10-07 Gradient Echo Sequences

For many clinical indications, rapid imaging sequences are essential to avoid long imaging times, which can cause motion artifacts and reduce patient through­put. Imaging time can drop from several minutes per standard SE ima­ge to seconds or even milliseconds. The number of specific indications of par­ti­cu­lar rapid pulse sequences has steadily increased over the last few years. GRE se­que­nces were the favorite rapid imaging sequences; however, the popularity of sequences in the RSE family is constantly increasing. For faster imaging, the num­ber of averages, the number of matrix points or image lines, or the re­pe­ti­tion have to be shortened. In general, signal-to-noise ratio and spatial resolution will worsen with faster imaging methods, but as stated, a good clinical diagnosis does not necessarily require beautiful image quality, but sufficient image qua­li­ty.

The main contrast parameters of conventional and rapid sequences are sum­ma­ri­zed in Table 10-05. Actual weighting of the sequences depends on a number of factors and might not be available on all MR imaging equipment. The T2*-de­pen­dent PS sequence equals a GRE sequence.

Table 10-05: Pulse sequences and their contrast de­pen­dence. The signal of all pulse sequences is influenced by ρ and by bulk flow.

GRE sequences take advantage of the saturation of the spin system when TR is shortened. The signal intensity after a series of 90º pulses becomes weaker, until an equilibrium (i.e., the saturation) is reached. Under these conditions, pulse ang­les smaller than 90° are more effective. It was unexpected that shor­te­ning TR below 100 ms, even below 10 ms, still provided images with a signal-to- noise ra­tio which was sufficient and allowed diagnostic assessment.

Gradient-echo sequences of this kind, with such short TR, have been dubbed FLASH sequences [⇒ Haase]. They are commercially available under several dif­fe­rent trade names (see Table 10-06 and List of Abbreviations).

Signal intensity in rapid imaging sequences can be calculated with the fol­lo­wing equation, if TR is shorter than T1 but longer than T2* or when gradient/RF spoiling is applied to remove transverse coherences:

SI = sinα × [1-exp(-TR/T1)] × exp(-TE/T2)/ 1 - cosα × exp(-TR/T1)

where α is the flip angle, TR the repetition time, T1 the longitudinal and T2 the trans­ver­sal relaxation time.

10-07-01 The Flip Angle

FLASH sequences add a fourth parameter to TR, TE, and TI: the pulse or flip an­gle α, also called FA. Similar to SE sequences, GRE sequences can be weigh­ted depending on repetition and echo times, the exact pulse sequence and the pulse angle.

However, there is one big difference: whereas SE and RSE sequences reflect true T2 in their T2-weighted images, GRE sequences show only T2* contrast. Figures 10-12 and 10-13 depict the typical signal intensity behavior of a GRE sequence, in this case a spoiled FLASH sequence. Commonly, the signal in­ten­si­ties reach a maximum between 30° and 60°. As we have seen with the signal in­ten­si­ty and contrast behavior of SE sequences, best contrast is not necessarily ob­tai­ned at the point of highest signal intensity. This is also the case in GRE se­que­nces, as the contrast behavior of the brain images of Figure 10-14 shows. At the greatest signal intensity, there is poor or no contrast.

Figure 10-12:
Gradient echo sequence (spoiled GRE). TR = 400 ms; TE = 20 ms. B₀ = 1.5 T.
Because of the three variables available, there are nearly unlimited possibilities for changing ima­ge contrast. Generally, at low flip angles proton density dominates contrast, at high flip angles T1 becomes more important.

Images (through the brain of a normal volunteer): (a) α = 15°; (b) α = 30° ; (c) α = 45°; (d) α = 60°; (e) α = 75°

Figure 10-13: Animation: Eleven images; α between 8° and 88°. Note that these are simulated images: Both on the picture sequence of Figure 10-12 and on the animated sequence image noise has been removed. Simulation software: MR Image Expert®

It turns out that images acquired using the Ernst angle tend to have rather poor contrast. Higher flip angles have to be used to improve the contrast. The effect of this is a reduction of the signal left along the z-axis after the RF pulse. Thus, the signal level depends on the rate at which the signal recovered during TR; it is strongly T1-dependent. The image series and the animated sequence in Figures 10-12 and 10-13 give an overview of how contrast changes with in­crea­sing flip angle. GRE sequences can provide sharp contrast between the CSF com­part­ment of the spine, the spinal cord, and the peripheral spinal column. The mye­lo­gram effect of the T2*-weighted images allows a fast screening for disk protrusions and is one example of clinical applications of GRE.

