TwinTree Insert

08-03 Gradient Echo Sequences


completely different approach to rapid imaging was used by the first pulse se­­quen­ces, which shortened imaging time in routine clinical settings. The ge­ne­ric name of these sequences is gradient-echo (GE) sequences or, better, gradient-recalled echo (GRE) sequences, and they come in a plethora of different acro­nyms. The basics of GRE were explained in Chapter 6.

The first sequence in this group was pre­sented in 1986 by Axel Haase and col­la­bo­ra­­tors and dub­bed FLASH [⇒ Haase 1986]. The FLASH (Fast Low Angle Shot) se­quen­ce is a saturation recovery sequence with a short repetition time (TR ≤200 ms), a low flip angle (≤90°), and a gradient echo for refocusing.

The application of flip angles different from 90° and 180° brought an end to the ideo­logy of long wait­ing times, which was based upon the belief that T1 is the li­mit­ing time factor of MR imaging. The reason for using a low flip angle is il­lu­stra­ted in Fi­gu­re 08-05.


Figure 08-05:
Principles of (a) a standard pulse sequence, compared to (b) a rapid imaging sequence of the FLASH type. In both cases, the net magnetization during equilib­rium is aligned with the z-axis. In the stan­dard se­quence, a 90° pulse tilts the mag­ne­ti­za­tion into the x'-y' plane. No lon­gi­tu­di­nal component re­mains. In the FLASH-type sequence, a flip angle α ≤90° is applied. Such a pulse divides the mag­­ne­­ti­­za­­tion in transverse and lon­gi­tu­di­nal components.
(c) In our example α equals 30°. This re­sults in a re­duction of the longitudinal mag­ne­ti­za­tion to 87%, where­as the transverse magnetization is 50% of the available lon­gi­tu­di­nal magnetization.
The flip angle which will give maximum signal is known as the Ernst angle.


When a 90° flip angle is applied, we convert all of the longitudinal mag­ne­ti­za­tion (in the z-axis) into transverse magnetization (signal in the x'-y' plane), while, e.g., for a 30° flip angle the amount of transverse magnetization is halved (sin 30°), but we still have 87% of the z-mag­ne­ti­za­tion (cos 30°). The z-mag­ne­ti­za­tion will recover at a rate determined by T1 during the interpulse interval. How­ever, since the TR is short in FLASH sequences, the z-magnetization left by the previous pulse becomes do­mi­nant and significantly increases the signal ob­tai­ned after the next RF pulse.

For a given repetition time, the flip angle which will give maximum signal can be cal­cu­lat­ed. It is known as the Ernst angle [⇒ Ernst 1966]:


Ernst angle = cos-1 [exp (-TR / T1)]

where TR is the repetition time and T1 the longitudinal relaxation time.


Figure 08-06 summarizes the main differences between a spin-echo and a gradient-echo (FLASH) pulse sequence.


Figure 08-06:
Principles of (a) a standard pulse sequence, compared to (b) a rapid imaging sequence of the FLASH type. (a) In the spin-echo pulse sequence, the echo is created by a 180° pulse. This involves relatively long time delays and high power deposition in the examined sample. Because of the dependence of TR on T1, TR has to be relatively long. (b) In the FLASH sequence, any pulse angle can be used instead of the initial 90° pulse. The echo is formed by gradient switching. This can be done faster and with less power deposition (potentially less hazardous for the patient). Thus, TR (and TE) can be shortened.
SE = spin echo; GRE = gradient (recalled) echo.


spaceholder redAs with all gradient-echo sequences, but unlike spin-echo sequences, the ef­fects of magnetic field inhomogeneities are not compensated so that short TE must be used if high-quality images are to be obtained. This rules out the pos­si­bi­li­ty of in­creas­ing the echo time to give T2 contrast.

Field inhomo­geneity effects can also be reduced by using small voxel sizes since this limits the de­pha­sing which occurs within a voxel.

