DOI: 10.1148/rg.275065204
RadioGraphics 2007;27:1445-1462
© RSNA, 2007
Body MR Imaging at 3.0 T: Understanding the Opportunities and Challenges1
Mara M. Barth, MD,
Martin P. Smith, MD,
Ivan Pedrosa, MD,
Robert E. Lenkinski, PhD, and
Neil M. Rofsky, MD
1 From the Department of Radiology, Beth Israel Deaconess Medical Center, 330 Brookline Ave, Boston, MA 02215. Presented as an education exhibit at the 2005 RSNA Annual Meeting. Received December 18, 2006; revision requested January 31, 2007; revision received March 14 and accepted March 23. N.M.R. has received research support from GE Healthcare and has served on the advisory board for Schering (Berlex) and as a consultant with CAD Sciences and EPIX Pharmaceuticals; all remaining authors have no financial relationships to disclose.
Address correspondence to M.M.B. (e-mail: mbarth{at}bidmc.harvard.edu).
 |
Abstract
|
|---|
The development of high-field-strength magnetic resonance (MR) imaging systems has been driven in part by expected improvements in signal-to-noise ratio, contrast-to-noise ratio, spatial-temporal resolution trade-off, and spectral resolution. However, the transition from 1.5- to 3.0-T MR imaging is not straightforward. Compared with body imaging at lower field strength, body imaging at 3.0 T results in altered relaxation times, augmented and new artifacts, changes in chemical shift effects, and a dramatic increase in power deposition, all of which must be accounted for when developing imaging protocols. Inhomogeneities in the static magnetic field and the radiofrequency field at 3.0 T necessitate alterations in the design of coils and other hardware and new approaches to pulse sequence design. Techniques to reduce total body heating are demanded by the physics governing the specific absorption rate. Furthermore, the siting and maintenance of 3.0-T MR imaging systems are complicated by additional safety hazards unique to high-field-strength magnets. These aspects of 3.0-T body imaging represent current challenges and opportunities for radiology practice.
© RSNA, 2007
|
Introduction
|
|---|
The magnetic resonance (MR) signal generally is derived from a small number of excess unpaired hydrogen atoms aligned in the direction of a magnetic field. The number of aligned protons, and therefore the intensity of the MR signal generated, is directly proportional to the strength of that field. The desire to increase the signal is the basis for the continuing drive to create higher-field-strength imaging systems.
Initial clinical MR imaging systems had a field strength of less than 0.6 T. In 1982, 1.5-T imaging systems were introduced, and 1.5 T soon became the reference standard for high-quality MR imaging. The first 3.0-T systems became available in 1999, but for practical reasons, including inadequacies in radiofrequency (RF) coil design and protocols, their use remained limited to research and brain imaging for several years. Even these limited applications demonstrated improvements in the signal-to-noise ratio (SNR), spatial and temporal resolution, the contrast-to-noise ratio, and spectral resolution, compared with the same parameters at 1.5 T. More recent research and development efforts have been focused on expanding the clinical applications of 3.0-T MR imaging in other regions of the body.
The transition, however, has not been easy. Although lessons learned from previous field strength increases have been helpful in some aspects of 3.0-T imaging, the higher magnetic field strength has introduced new and unexpected challenges. Along with the gain in SNR, there is an increase in magnetic field inhomogeneity. The higher resonance frequency at 3.0 T results in increased interference in RF transmission and reception, which may produce spurious signal intensity variations across the image. In addition, because the energy deposition is proportional to the square of the static magnetic field, pulse sequences at 3.0 T are much more likely to be limited by the Food and Drug Administration (FDA) guidelines on power deposition or specific absorption rate (SAR). However, these challenges are surmountable with new and improved coil and pulse sequence designs and the judicious selection of imaging parameters.
Other technical obstacles include altered tissue relaxation times at higher field strengths. The longer T1 times of tissues at 3.0 T may necessitate an increase in the repetition time (TR) and, thus, in acquisition time. This trade-off directly opposes one of the advantages of 3.0-T imaging, namely increased imaging speeds. In addition, because of the higher resonance frequency at 3.0 T, chemical shift artifacts are more pronounced, and the decrease in T2* exacerbates susceptibility effects. Finally, implanted devices that are MR safe at 1.5 T are not necessarily safe at higher field strengths.
Despite these challenges, the benefits of clinical body imaging at 3.0 T are already being realized. This article describes the advantages, disadvantages (and potential solutions to them), and future possibilities of 3.0-T imaging.
|
Advantages
|
|---|
The SNR describes the amount of usable signal, relative to the background noise, that is available to create the MR image. 
