Professional Documents
Culture Documents
Haase Snapshot FLASH Applications T1 T2 CS
Haase Snapshot FLASH Applications T1 T2 CS
A. HAASE
Snapshot FLASH magnetic resonance imaging techniques have been developed to en-
able real-time imaging of MR parameters. The first realization of the method is based on
a 64 X 128 FLASH tomogram acquired within 200 ms, using improved MR system
hardware conditions. The soft tissue contrast obtained in FLASH MRI almost disappears
by using flip angles of less than 5" and repetition times of 3 ms. This work describes
extensions of FLASH MRI placing conventional MR experiments before the whole im-
aging sequence. This creates images ofany desired contrast without changing the measur-
ing time. Examples of inversion-recoveryT1, spin-echoT2, chemical-shift-selective,and
spectroscopic FLASH MRI are presented. Further extensions to real-time MRI of blood
vessels, diffusion coefficients, combination with two-dimensional MR spectroscopy ex-
periments, and other nuclei are discussed. 0 1990 Academic Press, Inc.
INTRODUCTION
Several methods of fast MR imaging have been described ( 1-5). They can be classi-
fied as single-shot or multiple-shot MR imaging techniques. In single-shot MRI one
MR excitation of the plane of detection is followed by a series of MR signals, which
are created either by a frequent reversal of at least one magnetic field gradient ( 1 - 3 )
or by a series of spin echoes ( 4 ) .In a multiple-shot technique many MR excitations
of the plane of detection are needed, each followed by the acquisition of one MR
signal. We proposed the first multiple-shot technique, dubbed FLASH MRI (fast
low angle shot) ( S ) , in which low flip angle radiofrequency pulses were used. The
MR signal was created by reversing one magnetic field gradient. The measuring time
of single-shot imaging methods, e.g., echo planar (EPI), lies between 32 and 128 ms
(2, 3 ) , while conventional FLASH MRI is at least 30 times slower, with measuring
times of between 1 and 5 s. This large discrepancy is caused by the different hardware
conditions necessary for EPI techniques. For example, one precondition for EPI ex-
periments is a gradient coil system switchable within a fraction of I ms (6).
The purpose of the following experiments was to investigate FLASH MRI using
considerably improved MR system hardware. It can be demonstrated that under
these conditions, the repetition time can be reduced to a value of 3 ms, giving a total
77 0740-3 194190$3.00
Copyright 0 1990 by Academic Press, Inc.
All rights of reproduction in any form reserved.
78 A. HAASE
measuring time of 200 ms for a 64 X 128 pixel MR tomogram. This opens up a new
field of FLASH MR experiments giving any kind of parameter-selected snapshot
MR image.
THEORETICAL CONSIDERATIONS
EXPERIMENTAL DETAILS
FIG. 1. Radiofrequency (RF) pulse. and magnetic field gradient timing diagram for snapshot FLASH
MR imaging. The magnetic field gradient for the slice definition is Gs, the phase-encoding gradient Gp,
and the gradient for the production of the gradient echo G,. The gradient GSRserves for refocusing or
spoiling oftransverse magnetization, and can be omitted in many biomedical applications. The time period
for T, is 0.8 ms, the interval for phase-encoding Tpis 0.4 ms, the data acquisition period T, is 1.6 ms, and
the refocusing or spoiling period T,, is 0.3 ms. The flip angle of the RF pulse is less than 5". The repetition
time TR is 3 ms, and the read-out time of the gradient echo is 1.45 ms. The measuring time of snapshot
FLASH MRI is 192 ms using 64 phase-encoding steps.
cm-diameter inner gradient coil was used to supply three orthogonal gradients in the
x, y , and z direction of the magnet. A maximum gradient strength of up to 30 mT/
m was supplied by three gradient coils. Using a modified digital preemphasis unit,
magnetic field gradients could be switched within a time period of 0.1-0.2 ms. A 12-
bit ADC was used with a maximum sweep width of 80 kHz. In all 'H MR experiments
an 8-cm-diameter resonator tuned to 200 MHz was used.