However, SE and, in some in­stan­ces, RSE sequences commonly yield sharper spatial detail, and con­trast of GRE sequences ge­ne­ral­ly is in­fe­ri­or to that of SE se­quen­ces. Con­trast can be enhanced by contrast agents; by applying a high flip ang­le, the T1 effect of pa­ra­mag­ne­tic con­trast agents can be em­pha­si­zed. The creation of T2*-weighted con­trast is hampered by field in­ho­mo­ge­ne­ities, which are not refocused by the gradient echo. The in­ho­mo­ge­nei­ties solicit short echo times and limit the use of long echo times necessary for T2*-weighting. Re­duc­tion of TR shorter than T2 leads to the generation of transverse co­he­ren­ces which can either be spoi­led or refocused, as described in Chap­ter 8. Spoiled FLASH sequences (cf. Tab­le 10-06) remove the effect of the trans­ver­se coherences, usually by the application of spoiler gradients, to give genuine partial saturation contrast. Table 10-06: Some rapid imaging techniques correlated to the respective generic pulse sequence. * In this context, contrast-enhanced refers to the radiofrequency pulse sequence; is does not mean enhancement with a con­trast agent.

Refocusing GRE sequences incorporate the transverse coherences into the ob­ser­ved signal, and thus have a better signal-to-noise ratio. However, the basic refocused FLASH sequence generally has rather poor con­trast (which depends on T1/T2). The contrast-enhanced version, CE-FLASH, offers additional T2 con­trast, the amount of T2-weigh­ting being determined by TR and T2. The T2- weigh­ting is greatest at longer T2 values (e.g., 30-60 ms), but the signal-to- noise ratio is poorer than at short TR values.

GRE sequences are exquisitely sensitive to magnetic susceptibility (e.g., de­pic­ting hemorrhage and blood degradation products) and to flow phenomena (an­gio­graphy). Table 10-07 summarizes the features of a standard FLASH se­que­nce at high field (1.5 Tesla).

Table 10-07:
Approximate contrast characteristics in a standard GRE sequence at high field (1.5 Tesla).

As we have seen in the SE se­quen­ces, one can hide pathological chan­ges by choosing the wrong pulse se­quen­ce. This also holds for rapid se­quen­ces. If we select a T1-weighted se­que­nce, we cannot distinguish a lesion which possesses a similar T1 to its neighboring tissues. If we apply a T2*-weighted sequence, we cannot delineate a lesion with a T2 close to the T2 of its surroundings. The signal intensity of the vascular malformation in Figure 10-14 is a good example of this problem. Figure 10-14: Gradient echo pulse sequence (spoiled GRE) through the brain of a patient with a vascular malformation in the right occipital hemisphere. The left image series was ta­ken with an echo time TE = 20 ms, the right series with an echo time TE = 120 ms (B0 = 1.5 T). The lesion is nearly invisible in the image series with short TE, but well delineated in the series with long TE. The choice of the appropriate pulse se­quen­ce parameters is pivotal in MR ima­ging. Many different sequences can be applied for different diagnostic questions. In many instances, their contrast behavior has been recorded empirically and the se­quen­ce and specific sequence pa­ra­me­ters have been included in special clinical ima­ging protocols.

MP-RAGE and 3D-MP-RAGE. In snapshot gradient-echo scans, the signal evolves to different levels during the scan. Therefore, by manipulating the starting value one can alter the form of the evolution and thus the image contrast. The most commonly used pre­pa­ra­tion pulse is a 180° inversion pulse.

3D MP-RAGE (three-dimensional Magnetization-Prepared Rapid Gradient Echo) was introduced by Mugler and Brookeman in 1990 [⇒ Mugler]. The MP- RAGE se­quen­ce com­bines a 3D-inversion recovery pulse and N equally-spaced readout RF pulses of a specific flip angle with an echo spacing τ. The pulse cycle within the repetition time, TR, consists of the following components:

TR = TI + N×τ + TD

where τ is echo spacing time, N is the total number of readout RF pulses, TI is the time interval between the inversion recovery pulse and the first RF readout pulse, and TD is an adjustable delay time.

Image contrast is a function of N, TI, τ, the flip angle and the temporal position of the readout RF pulse, as well as the regular factors influencing contrast such as re­la­xa­tion times. Generally, the total number of RF pulses N is related to the spa­tial resolution along the slice direction. In commercial machines, the k-space strategy, including k-space trajectory and sampling order, is constrained to a few choi­ces. Often the theoretically achievable best signal cannot be reached.

In the MP-RAGE sequence, the effective inversion recovery time (TIeff) is a major determining factor of image contrast. It is defined as the time interval between the inversion recovery pulse and the RF read-out pulse for k-space center.

A good overview of the contrast behavior as well as basics of this pulse sequence is given by Wang [⇒ Wang 2014].

Figure 10-15:
3D-MP-RAGE. Simulated contrast between gray and white matter at 3.0 Tesla as function of TI for a total number of readout RF pulses of 176, 156, and 132, respectively. The interval time between readout RF pulses was set to 10 ms; the flip angle to 12°. To get decent T1-weighted contrast, the flip angle should be kept lower than 20° in this kind of pulse sequence. Contrast between tissues changes drastically with changes of TI as well as other parameters, e.g., T1 relaxation and field strength (modified from ⇒ Wang 2014).

Other Rapid Imaging Sequences. Chapter 8 describes a number of different fast imaging sequences, such as EPI. Contrast in EPI depends on the pre­pa­ra­tion module used before the EPI module. This can be an SE module, a GRE mo­du­le, or an IR module. The contrast of the EPI sequence will behave accordingly.