The SNR varies linearly with the field strength. At 3.0 T, twice as many protons as at 1.5 T are aligned along the magnetic field, a condition that creates the potential for a doubling of the derived signal (Fig 1). However, because of certain limitations of 3.0-T MR imaging, including alterations in relaxation times and in total body heating, the realized gain in SNR over that at 1.5 T is usually 1.7- to 1.8-fold (1). Within a particular examination, the higher SNR can be exploited in two different ways: either to increase the spatial resolution or indirectly to decrease the acquisition time.
View larger version (7K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1. Diagram shows the basis of increased signal at 3.0 T: a proportional increase in the number of protons aligned in the direction of the main magnetic field. The number of protons that contribute to the MR signal remains a fraction of the total number within the imaged object.
|
|
The improved spatial resolution at high magnetic field strengths is a function of the increased SNR, which allows larger matrix dimensions (ie, smaller pixels and thinner sections) for a given field of view (FOV). This increased spatial resolution at axial (in-plane) imaging has the potential to improve lesion visibility (Fig 2). The finer detail on reformatted (through-plane) images may aid in lesion characterization (Fig 3). Alternatively, improvements in SNR can be traded off for faster acquisition times to reduce motion artifacts by easing breath-hold requirements or to increase patient throughput.
View larger version (140K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2a. Comparison of prostate images acquired with an endorectal coil at 1.5 T (a, c) and 1 year later at 3.0 T (b, d) in a patient with benign prostatic hyperplasia shows improved resolution at 3.0 T, with sharper delineation of the margins of the central gland (arrowheads) and of nodules. The SNR remains robust despite a 44% reduction in voxel size at 3.0 T. Axial unenhanced images (a, b) were obtained with a fast SE sequence (a: TR msec/ echo time [TE] msec, 7000/161; section thickness, 3 mm; FOV, 16; matrix, 320 x 192; number of signals acquired [NSA], six; b: 3900/160; section thickness, 2.2 mm; FOV, 14; matrix, 320 x 192; NSA, four). Axial contrast-enhanced images (c, d) were obtained with a spoiled gradient-echo sequence (c: 9/4; section thickness, 3.2 mm; FOV, 16; matrix, 256 x 160; NSA, two; d: 7/2; section thickness, 3 mm; FOV, 14; matrix, 256 x 192; NSA, two).
|
|
View larger version (118K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2b. Comparison of prostate images acquired with an endorectal coil at 1.5 T (a, c) and 1 year later at 3.0 T (b, d) in a patient with benign prostatic hyperplasia shows improved resolution at 3.0 T, with sharper delineation of the margins of the central gland (arrowheads) and of nodules. The SNR remains robust despite a 44% reduction in voxel size at 3.0 T. Axial unenhanced images (a, b) were obtained with a fast SE sequence (a: TR msec/ echo time [TE] msec, 7000/161; section thickness, 3 mm; FOV, 16; matrix, 320 x 192; number of signals acquired [NSA], six; b: 3900/160; section thickness, 2.2 mm; FOV, 14; matrix, 320 x 192; NSA, four). Axial contrast-enhanced images (c, d) were obtained with a spoiled gradient-echo sequence (c: 9/4; section thickness, 3.2 mm; FOV, 16; matrix, 256 x 160; NSA, two; d: 7/2; section thickness, 3 mm; FOV, 14; matrix, 256 x 192; NSA, two).
|
|
View larger version (117K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2c. Comparison of prostate images acquired with an endorectal coil at 1.5 T (a, c) and 1 year later at 3.0 T (b, d) in a patient with benign prostatic hyperplasia shows improved resolution at 3.0 T, with sharper delineation of the margins of the central gland (arrowheads) and of nodules. The SNR remains robust despite a 44% reduction in voxel size at 3.0 T. Axial unenhanced images (a, b) were obtained with a fast SE sequence (a: TR msec/ echo time [TE] msec, 7000/161; section thickness, 3 mm; FOV, 16; matrix, 320 x 192; number of signals acquired [NSA], six; b: 3900/160; section thickness, 2.2 mm; FOV, 14; matrix, 320 x 192; NSA, four). Axial contrast-enhanced images (c, d) were obtained with a spoiled gradient-echo sequence (c: 9/4; section thickness, 3.2 mm; FOV, 16; matrix, 256 x 160; NSA, two; d: 7/2; section thickness, 3 mm; FOV, 14; matrix, 256 x 192; NSA, two).
