| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Perspective |
1 Department of Radiology, Duke University Medical Center, Duke North, Rm. 1417,
Box 3808, Erwin Rd., Durham, NC 27710.
2 Siemens Medical Solutions, USA, Cary, NC 27519.
Received June 1, 2005; accepted after revision July 20, 2005.
CME
OBJECTIVE. The purpose of our article is to describe the underlying physics concepts of abdominal MRI at 3.0 T and their impact on signal-to-noise ratio, susceptibility artifacts, chemical shift artifacts, and dielectric effects.
CONCLUSION. Abdominal MR sequence protocols optimized for 1.5-T scanners should not be transferred to 3.0 T without substantial modification. In addition, specific patient groups—for example, large patients with ascites—are not well suited to undergo an abdominal MRI study at 3.0 T.
Keywords: abdominal imaging • field strength • MRI • MRI technique • physics
Over the past 2 years, ultrahigh-field-strength whole-body 3.0-T MR systems have been installed in numerous institutions and are being increasingly used in clinics. Besides market considerations—for example, strategic investment to make a market statement and to stay competitive—the main reason to purchase an ultrahigh-field MR system is the anticipated twofold MR signal-to-noise ratio (SNR) compared with a standard 1.5-T MR scanner. This gain in SNR can be kept or traded for either speed, spatial resolution, or both. Although the number of accessory receiving coils has been limited in the past, the spectrum of dedicated receiver coils offered by vendors has increased significantly over the past 18 months, which allows almost all standard MRI examinations to be performed on a 3.0-T whole-body MR system. Although ultrahigh-field MR systems have already been shown to be advantageous for various indications in the brain and musculoskeletal system compared with standard high-field 1.5-T MR systems, only a few scientific studies have been published describing the use of 3.0-T MR systems in the chest, abdomen, and pelvis [1-5]. Unfortunately, insights gained in musculoskeletal or neuroimaging research at 3.0 T cannot simply be transferred to body MRI because MR sequence protocols and object sizes differ significantly in abdominal imaging. In addition, some artifacts are unique to ultrahigh-field abdominal MRI and are not seen in other regions of the body. It is also not clear which patient groups will benefit from an ultrahigh-field abdominal MRI study and which patient groups should remain on a 1.5-T MR scanner. This article will illustrate the underlying physics concepts of abdominal MRI at 3.0 T and their impact on SNR, susceptibility artifacts, chemical shift artifacts, and dielectric effects. On the basis of these fundamental considerations, basic recommendations will be provided for which patient groups will likely benefit from an ultrahigh-field MRI study and which patient groups should undergo a standard 1.5-T abdominal MRI examination.
3.0 T Offers Twice the SNR: A Persistent Myth
The idea that twice the magnetic field will give twice the SNR is
appealing, and at first it seems correct because the intrinsic SNR in MRI is
approximately proportional to the main magnetic field strength, B0
(equations 1 and 2) [6]. The
equation for spin-echo-based MRI sequences is
 | (1) |
where SNRSE = signal-to-noise ratio for a spinecho pulse sequence, B0 = main magnetic field strength, V = voxel volume, NPE = number of acquired phase encode lines, NPA = number of acquired partitions, NAV = number of signals averaged, BW = receiver bandwidth per pixel, T1 = longitudinal relaxation time, and T2 = transverse relaxation time.
The equation for gradient-echo-based MRI sequences is
 | (2) |
where SNRGRE = signal-to-noise ratio for a spoiled gradient-echo
sequence and 
=flip angle.
Note that, in both equations 1 and 2, the term under the square root is simply the total time spent acquiring data. Therefore, SNR is proportional to the main magnetic field strength, the voxel volume, the square root of the total sampling time, and some sequence-specific contrast-related terms. Some of these factors, such as the longitudinal relaxation time (T1), receiver bandwidth, and specific absorption rate limitations, can affect the SNR in a somewhat complicated manner by impacting other sequence-specific parameters (e.g., TR, flip angle).