To reduce the echo times, it is necessary to switch the gradients relatively quick­ly and to keep them stable after being switched. Gradient-switching re­qui­res less en­er­gy to create an echo than a 180° pulse. Thus, power de­po­si­tion in the body of a pa­­tient is re­duced, which is a major advantage of these sequences.


spaceholder redHow­ever, there are also a large number of disadvantages which have not yet re­sul­ted in FLASH replacing standard SE sequences in all instances. Because of the shor­ter TR, FLASH se­quences reduce not only the scan time but also the number of sli­ces that can be ac­quired.

Optimum repetition time has to be adjusted to the number of slices required and to other factors such as the duration of a breath hold for abdominal imaging or the heart rate in cardiac ima­ging.

When de­creasing the scan time, motion arti­facts tend to be reduced, while flow arti­facts will in­crease since the difference in signal in­ten­­si­ty between blood and sta­ti­o­nary tissue be­comes more marked at short repetition times.

The feature can be exploited in FLASH-based cine-MR imaging where 8-32 lines of the same slice are acquired during one car­diac cycle, then the sequence is repeated for each phase-encoding step to produce 8-32 images, each of which represents a dif­fe­rent stage of the cardiac cycle. The images can be presented — for instance — in the form of a closed movie loop, which depicts the function and dy­namics of the heart.


08-03-01 Transverse Coherences


When the repetition time for a FLASH se­quence is reduced to a level where it is shor­ter than T2, the relax­ation behavior is also influenced. This is due to the pre­sence of transverse coher­ences [⇒ Freeman 1971]. Their ex­ploi­ta­tion or suppression forms the basis of several fast imaging schemes based upon FLASH.


spaceholder redTo understand why transverse coherence occurs, we have to modify the sim­ple idea of a spin echo. After a 90° pulse the spins start dephasing. When a 180° pulse is ap­plied at a time τ after the 90° pulse, the ro­tation induced by the spin echo causes the magnetization to start refocusing and a spin echo forms at a time τ (= TE) after the 180° pulse. This model is very useful since it gives a clear picture for the formation of a spin echo. However, it is not so easy to visualize the effect of pulses that are smaller than 180°.

These ≤180°-pulses also form spin echoes. When the flip angle is not equal to 180°, the amplitude of the echoes is re­duced compared to that produced by a 180° re­fo­cus­ing pulse. In addition to the evolu­tion of the z-magnetization, there is now also evolution of the transverse magnetiza­tion. The signal received con­sists of con­tri­­bu­tions representing fresh transverse mag­netization and an echo term, which is the sum of all the possible echoes arising from combinations of the spin echoes creat­ed by pulses ≤180°.

By manipulating these parameters, three types of rapid FLASH imaging se­quen­ces can be defined.


Refocused FLASH (also known as FFE, FISP, FAST, GRASS, ROAST). These se­quences mea­sure the signal after the RF pulse, which corresponds to the combinati­on of the fresh transverse magnetization and the echo term [⇒ Frahm 1987, ⇒ Sekihara 1987].

They have a good signal-to-noise ratio, but generally rather poor con­trast. A very strong signal is obtained from flowing blood since the spins flowing into the slice will have equilibrium magnetization rather than the steady state magnetization of the stationary tissue (typically 10% of M₀).


Contrast-Enhanced (CE-) FLASH (or CE-FFE, PSIF, SSFP). The­se se­quences measure only the echo term [⇒ Hawkes 1987]. To avoid contamination from the fresh mag­ne­ti­za­tion present after the RF pulse, the echo term is ob­ser­ved prior to the RF pulse in the form of a gradient echo. CE-FLASH sequences provide good T2 contrast, but re­la­ti­ve­ly poor signal-to-noise. Shortening the TR improves the signal-to-noise, but also reduces the contrast. Flow artifacts are ge­ne­ral­ly absent from CE-FLASH since the blood flows out of the slice du­ring the TR interval and thus cannot be refocused to give an echo. These sequences are favorites for cardiac imaging at 1.5 Tesla and below; however, at ultrahigh fields (3 Tesla and more) they suffer from de­struc­ti­ve arti­facts and flip angle restrictions due to SAR problems.


