Gradient echo sequences

The most commonly used GE sequences for routine abdominal imaging are SGE and 3D-GE sequences.

SGE sequences (Figure 2.7) are two dimensional sequences and can also be used as a single (breathing independent) or multi acquisition (breath hold) technique. Image parameters for breath-hold SGE sequences involve (1) relatively long repetition time (TR) (approximately 150 ms) and (2) the shortest in-phase echo time (TE) (approximately 4.4 ms at 1.5 T, 2.2 ms at 3.0 T), both to maximize signal-to-noise ratio (SNR) and the number of sections that can be acquired in one multisection acquisition (7, 8). For routine T1-weighted images, in-phase TE are preferable to the shorter out-of-phase echo times (2.2 ms at 1.5 T, 1.1 ms at 3.0 T), to avoid both phase-cancellation artifact around the borders of organs and fat-water phase cancellation in tissues containing both fat and water protons. Flip angle should be approximately 70° to 90° to maximize T1-weighted information. Using the body coil, the SNR of 2D SGE sequences is usually suboptimal with section thickness less than 8 mm, whereas with the phased-array multicoil, section thickness of 5 mm results in diagnostically adequate images. On new MRI machines more than 22 sections may be acquired in a 20-s breath hold (7, 8).

In single-shot techniques, all of the k-space lines for one slice are acquired before acquiring the lines for another slice, that is, in sequential rather than interleaved fashion. SGE may be modified to a single-shot technique by minimizing TR, to achieve breathing-independent images for noncooperative patients (7, 8).

GE sequences can be performed as a 3D acquisition, which can be used both for volumetric imaging of organs such as the liver and for pre- and postgadolinium administration.

A 3D-GE sequence is typically performed with minimum TR and TE and a flip angle between 10° and 15°. The flip angle is related to the TR to maximize T1 signal. With a minimum TR, flip angle can also be relatively low and still achieve good T1 contrast. As TR and TE of these sequences have minimum values (TR < 5 ms and TE < 2.5 ms), the flip angle will mainly determine the contrast in the images. In 3D-GE, a volume of data is obtained rather than individual slices. There are several advantages of 3D versus 2D GE sequences: (i) higher inherent SNR than 2D GE sequences, (ii) higher in-plane resolution with larger matrices, (iii) contiguous, thinner sections for higher through plane resolution which facilitates reformats in other planes, (iv) homogenous and fast fat-suppression, which allows scan times that are still short enough to perform breath-hold imaging with fat suppression, (v) the enhancement of vessels and tissues with gadolinium is more obvious because of the fat suppressed nature of the sequence, (vi) gadolinium-enhanced 3D-GE sequences also are the most clinically effective techniques for MR angiography (MRA) of the body (7, 8).

3D-GE sequences, in combination with or without robust segmented fat-suppression techniques, overlapping reconstruction, and in-plane as well as through-plane interpolation of the MR data, allow high quality imaging with larger volume coverage in breath-hold times. 3D-GE sequences can be obtained with sufficient anatomic coverage, very thin sections, and high spatial resolution matrices in scan times of only 15–20 s.

Precontrast in-phase and out-of-phase 3D-GE sequences (Figure 2.7) can be acquired with high image quality, although the image quality is slightly below the image quality of 2D SGE. However, their higher spatial resolution is their main advantage. Additionally, the four-point Dixon method allows the acquisition of water-only and fat-only images (Figure 2.7).

3D-GE is particularly suitable for dynamic contrast-enhanced MR imaging because of its excellent inherent fat suppression and sensitivity to enhanced tissues and abnormalities. Currently, most of the 3D-GE sequences used for dynamic gadolinium-enhanced MR exams employ sequential k-space filling as opposed to the sequences with centric or elliptic-centric k-space filling used in MRA in most centers. These beneficial features, in combination with clinically viable acquisition breath hold times, make 3D-GE the primary technique over 2D SGE for dynamic contrast enhanced imaging of the abdomen and pelvis, including the liver (Figure 2.6) (7, 8).