The longitudinal relaxation time, T1, increases at a higher magnetic field strength, which causes a decrease in SNR [1, 7] (see equations 1 and 2) (Fig. 1A). The transverse relaxation time, T2, on the other hand, seems to be fairly independent of the main magnetic field strength [7]. However, one recently published study by de Bazelaire et al. [1] suggests a marked decrease of the transverse relaxation time (T2) at higher magnetic field strengths, which would further reduce the gain in SNR at ultrahigh-field-strength MRI (Fig. 1B) for long TE protocols. Given the optimistic assumption that the transverse relaxation time (T2) is independent of the main magnetic field strength and assuming only an increase of the longitudinal relaxation time (T1), equations 1 and 2 can be used to determine the theoretic maximum relative gain in SNR during MRI of the liver. For turbo spin-echo-based T2-weighted sequences with sequential acquisition such as HASTE sequences, an increase by a factor of approximately 1.8 in SNR can be obtained. For gradient-echo-based T1-weighted sequences such as in- and opposed-phase and VIBE (volume interpolated breath-hold examination) sequences, an increase by a factor of approximately 1.6-1.7 in SNR can be obtained. Thus, the theoretic twofold increase in SNR at 3.0 T compared with 1.5 T will not generally be obtained without sequence parameter optimization.
|
|
A second major factor with a negative impact on the gain in SNR is related to the specific absorption rate (SAR). When the main magnetic field strength is doubled, the SAR, a measure for energy deposition within the human body, increases by a factor of 4. Although the energy deposited at 3.0 T is still nonionizing, the increased SAR requires an increased concern for patient safety. Because body MRI at 3.0 T almost always runs at the upper limits of the allowed SAR deposition, patients are more likely to experience an uncomfortable sensation of warmth or heating. In addition, protocol adjustments are frequently necessary, such as an increase of the TR, a decrease in the number of slices, or a decrease of the flip angle. These adjustments are all undesirable because they increase scanning time, reduce anatomic coverage, alter contrast, or further reduce the gain in SNR at 3.0 T when compared with a standard 1.5-T MRI system. Finally, much of the radiofrequency transmitter and receiver technology at 1.5 T is relatively mature compared with the newer technology at 3.0 T.
|
|
|
|
Chemical Shift Artifacts at 3.0 T: A Double-Edged Sword
The chemical shift artifact of the first kind is due to a difference in the
resonant frequency between water and fat and is seen only along the
frequency-encoding axis and the slice-selection dimension
[8]. This difference in
resonant frequency is directly proportional to the main magnetic field
strength and has been measured as approximately 3.5 ppm, resulting in a
difference of about 225 Hz at 1.5 T, or a difference of about 450 Hz at 3.0 T.
This difference causes a chemical shift mis-registration, which is most easily
seen around the kidneys (Figs.
2A,
2B,
2C and
2D). The chemical shift
artifact of the first kind appears as a hypointense band, 1 to several pixels
in width, toward the lower part of the readout gradient field, and as a
hyperintense band toward the higher part of the readout gradient field. At a
constant field of view, base resolution, and receiver bandwidth, the chemical
shift artifact of the first kind will be twice as wide at 3.0 T as at standard
1.5-T imaging. Usually, this enlarged artifact does not cause substantial
problems in clinical body MRI at 3.0 T. However, it may be problematic in
selected cases such as the search for a subcapsular renal hematoma or an
intramural aortic hematoma. In these cases, the receiver bandwidth can be
increased to minimize the chemical shift artifact of the first kind.