|
|
View larger version (105K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2d. Comparison of prostate images acquired with an endorectal coil at 1.5 T (a, c) and 1 year later at 3.0 T (b, d) in a patient with benign prostatic hyperplasia shows improved resolution at 3.0 T, with sharper delineation of the margins of the central gland (arrowheads) and of nodules. The SNR remains robust despite a 44% reduction in voxel size at 3.0 T. Axial unenhanced images (a, b) were obtained with a fast SE sequence (a: TR msec/ echo time [TE] msec, 7000/161; section thickness, 3 mm; FOV, 16; matrix, 320 x 192; number of signals acquired [NSA], six; b: 3900/160; section thickness, 2.2 mm; FOV, 14; matrix, 320 x 192; NSA, four). Axial contrast-enhanced images (c, d) were obtained with a spoiled gradient-echo sequence (c: 9/4; section thickness, 3.2 mm; FOV, 16; matrix, 256 x 160; NSA, two; d: 7/2; section thickness, 3 mm; FOV, 14; matrix, 256 x 192; NSA, two).
|
|
View larger version (107K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3a. Comparison of axial (a, c) and coronal (b, d) reformatted images obtained at 1.5 T (a, b) and subsequently at 3.0 T (c, d) shows the clearer depiction of a right adrenal mass (arrow in a and c, arrowhead in b and d) at 3.0 T. Because of the higher SNR at 3.0 T, the voxel size could be reduced, allowing increased spatial resolution, while the SNR was maintained. In b, the lesion appears to be extra-adrenal and to have displaced the entire adrenal gland laterally. In d, it is clear that the mass arises from the medial limb of the adrenal gland and splays the limbs of the gland. The lesion was resected and was found at pathologic analysis to be an adrenal pheochromocytoma. Parameters at 1.5-T imaging were as follows: 4.0/1.9; matrix, 256 x 192; FOV, 31 cm; reformatted section thicknesses, 4 mm (a) and 2 mm (b). Parameters at 3.0-T imaging were as follows: 5.4/2.5; matrix, 320 x 224; FOV, 35 cm; reformatted section thicknesses, 3 mm (c) and 1.5 mm (d).
|
|
View larger version (130K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3b. Comparison of axial (a, c) and coronal (b, d) reformatted images obtained at 1.5 T (a, b) and subsequently at 3.0 T (c, d) shows the clearer depiction of a right adrenal mass (arrow in a and c, arrowhead in b and d) at 3.0 T. Because of the higher SNR at 3.0 T, the voxel size could be reduced, allowing increased spatial resolution, while the SNR was maintained. In b, the lesion appears to be extra-adrenal and to have displaced the entire adrenal gland laterally. In d, it is clear that the mass arises from the medial limb of the adrenal gland and splays the limbs of the gland. The lesion was resected and was found at pathologic analysis to be an adrenal pheochromocytoma. Parameters at 1.5-T imaging were as follows: 4.0/1.9; matrix, 256 x 192; FOV, 31 cm; reformatted section thicknesses, 4 mm (a) and 2 mm (b). Parameters at 3.0-T imaging were as follows: 5.4/2.5; matrix, 320 x 224; FOV, 35 cm; reformatted section thicknesses, 3 mm (c) and 1.5 mm (d).
|
|
View larger version (120K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3c. Comparison of axial (a, c) and coronal (b, d) reformatted images obtained at 1.5 T (a, b) and subsequently at 3.0 T (c, d) shows the clearer depiction of a right adrenal mass (arrow in a and c, arrowhead in b and d) at 3.0 T. Because of the higher SNR at 3.0 T, the voxel size could be reduced, allowing increased spatial resolution, while the SNR was maintained. In b, the lesion appears to be extra-adrenal and to have displaced the entire adrenal gland laterally. In d, it is clear that the mass arises from the medial limb of the adrenal gland and splays the limbs of the gland. The lesion was resected and was found at pathologic analysis to be an adrenal pheochromocytoma. Parameters at 1.5-T imaging were as follows: 4.0/1.9; matrix, 256 x 192; FOV, 31 cm; reformatted section thicknesses, 4 mm (a) and 2 mm (b). Parameters at 3.0-T imaging were as follows: 5.4/2.5; matrix, 320 x 224; FOV, 35 cm; reformatted section thicknesses, 3 mm (c) and 1.5 mm (d).
|
|
View larger version (137K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3d. Comparison of axial (a, c) and coronal (b, d) reformatted images obtained at 1.5 T (a, b) and subsequently at 3.0 T (c, d) shows the clearer depiction of a right adrenal mass (arrow in a and c, arrowhead in b and d) at 3.0 T. Because of the higher SNR at 3.0 T, the voxel size could be reduced, allowing increased spatial resolution, while the SNR was maintained. In b, the lesion appears to be extra-adrenal and to have displaced the entire adrenal gland laterally. In d, it is clear that the mass arises from the medial limb of the adrenal gland and splays the limbs of the gland. The lesion was resected and was found at pathologic analysis to be an adrenal pheochromocytoma. Parameters at 1.5-T imaging were as follows: 4.0/1.9; matrix, 256 x 192; FOV, 31 cm; reformatted section thicknesses, 4 mm (a) and 2 mm (b). Parameters at 3.0-T imaging were as follows: 5.4/2.5; matrix, 320 x 224; FOV, 35 cm; reformatted section thicknesses, 3 mm (c) and 1.5 mm (d).