Additionally, 3D-GE sequences with radial acquisition have recently been used in noncooperative patients who cannot hold their breaths. In noncooperative patients, pre and postgadolinium fat-suppressed T1-weighted imaging can be performed with 3D-GE sequences acquired with radial acquisition without breath holding.

Fat-suppressed gradient echo sequences

Fat-suppressed SGE or 3D-GE sequences are routinely used as precontrast images (Figure 2.8). They are important for the detection of subacute blood, high protein content, and fat containing lesions, and particularly for the evaluation of pancreas (Figure 2.8). Image parameters are similar to those for standard SGE or 3D-GE; however, it is advantageous to employ a lower out-of-phase echo time (2.2 to 2.5 ms at 1.5 T) for SGE, which benefits from additional fat-attenuating effects, and also increases SNR and the number of sections per acquisition. Fat-suppressed 3D-GE sequences are now routinely used for the acquisition of fat-suppressed precontrast T1-weighted images. Although the contrast resolution of fat-suppressed 3D-GE is mildly lower compared to fat-suppressed SGE, the acquisition of thin sections with higher matrices is advantageous (7, 8).

Fat-suppressed-SGE and 3D-GE images are often used to acquire postgadolinium images (Figure 2.6). Fat-suppression is particularly essential for interstitial-phase gadolinium-enhanced images. The complementary roles of gadolinium enhancement, which generally increases the signal intensity of disease tissue, and fat suppression, which diminishes the competing high signal intensity of background fat, are particularly effective at maximizing conspicuity of diseased tissue (7, 8).

Out-of-phase gradient echo sequences

Out-of-phase (opposed-phase) SGE images are useful for demonstrating diseased tissue in which fat and water protons are present within the same voxel. A TE = 2.2 ms is advisable at 1.5 T. A TE = 6.6 ms is also out-of-phase at 1.5 T but the shorter TE of 2.2 ms is preferable as more sections can be acquired per sequence acquisition, the signal is higher, the sequence is more T1-weighted, and in combination with a T2-weighted sequence, it is easier to distinguish fat and iron in the liver. At TE = 6.6 ms both fat and iron in the liver result in signal loss relative to the TE = 4.4 ms in-phase sequence; while on TE = 2.2 ms out-of-phase sequences fat is darker and iron is brighter relative to TE = 4.4 ms, facilitating their distinction (Figure 2.9). At 3.0 T, the shortest in-phase and out-of-phase values are 2.2 and 1.1 ms, respectively. It is not possible to acquire the first echoes at these optimal in-phase and out-of-phase TE times due to gradient capabilities, unless separate in-phase and out-of-phase acquisitions are used or highly asymmetric echoes are utilized in a single scan. Therefore, asymmetric TE values can be preferred for in-phase (2.5 ms) and out-of-phase (1.58 ms) echoes and these values vary depending on the vendor. When acquiring in-phase/out-of-phase GE sequences, it is essential to use an opposed-phase TE which is less than the in-phase TE. Additionally, it is also possible to acquire in-phase and out-of-phase images with high in-plane and through-plane spatial resolution with the help of 3D-GE sequences; however, the image quality of this technique is still developing (7, 8).