Figure 1 shows the timing diagram of 'H FLASH MRI under these improved hard-
ware conditions. A Gaussian-shaped RF pulse of 0.5 ms duration was applied to
select a detection slice of 8 mm. The time period T, for the whole slice selection
procedure amounted 0.8 ms. Sixty-four phase-encoding steps within a time interval
Tpof 0.4 ms were used and 128 real and 128 imaginary data points were acquired in
the interval T, of 1.6 ms. At the end of data acquisition a time interval T,, of 0.3 ms
was introduced, either for spoiling transverse magnetization ( 10) or for refocusing
transverse magnetization by phase-encoding pulses ( 7). In many biological applica-
tions this part was not needed. The resulting cross-sectional snapshot FLASH MR
image was zero-filled to 128 X 128 pixel elements giving a spatial resolution of 1 X 1
mm. No further postprocessingwas needed.
The electronic devices and gradient system, described here, show that MR system
hardware conditions did not reach the high standard required for EPI. A completely
optimized MR system (gradient switching time of 0.2 ms, increased detection
bandwidth) would reduce the read-out time of the gradient echo from 1.45 ms in the
present stage to about 0.8 ms. The repetition time could be reduced from 3 ms to
80 A. HAASE
FIG.2. Snapshot FLASH 200 MHz ‘H MR image ofa phantom composed of six tubes of I-cm diameter
containing water doped with different concentrations of Gd(DTPA). The TI values are ( A ) 0.14s, ( B )
0.3s,(C)O.51s,(D)l.Os,(E)1.7s,and(F)2.4s.
about 1.4 ms, giving sampling times of less than 100 ms for the same spatial res-
olution.
'7
FLASH-MRI
<
Tir
I spoiling
FLASH-MRI
1
FLASH-MRI
2
....
IFLASH-MRI
Tir 1
FIG. 3. Timing diagram for two pulse programs for inversion-recovery (IR) snapshot FLASH MR im-
aging. The rectangle "FLASH MRI" represents the application of the complete snapshot FLASH experi-
ment using the timings as shown in Fig. 1. The spoiling period is applied to destroy transverse magnetiza-
tion caused by RF pulse imperfections. ( a ) Single-point IR FLASH MRI using one 180" inversion pulse
and an inversion-recoverywaiting delay T,r.The complete measurement of the inversion-recovery curve
needs the repetition of the experiment using different values of Ti,. (b) Single-shot IR FLASH MRI using
one 180" inversion pulse and the measurement ofthe inversion-recoverycurve with n FLASH MR images.
A single experiment is needed to measure the inversion-recoverybehavior represented by n FLASH images.
The total measuring time and number of images n are given by the largest T1 value.
FIG.4. Series of cross-sectional snapshot IR FLASH 200 MHz 'H MR images of the phantom, shown
in Fig. 2. The images are obtained using the single-shot IR FLASH technique, described in Fig. 3b. The
numbers indicate the order in which the images have been measured following a single 180".
images. This type of technique is the first snapshot T 1 imaging technique described
to date. It could be argued that FLASH experiments under these nonequilibrium
conditions resulted in a changed spatial resolution, due to a distorted point-spread
function. This effect could be observed in the case of short T1 values with respect to
the measuring time of 200 ms. However, for biological applications this effect causes
negligible problems and would probably be eliminated by using shorter repeti-
tion times.
FIG.5. Time-dependent IR-intensity profiles of selected regions within the phantom shown in Fig. 4.
The intensitiesof 16 images of Fig. 4 are displayed. The time resolution is 220 ms, including a 20-rns time
delay between the images. The curves are from (A) tube A with T1 = 0.14 s, (B) tube B with TI = 0.3 s,
(C)tubeCwith T1 = 0.51 s, ( D )tube D withT1 = 1.0 s, (E) tube E with T1 = 1.7 s, and(F) tube Fwith
TI = 2.4s.
both bottles. The T2 values were 800 and 200 ms. It has been measured that the T1
values were identical. This can be seen by the IR FLASH measurement in Fig. 8.