|
|
The contrast-to-noise ratio describes the extent to which different objects on an image can be distinguished. This ability is one of the principal advantages of MR imaging over modalities such as computed tomography and ultrasonography. Contrast in MR imaging is derived mainly from intrinsic tissue relaxation kinetics, which may be supplemented by the effects of exogenous contrast media. The intrinsic tissue relaxation kinetics defined by T1, T2, and T2* values varies slightly at higher field strengths, causing a decrease in intrinsic image contrast (Fig 4). However, pulse sequences may be adapted to exploit these differences in relaxation kinetics so as to minimize intrinsic tissue contrast losses at 3.0 T (2,3). The improved image contrast at higher field strengths is largely the result of exogenous contrast media such as gadolinium, a paramagnetic substance that disrupts the local magnetic field and leads to T1 shortening (4). T1 generally is lengthened at 3.0-T imaging, even with the use of a paramagnetic agent such as gadolinium. However, because the T1 of gadolinium is shorter than that of the soft tissues, gadolinium-enhanced tissues stand out markedly against the background. Diagnostic sensitivity is improved by the enhanced contrast (Fig 5), and this improvement in turn provides an opportunity for gadolinium dose reduction (5).
View larger version (118K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4a. Comparison of liver-spleen contrast on T1-weighted images obtained at 1.5 T (a) and at 3.0 T (b) shows decreased contrast in b because of increased T1 at the higher field strength. The decrease in contrast might have been mitigated by altering the pulse sequence used at 3.0-T imaging. Parameters at 1.5-T imaging were as follows: 180/2.34; section thickness, 7 mm; matrix, 256 x 123; flip angle, 70°. Parameters at 3.0-T imaging were the same, except for the matrix, which was 320 x 192.
|
|
View larger version (121K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4b. Comparison of liver-spleen contrast on T1-weighted images obtained at 1.5 T (a) and at 3.0 T (b) shows decreased contrast in b because of increased T1 at the higher field strength. The decrease in contrast might have been mitigated by altering the pulse sequence used at 3.0-T imaging. Parameters at 1.5-T imaging were as follows: 180/2.34; section thickness, 7 mm; matrix, 256 x 123; flip angle, 70°. Parameters at 3.0-T imaging were the same, except for the matrix, which was 320 x 192.
|
|
View larger version (118K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5a. Comparison of contrast-enhanced three-dimensional T1-weighted fat-saturated images obtained at 1.5 T (a) and 3.0 T (b) in a patient with focal nodular hyperplasia demonstrates an improved contrast-to-noise ratio at 3.0 T. Although the intrinsic image contrast is decreased at 3.0 T, the T1 shortening effect of gadolinium relative to adjacent tissues is more pronounced; therefore, the liver lesion (arrow) stands out more markedly from the adjacent parenchyma, and the outlines of the portal vein (arrowhead) are more readily apparent at 3.0 T than at 1.5 T. Parameters at 1.5-T imaging were as follows: 4.3/1.98; section thickness, 4.4 mm; matrix, 256 x 154. Parameters at 3.0-T imaging were as follows: 3.9/1.06; section thickness, 3.6 mm; matrix, 320 x 224.
|
|
View larger version (147K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5b. Comparison of contrast-enhanced three-dimensional T1-weighted fat-saturated images obtained at 1.5 T (a) and 3.0 T (b) in a patient with focal nodular hyperplasia demonstrates an improved contrast-to-noise ratio at 3.0 T. Although the intrinsic image contrast is decreased at 3.0 T, the T1 shortening effect of gadolinium relative to adjacent tissues is more pronounced; therefore, the liver lesion (arrow) stands out more markedly from the adjacent parenchyma, and the outlines of the portal vein (arrowhead) are more readily apparent at 3.0 T than at 1.5 T. Parameters at 1.5-T imaging were as follows: 4.3/1.98; section thickness, 4.4 mm; matrix, 256 x 154. Parameters at 3.0-T imaging were as follows: 3.9/1.06; section thickness, 3.6 mm; matrix, 320 x 224.
|
|
At MR spectroscopy, the increased SNR translates into increased sensitivity and specificity. Because the amount of signal derived from each metabolite is increased, the metabolite peaks are more easily differentiated from the background. In addition, the increased frequency spread between individual metabolites at 3.0 T results in improved distinction between them (Fig 6). Finally, when the SNR is higher, the measurement times for acquiring specific data can be reduced. Such reductions may be particularly advantageous for in vivo imaging, in which data acquisition may be limited by patient motion.
