventricular volumes6 and thus ejec- dard for cine cardiac imaging, and
enough, steady-state sequences, which tion fraction (ventricular volumes de- have other applications as well. These
were previously explicitly avoided, termined by steady-state sequences are sequences are very fast and robust, but
turned out to be very useful.   larger than those determined by spoiled they are limited to applications where
  There are a few different steady- sequences). However, not everything T2/T1-image contrast is acceptable.   
state sequences.1 These are designed is advantageous. As steady-state se-
to maintain the steady state rather than quences are not flow-sensitive, it is
spoil it. Some of these sequences have harder to visualize turbulent flow (eg REFERENCES 
1. Boyle GE, Ahern M, Cooke J, et al. An interac-
minimal clinical use, others have spe- from regurgitation) (Figure 9). tive taxonomy of MRI sequences. Radiographics.
cialized applications (eg, DESS or Steady-state sequences can also be 2006;26:e24.
CISS). The relevant steady-state se- used to visualize blood outside of the 2. Lee VS. Fast scanning and k-space shortcuts.
In: Lee VS. Cardiovascular MRI: Physical prin-
quence for body MRI applications is a heart; that is, for nonenhanced-MR an- ciples to practical protocols. Philadelphia, PA: Lip-
fully refocused or balanced sequence, giography. This may be more relevant pincott Williams and Wilkins. 2006:138-158.
in which the steady state is maintained with the recently reported relation- 3. Chavan GB, Babyn PS, Jankharia BG, et al.
Steady-state MR imaging sequences: Physics,
with balanced gradients that minimize ship between gadolinium and nephro- classification, and clinical applications. Radio-
dephasing (Figure 7). genic systemic fibrosis. Steady-state graphics. 2008;28:1147-1160.
Steady-state sequences have essen- sequences are the most common se- 4. Mitchell DG, Cohen MS. Pulse sequences:
Gradient echo and spin echo. In: MRI Principles,
tially replaced spoiled gradient-echo quences used to obtain “scout” im- 2nd Edition. Philadelphia, PA: Elsevier Science;
sequences for cine cardiac imaging in ages in body MR imaging. As very 2003:400.
recent years. Steady-state sequences fast sequences with high SNR, they 5. Mitchell DG, Cohen MS. MRI principles. Phila-
delphia, PA: Saunders; 2004:163-176.
are faster, and have higher SNRs and are optimally suited for this purpose. 6. Mitchell DG, Cohen MS. Proton environments
contrast-to-noise ratios. With a spoiled As steady-state sequences are very fast and T1 relaxation. Philadelphia, PA: Saunders;
sequence, the “bright blood” comes sequences where fluid is bright, they 2004:177-185.
7. Mitchell DG, Cohen MS. MRI principles. Phila-
from through plane-flow enhancement can be viewed as a complement or al- delphia, PA: Saunders; 2004:21-34.
(similar to time-of-flight angiography). ternative to single-shot fast spin-echo 8. Hudsmith LE, Petersen SE, Tyler DJ, et al.
Steady-state gradient-echo sequences sequences for body MRI applications Determination of cardiac volumes and mass with
FLASH and SSFP cine sequences at 1.5 vs 3
are not flow-sensitive, and the “bright (eg, for MRCP or fetal imaging). Sin- tesla: A validation study. J Magn Reson Imaging.
blood” comes from intrinsic contrast. gle-shot fast spin-echo sequences are 2006;24:312-318.

January–February 2012 23
©

www.appliedradiology.com APPLIED RADIOLOGY n