Spoiled FLASH. This type of rapid pulse sequence observes only the fresh trans­­verse mag­ne­ti­za­tion. The echo term is re­moved (spoiled) by the use of either spoil­ing gra­dients or phase spoil­ing tech­­ni­ques. When a high flip angle is used, the spoiled FLASH sequence can give good T1 contrast.


Two other variants of FLASH sequences are the FADE sequence [⇒ Redpath 1988] and the FISP sequence [⇒ Oppelt 1986].

The FADE sequence combines the refo­cused and CE-FLASH se­quen­ces into a sin­gle se­quen­ce in which the two resulting sig­nals are observed in separate ac­qui­si­tion peri­ods during a single interpulse interval. Therefore, the minimum TR is longer, but the sequence is more efficient because we obtain two images with different con­trast.

The FISP sequence is designed to super­impose the two signals that are se­pa­ra­te­ly ac­quir­ed in FADE to give a single signal with excellent signal-to-noise ratio. Un­for­­tu­na­te­ly, the sequence is not practical since, unless the two images are perfectly align­ed, artifacts will result [⇒ van Vaals 1993].


08-03-02 Ultrafast Gradient-Echo Sequences


The use of very rapid FLASH sequences (with TR in the range of 4-10 ms) allows to pro­duce images in seconds or even in less than a second. These sequences are, for in­stan­ce, used for abdominal imaging, commonly as a single-slice technique. They allow breath-holding and thus can eliminate ghost artifacts and blur­ring from re­spi­ra­to­ry motion.

In the basic Snapshot FLASH sequence, no spoiler or refocusing gradients are in­clud­ed and a very low flip angle corresponding to the Ernst angle for such short TR va­lues is used. Now, little or no transverse coherence is generated and the re­sul­ting images are essentially proton-density-weighted. To improve the contrast in these exa­mi­na­tions, a pre­pa­ra­tion pulse can be used. Its function is to pre­pare the z-mag­ne­ti­za­tion prior to starting the examination [⇒ Haase 1990]. The scan times for 128×128 matrices vary between 0.5 and 1.0 seconds on clinical systems.

The main applications for Snapshot FLASH sequences are in abdominal ima­ging, car­diac studies and functional (dynamic) imaging using contrast agents. In the first two cases, other techniques suffer from motion artifacts or long scan times when trig­ger­ing is used. For dynamic imaging, the time resolution re­qui­red (1-3 seconds) means that snapshot sequences have to be used if a rea­son­able (128×128) resolution is to be obtained.

spaceholder redThe advent of gradient-echo techniques with a preparation pulse, such as Turbo-FLASH, snapshot FLASH, and MP-RAGE allowed a shortening of examinations times main­tain­ing the level of the signal-to-noise ratio — even for 3D imaging.

The Ultrafast 3D Gradient Echo (also knows as, e.g., 3D MP-RAGE, 3D TFE, 3D-FGRE) is the 3D version of Turbo-FLASH and has developed into one of the most fa­vor­ed sequences for T1-weighted brain imaging. Details of this sequence are dis­cus­sed in Chapter 10.

spaceholder redAn overview of many acronyms and abbreviations can be found in the List of Ab­bre­vi­a­tions.

spaceholder redIt is worth noting that the acronym FISP is used to refer to two different se­­quen­­ces (i.e., this sequence and refocused FLASH). A modification of refocused FLASH with refocusing of all three gradients is known as True FISP (and also Balanced FFE). This sequence is used in cardiac ima­ging.

spaceholder red

When acronyms cause confu­sion about termi­nology:
Acronyms and abbre­viations in magnetic reso­nance imaging.

inkpot Alphabet soup (with comments from Hamlet)

spaceholder red