View larger version (110K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6a. Breath-hold MR spectroscopy at 3.0 T. (a) Single-shot fast SE image shows a voxel ( ) selected for spectroscopic analysis within a renal cell carcinoma metastasis in the right adrenal gland. (b) Spectrum from the selected voxel shows clear separation of the metabolite peaks, which are easily distinguishable from background noise. Note the marked trimethylamine (TMA) or choline peak at 3.2 ppm, a feature associated with malignancy. The increased SNR and increased spectral dispersion at 3.0 T allow clearer identification and separation of metabolites. In addition, the increased SNR may enable a reduction in acquisition time, allowing spectroscopy to be performed in a single breath hold and thus eliminating respiratory motion.
|
|
View larger version (36K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6b. Breath-hold MR spectroscopy at 3.0 T. (a) Single-shot fast SE image shows a voxel () selected for spectroscopic analysis within a renal cell carcinoma metastasis in the right adrenal gland. (b) Spectrum from the selected voxel shows clear separation of the metabolite peaks, which are easily distinguishable from background noise. Note the marked trimethylamine (TMA) or choline peak at 3.2 ppm, a feature associated with malignancy. The increased SNR and increased spectral dispersion at 3.0 T allow clearer identification and separation of metabolites. In addition, the increased SNR may enable a reduction in acquisition time, allowing spectroscopy to be performed in a single breath hold and thus eliminating respiratory motion.
|
|
|
Disadvantages
|
|---|
Before the benefits of 3.0-T imaging can be attained, the limitations of high field strength must be surmounted. Although the limiting factors overlap and interact, for ease of discussion they are considered here in the following categories: physics and technology, sequence optimization, artifacts, and safety.
Physics and Technology
RF Field Inhomogeneity.—
RF field inhomogeneity may represent the most formidable challenge to clinical imaging at 3.0 T, particularly in the abdomen. The increase in field strength translates into an increase in resonance frequency and, therefore, a decrease in the RF wavelength. In water and human tissue, the decreased RF wavelength may approximate the size of the field of view. When this occurs, the result is a standing wave pattern across the image—often referred to as the dielectric effect. Constructive or destructive interference from standing RF waves results in areas of brightening or darkening, respectively. The larger the region of interest in comparison with the wavelength, the more pronounced the artifact. For this reason, standing wave artifacts are seen more often in obese patients with a distended abdomen than in nonobese patients (Fig 7) (1,3).
View larger version (151K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7a. RF magnetic field inhomogeneity. Comparison of coronal single-shot fast SE images acquired at 1.5 T (a) and 24 hours later at 3.0 T (b) shows improved contrast of a tumor thrombus (arrow) in the left renal vein and an overall increase in the SNR with less graininess at 3.0 T. However, a standing wave artifact at 3.0 T (* in b) obscures the liver, particularly the dome, which has more-uniform signal intensity at 1.5 T (* in a). Parameters at 1.5-T imaging were as follows: 911/76; section thickness, 5 mm; matrix, 256 x 205; per-pixel bandwidth, 488 Hz. Parameters at 3.0-T imaging were as follows: 1168/59; section thickness, 4.6 mm; matrix, 256 x 192; per-pixel bandwidth, 651 Hz.
|
|
View larger version (149K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7b. RF magnetic field inhomogeneity. Comparison of coronal single-shot fast SE images acquired at 1.5 T (a) and 24 hours later at 3.0 T (b) shows improved contrast of a tumor thrombus (arrow) in the left renal vein and an overall increase in the SNR with less graininess at 3.0 T. However, a standing wave artifact at 3.0 T (* in b) obscures the liver, particularly the dome, which has more-uniform signal intensity at 1.5 T (* in a). Parameters at 1.5-T imaging were as follows: 911/76; section thickness, 5 mm; matrix, 256 x 205; per-pixel bandwidth, 488 Hz. Parameters at 3.0-T imaging were as follows: 1168/59; section thickness, 4.6 mm; matrix, 256 x 192; per-pixel bandwidth, 651 Hz.
|
|
A related artifact is thought to be caused by the interference of electric currents produced in highly conductive tissue (ie, in ascites) in RF send-receive transmissions. A rapidly changing magnetic field like that in RF transmissions induces a circulating electric field. When this happens in a conductive medium, a circulating electric current is established. This current acts as an electromagnet that opposes the changing magnetic field, reducing the amplitude and dissipating the energy of the RF field. The more conductive the medium, the stronger the opposing electromagnet and the greater the attenuation of the RF field at that location. An archetype of 3.0 T–specific artifacts often is seen in patients with ascites, in whom abdominal distention produces the standing wave phenomenon, while the highly conductive ascitic fluid may cause local regions of signal loss in the abdomen (Fig 8) (3,6,7).