The most common indication for an out-of-phase sequence is the presence of fat in the organs or lesions such as fatty infiltration of the liver, or hepatic adenoma, or HCCs (Figures 2.3, 2.9, and 2.10). Another useful feature is that the generation of a phase-cancellation artifact around high signal intensity masses, located in water-based tissues, confirms that these lesions are fatty. An example of this is angiomyolipoma of the liver (Figure 2.10). In addition to out-of-phase effects, the different TE, for the out-of-phase sequence compared to the in-phase sequence, provides information on magnetic susceptibility effects that increase with increase in TE. This can be used to distinguish iron-containing structures (e.g., surgical clips, iron deposition in the liver, gamna gandy bodies in the spleen) from nonmagnetic signal void structures (e.g., calcium) (Figures 2.9 and 2.11). To illustrate this point, the signal void susceptibility artifact from surgical clips increases in size, from a shorter TE (e.g., 2.2 ms) out-of-phase sequence to a longer TE (e.g., 4.4 ms) in-phase sequence, whereas the signal void from calcium remains unchanged (7, 8). Therefore, 3D-GE or SGE with out-of-phase TE can be used for pre- and postcontrast imaging in patients with prominent susceptibility artifacts which arise from metallic implants, clips, prostheses, or air (free air or bowel gas) and impair the evaluation of adjacent structures, because short TE times decrease the size of those artifacts compared to long TE times as on in-phase images.

Magnetization-prepared rapid-acquisition gradient echo (MP-RAGE) sequences

One limitation of GE sequences, both SGE and 3D-GE, is relative motion sensitivity, and requirement for cooperation by the patient in following breathing instructions. Despite using relatively short breath-hold times, motion artifacts are still the most common and important obstacle for the acquisition of good image quality (Figure 2.12). Therefore, the development of extremely fast breath-hold and/or breathing-independent sequences, which would dramatically reduce the motion artifacts, is still in progress.

In uncooperative patients, GE sequences may be modified as a single-shot technique using the minimum TR to achieve breathing-independent images. Such sequences have included so-called magnetization prepared rapid acquisition gradient echo (MP-RAGE). MP-RAGE sequences include turbo fast low-angle shot (turboFLASH) sequence. In 2D multi-section varieties, these techniques are generally performed as a single shot, with image acquisition duration of 1 to 2 s, which renders them relatively breathing-independent. Magnetization preparation is currently performed with a 180° inversion pulse to impart greater T1-weighted information. The inversion pulse may be either slice or non-slice selective. Slice-selective means that only the tissue section that is being imaged experiences the inverting pulse, while non-slice-selective means that all the tissue in the bore of the magnet experiences the inverting pulse. The advantage of a non-slice-selective inversion pulse is that no time delay is required between acquisition of single sections in multiple single-section acquisition. A stack of single-section images can be acquired in a rapid fashion. This is important for dynamic gadolinium-enhanced studies. A non-slice-selective inversion pulse results in slightly better image quality, particularly because flowing blood is signal-void. Approximately 3 s of tissue relaxation is required between acquisition of individual sections, which limits the usefulness of this sequence for dynamic gadolinium-enhanced acquisitions. Current versions of MP-RAGE are limited because of low signal-to-noise, varying signal intensity and contrast between sections, unpredictable bounce-point boundary artifacts due to signal-nulling effects caused by the inverting 180° pulse, and unpredictable nulling of tissue enhanced with gadolinium (Figure 2.13). Research is ongoing to solve these problems with MP-RAGE so that it may assume a more important clinical role. Routine use of a high quality MP-RAGE sequence would further increase the reproducibility of MR image quality by obviating the need for breath-holding, particularly in patients unable to suspend respiration (Figure 2.13). 3.0 T MR imaging particularly improves the image quality of MPRAGE sequences due to higher signal to noise ratio and greater spectral separation (Figure 2.13) (7, 9, 10). Additionally, the use of spatial-spectral selective water excitation (WE), which is a special fat suppression technique compatible with MPRAGE sequence, allows the use of fat suppression with MPRAGE, particularly on postgadolinium sequences (Figure 2.13). Spatial-spectral selective WE is a newer fat-attenuation technique which selectively excites water protons, leaving protons in fat unperturbed, rather than exciting all protons and then spoiling the signal from fat protons (10).

Echo-train spin-echo sequences

Echo-train spin-echo (ETSE) sequences are termed fast spin-echo, turbo spin-echo, or rapid acquisition with relaxation enhancement (RARE) sequences. The principle of ETSE sequences is to summate multiple echoes within the same repetition time interval to decrease examination time, increase spatial resolution, or both.