This experiment demonstrates that exact T2 measurements can be performed by
using a DEFT preceded high-speed FLASH method. There is no need to increase the
read-out time of the gradient echo to improve T2 contrast. The definite advantage
over other high-speed imaging techniques is the ability to observe extremely short T2
times. In alternative techniques, e.g., EPI, only parts of the object with long T2 values
will give the MR signal, while short T2 values will never be detected.
84
SNAPSHOT FLASH MRI 85
- r--- -
FLASH-MRI
spoiling
TE
FIG.7. Timing diagram for spin-echo snapshot FLASH MR imaging. The rectangle "FLASH MRI"
represents the application of the complete snapshot FLASH MR experiment shown in Fig. I. The spoiling
period is applied to destroy transverse magnetization caused by RF pulse imperfections. The RF pulse
sequence prior to the FLASH experiment uses a 90'-180"-90" technique to label the longitudinal magne-
tization with T2-dependent magnetization. The complete measurement of the T2 decay curve needs the
repetition of the whole experiment using different values of TE. It is also possible to apply a complete
CPMG sequence in front of the FLASH imaging procedure.
resonance line; the image ( C ) is the conventional high-speed FLASH image showing
both water and fat containing structures.
FIG. 8. Time-dependent spin-echo intensity profiles using the pulse sequence described in Fig. 7 of
selected regions of a phantom composed of two tubes containing water and agarose gel. (A) The curve
from the tube containing agarose gel shows a monoexponential T2 decay with T2 = 200 ms, and (B) the
curve from the tube containing water shows a monoexponential T2 decay with T2 = 800 ms. The curve
in (C) shows IR FLASH measurement of the T1 relaxation time of agarose gel and (D) of water. Both
tubes give identical TI values of 2.6 s.
86 A. HAASE
CHESS I------
9oo FLASH-MRI
spoiling
FIG.9. Timing diagram for chemical-shift-selective(CHESS) snapshot FLASH MR imaging. The rec-
tangle “FLASH MRI” represents the application ofthe complete snapshot FLASH MR experiment shown
in Fig. 1. The spoiling period is applied to destroy transverse magnetization caused by the CHESS pulse.
One single CHESS RF pulse excites unwanted spectral regions prior to the complete FLASH experiment.
T, will provide the desired MR spectra for each image element. The experiment re-
quires short 90” RF preparation pulses in order to avoid off-resonance effects. The
‘H MR spectrum at 4.7 T field strength covers a spectral region of *lo00 Hz. Here
the RF preparation pulse length should be less than 0.25 ms, which works even under
the conditions of whole-body experiments.
Figure 12 shows a spectroscopic FLASH experiment on a water-oil phantom. Im-
age (A) shows a two-dimensional representation of the acquired three-dimensional
data set with one spectroscopic axis and one spatial axis. The summed spectroscopic
images are shown in image (B), which represents the cross section of the phantom.
The data set has been acquired within 16 s, using 16 different time intervals T,, and
a repetition time for the whole pulse program of 1 s.
CONCLUSIONS
FIG.10. Three cross-sectional snapshot CHESS FLASH 200 MHz ‘H MR images of the human hand
after the application of a single CHESS pulse. ( A ) The CHESS pulse was centered on resonance of water,
giving a “fat” image, (B) the CHESS pulse was centered on resonance of CH2, giving a “water” image,
and (C) the CHESS pulse was centered 4 kHz off resonance to CH2resonance, giving the composed image
of water and fat regions.