Unfortunately, this comes at the expense of SNR: doubling the receiver
bandwidth will decrease the SNR by approximately 30% (see equations 1 and 2
)
(Figs. 3A,
3B and
3C). Another option is to
repeat the MR pulse sequence with either chemical shift fat saturation,
inversion nulling, or water excitation, which will eliminate chemical shift
artifacts effectively, allow imaging at the lower bandwidth, and return the
30% loss in SNR.
|
|
|
Fortunately, the increased difference in resonant frequency between water and fat at 3.0 T may also be advantageous because it allows a better separation of the fat and water peak during MR spectroscopy, and a better or faster fat suppression using other chemical shift techniques as well—for example, fat saturation and water excitation.
Susceptibility Artifacts: A Closer Look
Magnetic susceptibility is the extent to which a material becomes magnetized when placed in a magnetic field. Susceptibility artifacts occur as the result of microscopic gradients or variations in the magnetic field strength that occur near the interfaces of materials of different magnetic susceptibility. These artifacts are usually caused by metallic objects from previous surgical or interventional procedures near or in the imaging field of view because the susceptibility of metal is much higher than that of soft tissue. Susceptibility artifacts increase with the main magnetic field strength and are approximately twice as large in terms of volume at 3.0 T as at standard 1.5-T MRI [9] (Fig. 4). This may be advantageous in selected cases because metal-related susceptibility artifacts from surgical clips or surgical debris—for example, prior cholecystectomy or prior hepatic resection—may be better seen (Figs. 5A and 5B). However, it is possible that enlarged susceptibility artifacts may obscure important findings at 3.0-T MRI that may have been visualized at standard 1.5-T MRI. It must be clearly stated here that metal-containing devices that are considered MR safe at a field strength of 1.5 T are not necessarily safe at 3.0 T [10-15]. All these devices need to be rigorously tested at 3.0 T as well before affected patients can undergo an MRI examination at this field strength.
|
|
|
|
|
|
|
|
|
|
|
|
In addition to the exacerbation of artifacts that are seen at 1.5 T, some new artifacts also begin to appear at 3.0 T. These artifacts are related to the higher frequency B1 transmit fields that are used at 3.0 T. The wavelength of the radiofrequency field at 128 MHz is 234 cm in free space, which is much larger than the field of view for clinical body imaging. However, water (and most body tissue) has a rather high dielectric constant, which reduces both the speed and wavelength of electromagnetic radiation. For visible light, this effect causes a straight stick entering water at an angle to appear bent. For MRI, this effect reduces the radiofrequency field wavelength from 234 cm in free space to about 30 cm in most human tissues—that is, water-containing tissues [16]. This size is approximately the size of the field of view for many body applications and can result in a so-called standing wave effect (often incorrectly called a "dielectric resonance" effect) [17]. As a result, strong signal variations across an image can be seen, especially brightening or dark "holes" in regions away from the receive coil caused by constructive or destructive interference from the standing waves. These artifacts become more pronounced the larger the region of interest is relative to the wavelength—that is, they are seen more in obese patients with a distended abdomen than in thin patients.
A rapidly changing magnetic field, like the radiofrequency transmit field, will induce a circulating electric field (Fig. 8). When this happens in a conductive medium, a circulating electric current is established. This current in turn acts like an electromagnet that opposes the changing magnetic field, reducing the amplitude and dissipating the energy of the radiofrequency field. The more conductive the medium, the stronger the opposing electromagnet and therefore the greater the attenuation of the radiofrequency field. In construction of the MRI suite, this principle is used by encasing the room in a copper conductor. Because of the high conductivity of copper, any incoming radiofrequency waves are almost completely attenuated and the magnet is shielded from external interference. To a lesser extent, large amounts of relatively highly conductive tissues can cause similar shielding effects, resulting in hypointense areas in the image where the radiofrequency field is partially attenuated [16].
These two effects combine to cause particularly strong artifacts for 3.0-T body MRI in pregnant patients and in patients with ascites (Figs. 9A, 9B, 9C and 10). In both cases, not only are the standing wave effects more pronounced because of the enlarged abdomen, but greater radiofrequency field attenuation is also present because of the increased amounts of highly conductive amniotic or ascitic fluid.