View larger version (175K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 8a. Signal loss due to ascites at 3.0 T. Comparison of coronal images obtained with a single-shot fast SE sequence at 1.5 T (a) and 3.0 T (b) shows dramatic standing wave and conductivity effects at the higher field strength because of ascitic fluid, which produced a nonuniform lower RF field centrally, seen in b. The programmed excitation and refocusing pulse angles were effectively reduced, yielding signal reduction or loss. In this case, the patient was first examined at 3.0 T and then was transferred to a 1.5-T system to obtain images of higher diagnostic quality. Parameters at 1.5-T imaging were as follows: 1157/58; section thickness, 4 mm; matrix, 256 x 256. Parameters at 3.0-T imaging were as follows: 925/58; section thickness, 4.6 mm; matrix, 256 x 256.
|
|
View larger version (151K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 8b. Signal loss due to ascites at 3.0 T. Comparison of coronal images obtained with a single-shot fast SE sequence at 1.5 T (a) and 3.0 T (b) shows dramatic standing wave and conductivity effects at the higher field strength because of ascitic fluid, which produced a nonuniform lower RF field centrally, seen in b. The programmed excitation and refocusing pulse angles were effectively reduced, yielding signal reduction or loss. In this case, the patient was first examined at 3.0 T and then was transferred to a 1.5-T system to obtain images of higher diagnostic quality. Parameters at 1.5-T imaging were as follows: 1157/58; section thickness, 4 mm; matrix, 256 x 256. Parameters at 3.0-T imaging were as follows: 925/58; section thickness, 4.6 mm; matrix, 256 x 256.
|
|
Improved coil design may compensate for some of these effects. While the SNR advantages of a phased-array coil over traditional body coils are well known (Fig 9) (8), phased-array coils do little to correct the dielectric effect. However, alternative transmission coil designs, such as a spiral configuration, may alter current patterns and influence B1 (9). Multiple transmission coils also offer some improvement. An off-resonance coil placed between the transmission coil and the patient may function as a dielectric, changing the pattern of RF transmission and thus favorably altering the B1 field (10). Newer coils, such as the transverse electromagnetic body coil, may reduce RF field inhomogeneity in body imaging at a field strength of 3.0 T and above (11). The shield or cavity wall integral to the transverse electromagnetic body coil design effectively suppresses eddy currents that may interfere with anatomic and spectroscopic applications at higher field strengths.
View larger version (132K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 9a. Effect of coil design on SNR. Comparison of images obtained with a body coil (a, c) and with a phased-array coil (b, d) at 1.5 T (1204/60.1; matrix, 256 x 192) (a, b) and at 3.0 T (25,224/65.0; matrix, 256 x 192) (c, d) shows that a small surface coil has an inherent SNR advantage over a body coil at both field strengths. Image quality with use of the body coil at 3.0 T (c) approaches that with the phased-array coil at 1.5 T (b), but the best image quality is demonstrated with the phased-array coil at 3.0 T (d).
|
|
View larger version (118K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 9b. Effect of coil design on SNR. Comparison of images obtained with a body coil (a, c) and with a phased-array coil (b, d) at 1.5 T (1204/60.1; matrix, 256 x 192) (a, b) and at 3.0 T (25,224/65.0; matrix, 256 x 192) (c, d) shows that a small surface coil has an inherent SNR advantage over a body coil at both field strengths. Image quality with use of the body coil at 3.0 T (c) approaches that with the phased-array coil at 1.5 T (b), but the best image quality is demonstrated with the phased-array coil at 3.0 T (d).
|
|
View larger version (128K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 9c. Effect of coil design on SNR. Comparison of images obtained with a body coil (a, c) and with a phased-array coil (b, d) at 1.5 T (1204/60.1; matrix, 256 x 192) (a, b) and at 3.0 T (25,224/65.0; matrix, 256 x 192) (c, d) shows that a small surface coil has an inherent SNR advantage over a body coil at both field strengths. Image quality with use of the body coil at 3.0 T (c) approaches that with the phased-array coil at 1.5 T (b), but the best image quality is demonstrated with the phased-array coil at 3.0 T (d).
|
|
View larger version (116K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 9d. Effect of coil design on SNR. Comparison of images obtained with a body coil (a, c) and with a phased-array coil (b, d) at 1.5 T (1204/60.1; matrix, 256 x 192) (a, b) and at 3.0 T (25,224/65.0; matrix, 256 x 192) (c, d) shows that a small surface coil has an inherent SNR advantage over a body coil at both field strengths. Image quality with use of the body coil at 3.0 T (c) approaches that with the phased-array coil at 1.5 T (b), but the best image quality is demonstrated with the phased-array coil at 3.0 T (d).
|
|
Improved coils alone may not remove all the inhomogeneity. Thus, recent pulse sequence approaches, including adiabatic pulses (12), impulse two-dimensional pulses (13,14), and, most recently, three-dimensional tailored RF pulses (15), have been proposed and have proved particularly useful in body imaging. However, these methods are specific to the coil type and the imaging specifications.