SNAPSHOT FLASH MRI 87
- -
FLASH-MRI
I spoiling
FIG.1 1. Timing diagram for spectroscopicsnapshot FLASH MR imaging.The rectangle "FLASH MRI"
represents the application of the complete snapshot FLASH experiment shown in Fig. 1. The spoiling
period is applied to destroy transverse magnetization caused by RF pulse imperfections. The FW pulse
sequence prior to the FLASH experiment applies a 90"-90" technique to label the longitudinal magnetiza-
tion with chemical-shift-dependent magnetization. The complete measurement of the spectroscopicinfor-
mation needs the repetition of the whole experiment using different values of T, and a Fourier transform
with respect to T,.
tions. These applications would benefit, if the available hardware could be improved
to the high standard which is necessary to apply single-shot MRI, e.g., echo-planar
imaging.
It should be emphasized that EPI is the fastest technique to measure a cross section
of an object. FLASH sequences are a factor of 1.3 to 1.4 slower under the same hard-
ware conditions, mainly because of the frequent slice selections which are necessary
for the experiment. In almost all biomedical applicationsthis factor can be tolerated.
Furthermore, the signal strength of EPI can be at least a factor of 10 higher, since
snapshot FLASH needs excitation flip angles of less than 5". However, if high spatial
resolution is desired, a large number echoes are needed for EPI. This requirement
results in a signal loss of the echo-planar image due to T2 * decay. However, a com-
plete comparison of the sensitivity of snapshot FLASH and EPI is beyond the scope
of the present work.
Different applications of snapshot FLASH MRI have been presented. These have
been selected to show the principal characteristics of the whole new field of MRI
FIG. 12. Spectroscopic cross-sectional snapshot FLASH 200 MHz 'H MR images of a phantom com-
posed ofwater (below) and cooking oil (above). (A) The image shows a two-dimensional representation
of a three-dimensionalspectroscopic data set showing one spatial axis ( y ) and one spectroscopicaxis (F).
The resonance lines of fat (upper part) and water (lower part) are visible. The resonance lines are curved
due to magnetic field inhomogeneity within the phantom. (B) the image shows a combined cross section
through the same data set by summation of all spectroscopicimages.
88 A. HAASE
experiments which are now available. The principal idea of all methods is that the
imaging measuring time is no more the time-limiting factor of the MR experiment.
Furthermore, the imaging technique itself produces only minor modifications of the
longitudinal magnetization. For example, a frequent application of 64 3" pulses
would result in a 6%reduction of the total longitudinal magnetization (assuming that
no relaxation processes occur within 200 ms). The short measuring time shown in
these experiments is comparable to the acquisition of a single FID signal in a conven-
tional NMR technique.
The total investigation time using snapshot FLASH MRI is given by the desired
resolution of a parameter. The desired contrast can be selected by the MR experiment
in front of the complete FLASH sequence. Therefore, in contradiction to statements
made by other authors (2), any image contrast can be produced without changing
the measuring time of the FLASH MR experiment. In the case of T2 measurements,
a series of different TE values are needed for the DEFT sequence prior to FLASH
imaging. This gives access to high-speed images even from substances with short T2
values, below 30 ms. Other high-speed techniques, e.g., EPI, do not allow this experi-
ment. Two TE values are often used in clinical MRI applications to observe patholo-
gies. Using DEFT experiments prior to snapshot FLASH MRI this experiment would
need about 2 s.
Snapshot FLASH MRI obviously presents possibilities for combining imaging
techniques with any kind of other MR technique: ( 1) Quantitative T 1 imaging can be
performed either by combination with inversion-recovery experiments, as described
above, or by combination with saturation-recovery MR. ( 2 ) Quantitative T2 imaging
is possible by using combined FLASH and DEFT techniques. Moreover this tech-
nique is the only one described to date to measure short T2 values with a high-speed
technique. ( 3 ) Spectroscopic high-speed FLASH is applicable to 'H MR and other
nuclei, e.g., 31P,I9F,and 23Na,to allow fast determination of the regional distribution
of metabolites of biochemical interest. ( 4 ) Quantitative measurements of self-diffu-
sion coefficientswill be possible by the combination ofthe T2 FLASH imaging proce-
dure described in this work with magnetic field gradient pulses (13, 1 4 ) . ( 5 ) Real-
time measurements of flow velocities and angiograms will be possible with the DEFT
combined snapshot FLASH method using flow-sensitive magnetic field gradient
pulse techniques ( 1 5 ) . ( 6 ) The high-speed FLASH sequence can be combined with
many conventional 2D MR spectroscopy techniques (16). ( 7 ) Snapshot FLASH can
reduce the measuring time of a 3D imaging experiment to less than 15 s using the
hardware presented here.