ETSE sequences allows the acquisition of higher matrices with shorter acquisition times and resultant decreased motion artifacts compared to conventional T2-weighted spine echo sequences. Due to this advantage, ETSE sequences are generally used for evaluation of the pancreaticobiliary ductal system.

ETSE sequences, acquired as contiguous thin 2D sections, 3D thin section, or as a thick 3D volume slab, form the basis for MR cholangiography (Figure 2.14). The greatest concern with high resolution 3D ETSE sequences is the significant increase in specific absorption rate (SAR). This is particularly relevant as most 3D implementations must increase the echo-train length compared to 2D varieties because of the added data that needs to be acquired. Recently, newer MR systems, in combination with advances in gradient and software design, have allowed the development of efficient long echo-train 3D ETSE using variable flip angle refocusing pulses. Concerns of SAR are offset by using smaller (constant or variable) flip angles during data acquisition. With this strategy, T2 information can be maintained throughout the echo-train, or used more effectively, for high SNR imaging. Efficiency is further aided by implementing an additional “restore” pulse at the end of data collection to expedite magnetization recovery, allowing shorter TRs (∼1200 ms) to be used for T2 imaging. These recent modifications to conventional 3D-ETSE imaging have made high resolution isotropic T2 imaging of the pelvis and biliary system feasible in a relatively short acquisition time (∼5 min) (7, 11).

Single-shot echo-train spin-echo sequence

Single-shot ETSE sequence (e.g.: half-fourier acquisition single shot turbo spin-echo (HASTE), single shot fast or turbo spin echo (SSFSE or SSTSE) or single shot echo-train spin echo (SS-ETSE) is a breathing independent T2-weighted sequence that has had a substantial impact on abdominal imaging (Figures 2.4 and 2.5). Typical imaging involves a 400 ms image acquisition time, in which k-space is completely filled using half-Fourier reconstruction. Shorter effective echo time (e.g., 60 ms) is recommended for bowel-peritoneal disease, and longer effective echo time (e.g., 100 ms and greater) is recommended for liver-biliary disease. Since the technique is single-shot, the echo train length is typically greater than 100 echoes, while the effective TR for one slice is infinite. A stack of sections should be acquired in single-section mode in one breath-hold to avoid slice misregistration; however, the method lends itself to free-breathing application under uncooperative imaging circumstances. Recently, 3D versions of this technique have been implemented. Motion artifacts from respiration and bowel peristalsis are obviated; chemical-shift artifact is negligible; and susceptibility artifact from air in bowel, lungs, and other locations is minimized, such that the bowel wall is clearly demonstrated (Figure 2.15). Similarly, susceptibility artifact from metallic devices such as surgical clips or hip prostheses is minimal (Figure 2.15). All of these effects render SS-ETSE an attractive sequence for evaluating abdomino-pelvic disease. In patients with implanted metallic devices and extensive surgical clips, SS-ETSE is the sequence least affected by metal susceptibility artifact (7, 11).

However, one disadvantage of SS-ETSE sequences is that T2 differences between tissues are reduced in part due to averaging of T2 echoes, which has the greatest effect of diminishing contrast between background organs and focal lesions with minimal T2 differences. This generally is not problematic in the pelvis due to substantial differences in the T2 values between diseased and normal tissue. In the liver, however, the T2 difference between diseased (solid tissue) and background liver may be small and the T2-averaging effects of summated multiple echoes blur this T2 difference. These effects are most commonly observed with HCC. Fortunately, diseases with T2 values similar to those of the liver generally have longer T1 values than the liver, so the lesions poorly visualized on ETSE are generally apparent on precontrast GE as low signal lesions or on postgadolinium GE images as low or high signal lesions, depending on the enhancement features (7, 11).