Energy Deposition.—
RF pulses are used to stimulate the proton spins of a particular object in a magnetic field. This leads to energy transfer from the RF pulse to the investigated object, which generates heat. If not controlled, the heat produced can have detrimental physiologic effects, including changes in mental function and cardiac output (16). The SAR provides an estimate for the energy deposited in tissue by the RF pulse and the potential for heating the tissue. SAR limits are set by the FDA to prevent total body heating by more than 1°C or 4 W/kg averaged over the whole body for 15 minutes (17).
The SAR increases with the square of the resonance frequency and, therefore, the square of the magnetic field. The SAR also increases with the square of the flip angle, the size of the patient, and the duty cycle of the RF pulse. This is especially true for SAR-intensive sequences such as fast spin-echo (SE), balanced steady-state, or magnetization transfer sequences, as well as for sequences that include a fat saturation pulse. Proposed solutions to mitigate SAR increases incurred with higher field strengths usually involve undesirable trade-offs such as increased image acquisition times, decreased in-plane and through-plane resolution, or decreased SNR. For example, lower or asymptotic flip angles may decrease signal and image contrast, whereas respiratory triggering, shorter echo train lengths, wider interecho spacing, insertion of dead time, and lengthening of TR would increase the acquisition time (18). New or modified pulse sequence designs, RF pulse designs, acquisition techniques, and hardware designs are being developed to allow better management of the SAR at high-field-strength imaging.
Parallel imaging provides an elegant solution to this trade-off. Unlike sequential acquisitions, parallel imaging is based on the use of coils with multiple small detectors that operate simultaneously to acquire MR data. Each of these detectors contains spatial information that can be used as a substitute for time-consuming phase encoding steps, thereby allowing both the acquisition time and the SAR to be reduced (19,20). However, parallel imaging also has inherent drawbacks, including a decrease in the SNR. While this effect is somewhat counterbalanced by the inherent increase in the SNR at high field strength, further strategies are needed to maximize all the benefits of 3.0-T imaging simultaneously.
One such strategy involves the combined use of parallel imaging with a single-shot approach to regain the lost SNR by decreasing the bandwidth. When used with single-shot T2-weighted echo-train MR imaging sequences such as half-Fourier rapid acquisition with relaxation enhancement (eg, HASTE) and single-shot fast SE, parallel imaging can save time by reducing the number of echoes in each echo train. This strategy removes the later, low-amplitude echoes that cause image blurring and thus increases the image sharpness without a change in the matrix size. A penalty is paid in the form of a decrease in the SNR because fewer echoes are sampled; however, the loss of signal can be recouped by reducing the receiver bandwidth. Although the reduction of bandwidth lengthens the duration of the echo train by increasing the sampling time and thus the echo spacing, the bandwidth reduction needed to recover the loss in SNR does not decrease image sharpness to the same degree as would nonparallel acquisition (21) (Fig 10).
View larger version (11K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 10a. Diagrams show the use of a parallel imaging technique with a single-shot fast SE sequence to decrease acquisition time and motion-related artifacts by summing multiple echoes (TE) within a single TR interval. (a) Single-shot fast SE sequence without the use of parallel imaging. (b–d) Use of a parallel imaging technique with the sequence diagrammed in a: Intermittent phase encoding steps (lower dashed arrows in b) and received echoes (upper dashed arrows in b) are excluded to decrease the echo train duration and, thus, the image acquisition time, producing the result shown in c. The missing data are then virtually recreated by using the known sensitivity profiles of the individual elements in the multiple-element receiver coil. The reduction in the SNR that occurs with a decrease in the number of phase encoding steps may be mitigated by decreasing the bandwidth. Although a decrease in the bandwidth in turn results in an increase in the echo spacing and, thus, the echo train duration (d), the trade-off yields images with less motion-related artifact and a lesser penalty on the SNR than those acquired without the use of parallel imaging.