Snapshot FLASH MRI offers a solution to many problems for biomedical MR. It
is applicable to small-bore MR systems in animal studies and offers a straightforward
solution to flow imaging artifacts ( 17). Moreover it could be applied to whole-body
MRI, when gradient systems and electronic data acquisition of conventional MR
systems can be modified, e.g., to perform techniques such as EPI. The economic and
scientific applications of MRI techniques should benefit from this approach.
ACKNOWLEDGMENTS
This work was supported by a Heisenberg grant of the Deutsche Forschungsgemeinschaft(DFG) and
funds of DFG ( Le 277/ 7-2 ) and of BundesministeriumFur Forschung und Technologie ( BMFT) ( 0 1-VF-
SNAPSHOT FLASH MRI 89
85 14). This work has been completed using the facilitiesprovided by the NMR group of the University of
Bremen (Professor D. Leibfritz). The author thanks Professor D. Leibfritz and Dr. D. G. Noms, Univer-
sity of Bremen, FKG, for critically reading this manuscript, and Dr. D. Matthaei, University ofGottingen,
FRG, for many helpful discussions during the preparation of this work.
REFERENCES
1 . P. MANSFIELD, J. Phys. C 10, L55 ( 1977).
2. R. J. ORDIDGE,R. COXON,A. HOWSEMAN, B. CHAPMAN, R. TURNER, M. STEHLING, AND P. MANS-
FIELD,Magn. Reson. Med. 8, 1 10 ( 1988 ).
3. I. L. PYKETT,AND R. R. RZEDZIAN, Mugn. Reson. Med. 5,563 ( 1987).
4. J. HENNIG,A. NAUERTH,AND H. FRIEDBURG, Mugn. Reson. Med. 3,823 ( 1986).
5. A. HAASE,J. FRAHM,D. MATTHAEI, W. HANICKE,AND K. D. MERBOLDT, “Book of Abstracts,” 4th
Annual Meeting of the SMRM, London, August 19-23, 1985, p. 980; J. Magn. Reson. 67, 258
(1986); US Patent No. 4,707,658.
6. P. MANSFIELD AND B. CHAPMAN, J. Phys. E 19,540 ( 1986).
7. M. L. GYNGELL,N. D. PALMER,AND L. M. EASTWOOD, “Book of Abstracts,” 5th Annual Meeting
ofthe SMRM, Montreal, 1986, p. 666.
8. J. S. WAUGH,J. Mol. Spectrosc. 35,298 ( 1970).
9. P. MANSFIELD AND P. G. MORRIS,“NMR Imaging in Biomedicine,” pp. 68-83, Academic Press,
New York, 1982.
10. J. FRAHM,W. HANICKE,AND K. D. MERBOLDT, J. Mugn. Reson. 72,307 ( 1987).
11. A. HAASE,J. FRAHM,W. HANICKE,AND D. MATTHAEI, Phys. Med. Biol. 30,341 (1985).
12. T. R. BROWN,B. M. KINCAID,AND K. UGURBIL,Proc. Natl. Acad. Sci. USA 79,3523 ( 1982).
I S . E. 0.STEJSKALAND J. E. TANNER,J. Chern.Phys. 42,288 ( 1965).
14. A. HAASE,in preparation.
15. P. R. MORAN,Mugn. Reson. Zrnagingl, 197 (1982).
16. A. HAASE,in preparation.
17. A. HAASE,D. MATTHAEI, D. HENRICH, submitted for publication.