Fat-suppressed (FS) echo-train spin-echo sequences

Fat suppression in spin echo sequences is useful for investigating focal liver disease to attenuate the high signal intensity of fatty infiltration, if present. Fatty liver is high in signal intensity on ETSE, in particular single shot versions, which lessens the conspicuity of high signal intensity liver lesions. Diminishing fat signal intensity with fat suppression accentuates the high signal intensity of focal liver lesions (Figure 2.16). FS SS-ETSE is also useful for evaluating the biliary tree. Fat suppression appears to diminish the image quality of the bowel due to susceptibility artifact from air–bowel wall interface and is not recommended for bowel studies, although bowel-related extraluminal disease (such as appendicial abscess) may be well shown with this technique.

Fat suppression in ETSE sequences is achieved through either spectrally-selective or non-spectral selective preparation pulses. Non-spectral selective fat suppression, which is termed short tau inversion recovery (STIR), is performed with a slice-selective inversion pulse timed to suppress the T1 of fat (TI = 150 ms at 1.5 T) (Figure 2.16). As the pulse is broadband, both water and fat are magnetization-prepared, resulting in suppression of fat signal, alteration of contrast from water signal, and depression of water signal from equilibrium. While fat signal is uniformly suppressed, the remaining soft tissue (water signal) has lower signal as a direct consequence of the inversion pulse. As soft tissue in the abdomen has long T1 relative to fat, adequate SNR is still maintained. Although STIR is sensitive to non-uniform inversion (due to B1 field effects), the method is well-accepted as a robust technique for fat suppression in the abdomen, using spin echo-based sequences. As the sequence is fundamentally different from SS-ETSE, it may be useful to combine both short duration sequences for the liver in place of longer, breathing-averaged ETSE (7, 12).

An alternate method of fat suppression involves spectral selection of fat resonances exclusively. Preparation pulses (inversion or saturation) can be spectrally tuned to fat signal, while avoiding undue suppression or alteration of water signal, as seen with STIR techniques. Conceptually, the technique is similar to STIR, in which magnetization (in this case, only fat) is prepared with an inversion pulse timed to suppress fat. Ideally, other soft tissue will remain unaffected, preserving high contrast-to-noise ratio. Since spectral fat suppression is frequency-specific, it is sensitive to spatial susceptibility, which creates an inhomogeneous magnetic field (Figure 2.17). The water and fat resonances are not clearly delineated in this circumstance, causing fat–water frequency “overlap”, and potentially a significant number of fat spins to be unsuppressed. This effect is also prevalent at the edges of the FOV, or any regions away from the magnet isocenter (Figure 2.17) (7, 12).

Further improvement in uniform fat suppression can be achieved by incorporating adiabatic pulses, which are a specially designed class of RF pulses that provide B1 insensitive spin nutation. Recently, spectral (fat)-selective adiabatic inversion pulses, termed SPAIR, have been used in T2-weighted imaging of the abdomen (Figure 2.18). The uniformity of fat suppression is more robust than traditional spectral-selective techniques, making it the method of choice for liver and bowel imaging. Moreover, its inherent high CNR provides important diagnostic advantages over STIR, since soft tissue signal is preserved. The effects of inhomogeneous main magnetic field still pose challenges to SPAIR, due to spectral overlap. But the improved inversion profile and frequency cutoff of SPAIR alleviates the degree of poor fat suppression. Implementation of SPAIR fat suppression takes longer than its other counterparts, due to the longer pulse length required for the adiabatic condition. This may increase scan times for segmented T2 and T1 acquisitions (7, 12).

Alternative fat-suppressed T2-weighted sequences for the liver, such as echo-planar imaging, have also been employed at some centers.

Focal or diffuse lesions may show mild, moderate, or markedly high signal, depending on fluid content on non-fat-suppressed and fat-suppressed images; high or low signal, depending on fat content on non-fat-suppressed images or fat-suppressed images, respectively; and low signal, depending on iron, fibrosis, or protein content on non-fat-suppressed and fat-suppressed images (Figures 2.15, 2.16, and 2.19).