Summary and General Recommendations
Body MRI at 3.0 T is still in its infancy and will improve substantially over the next several years. However, radiologists need to know these several limitations based on the laws of physics:
First, overall, the gain in SNR at 3.0 T will be less than twofold compared with a standard 1.5-T MR system because of the inescapable increase of the longitudinal relaxation time T1. Also, the increased SAR deposition at ultrahigh field strength often requires protocol adjustments that can further reduce the anticipated gain. The gain in SNR will be higher in T2-weighted sequences than in T1-weighted sequences because longer TRs allow a more complete recovery of the longitudinal magnetization, and transverse relaxation times (T2) are fairly independent of the main magnetic field strength. Thus, patients referred for MR cholangiography may benefit from an ultrahigh-field-strength MR examination.
Second, chemical shift artifacts of the first kind are twice as large on ultrahigh-field MRI as on standard 1.5-T MRI. Chemical shift artifacts of the second kind, on the other hand, do not increase in size, although the timing is altered. Fortunately, the increased difference in resonant frequency between water and fat at 3.0 T is also advantageous because it allows a better separation of the fat and water peaks during MR spectroscopy and a better or faster fat suppression using chemical shift techniques.
Third, susceptibility artifacts are twice as big on 3.0-T MRI. Although patients referred for a colon study may be challenging, the search for gas—for example, free air or pneumobilia—should be easier on 3.0-T MRI. Patients with metal implants should undergo an MR examination at 3.0 T only if the metal-containing device has been proven to be MR safe for this field strength.
Fourth, standing wave and conductivity effects are usually not seen at a field strength of 1.5 T. At 3.0 T, these artifacts are most pronounced in pregnant women in the second and third trimesters because of the large amount of amniotic fluid and the increased size of the abdomen. Fetal MRI should therefore not be performed at 3.0 T because of these severe artifacts and the increased safety concerns. The same holds true for patients with a large amount of ascites, who are also not well suited for an ultrahigh-field MRI examination.
Finally, most patients can undergo an abdominal MRI study at 3.0 T with a reasonable outcome in terms of image quality.
Acknowledgments
We thank David E. Purdy and H. Cecil Charles for reviewing the manuscript and the subsequent stimulating discussions.
References
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Hori, T. Kim, T. Murakami, I. Imaoka, H. Onishi, K. Tomoda, T. Tsutsui, T. Enomoto, T. Kimura, and H. Nakamura Uterine Cervical Carcinoma: Preoperative Staging with 3.0-T MR Imaging--Comparison with 1.5-T MR Imaging Radiology, April 1, 2009; 251(1): 96 - 104. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. T. Schindera, B. J. Soher, D. M. Delong, B. M. Dale, and E. M. Merkle Effect of Echo Time Pair Selection on Quantitative Analysis for Adrenal Tumor Characterization with In-Phase and Opposed-Phase MR Imaging: Initial Experience Radiology, July 1, 2008; 248(1): 140 - 147. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. M. Akisik, K. Sandrasegaran, A. M. Aisen, C. Lin, and C. Lall Abdominal MR Imaging at 3.0 T RadioGraphics, September 1, 2007; 27(5): 1433 - 1444. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. M. Barth, M. P. Smith, I. Pedrosa, R. E. Lenkinski, and N. M. Rofsky Body MR Imaging at 3.0 T: Understanding the Opportunities and Challenges RadioGraphics, September 1, 2007; 27(5): 1445 - 1462. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. I. Sher Invited Commentary RadioGraphics, September 1, 2007; 27(5): 1462 - 1464. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
), together with
data based on six volunteers acquired by de Bazelaire et al.
[
). Note nonlinear
increase of T1 of liver tissue with main magnetic field strength. Also shown
are nonlinear regression (solid line) and 95% confidence interval for
data (dashed lines) as described by Bottomley et al.