|
|
View larger version (12K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 10b. Diagrams show the use of a parallel imaging technique with a single-shot fast SE sequence to decrease acquisition time and motion-related artifacts by summing multiple echoes (TE) within a single TR interval. (a) Single-shot fast SE sequence without the use of parallel imaging. (b–d) Use of a parallel imaging technique with the sequence diagrammed in a: Intermittent phase encoding steps (lower dashed arrows in b) and received echoes (upper dashed arrows in b) are excluded to decrease the echo train duration and, thus, the image acquisition time, producing the result shown in c. The missing data are then virtually recreated by using the known sensitivity profiles of the individual elements in the multiple-element receiver coil. The reduction in the SNR that occurs with a decrease in the number of phase encoding steps may be mitigated by decreasing the bandwidth. Although a decrease in the bandwidth in turn results in an increase in the echo spacing and, thus, the echo train duration (d), the trade-off yields images with less motion-related artifact and a lesser penalty on the SNR than those acquired without the use of parallel imaging.
|
|
View larger version (6K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 10c. Diagrams show the use of a parallel imaging technique with a single-shot fast SE sequence to decrease acquisition time and motion-related artifacts by summing multiple echoes (TE) within a single TR interval. (a) Single-shot fast SE sequence without the use of parallel imaging. (b–d) Use of a parallel imaging technique with the sequence diagrammed in a: Intermittent phase encoding steps (lower dashed arrows in b) and received echoes (upper dashed arrows in b) are excluded to decrease the echo train duration and, thus, the image acquisition time, producing the result shown in c. The missing data are then virtually recreated by using the known sensitivity profiles of the individual elements in the multiple-element receiver coil. The reduction in the SNR that occurs with a decrease in the number of phase encoding steps may be mitigated by decreasing the bandwidth. Although a decrease in the bandwidth in turn results in an increase in the echo spacing and, thus, the echo train duration (d), the trade-off yields images with less motion-related artifact and a lesser penalty on the SNR than those acquired without the use of parallel imaging.
|
|
View larger version (7K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 10d. Diagrams show the use of a parallel imaging technique with a single-shot fast SE sequence to decrease acquisition time and motion-related artifacts by summing multiple echoes (TE) within a single TR interval. (a) Single-shot fast SE sequence without the use of parallel imaging. (b–d) Use of a parallel imaging technique with the sequence diagrammed in a: Intermittent phase encoding steps (lower dashed arrows in b) and received echoes (upper dashed arrows in b) are excluded to decrease the echo train duration and, thus, the image acquisition time, producing the result shown in c. The missing data are then virtually recreated by using the known sensitivity profiles of the individual elements in the multiple-element receiver coil. The reduction in the SNR that occurs with a decrease in the number of phase encoding steps may be mitigated by decreasing the bandwidth. Although a decrease in the bandwidth in turn results in an increase in the echo spacing and, thus, the echo train duration (d), the trade-off yields images with less motion-related artifact and a lesser penalty on the SNR than those acquired without the use of parallel imaging.
|
|
Sequence Optimization
Intrinsic tissue relaxation parameters all vary slightly with increasing magnetic field strength. In particular, with a field strength increase from 1.5 T to 3.0 T, the T1 is increased, T2* is decreased, and T2 is decreased slightly or stays the same (22–24).
T1, also known as longitudinal or spin-lattice relaxation time, is a property of a proton within a given molecular environment but also depends on the static magnetic field. With an increase in field strength from 1.5 T to 3.0 T, the T1 in soft tissues increases; this change may result in reduced relative signal intensity and contrast at 3.0-T imaging if T1-weighted sequences are applied with the same TR as at 1.5-T imaging (Fig 5) (11,17–20,25–27). The T1-weighted sequences that have been developed for use at 1.5 T must be modified for optimal imaging at 3.0 T. While it might make sense to simply increase the TR of all pulse sequences at 3.0 T, this would have the undesirable effect of increasing the acquisition time. Again, parallel imaging may provide a solution to this problem of increased time, but with the trade-off of a decreased SNR. Alternatively, specific inversion recovery or magnetization-prepared techniques may be implemented to achieve the desired resolution or contrast. Knowledge of the T1 values of tissues may facilitate an objective selection of TR, flip angle, and—particularly for inversion recovery sequences—inversion time to optimize the contrast between selected tissues or organs (20). For example, at short inversion time inversion-recovery imaging of the liver with the 3.0-T system at our institution, we found that an inversion time of 170 msec resulted in strong suppression of the fat signal and high conspicuity of focal liver masses.
T2, also known as transverse or spin-spin relaxation time, is a property that reflects a particular local microscopic environment. Although it is subject to the effects of the magnetic field, T2 typically is unchanged or only slightly decreased with increasing magnetic field strengths. This is most likely because some mechanisms of T2 are prolonged while others become more efficient with increased field strength.
T2* is the observed or effective T2 value; it is a composite of both the intrinsic T2 of a tissue and the superimposed relaxation effects from local field inhomogeneities. The effect of T2* (proton dephasing) is more pronounced at 3.0 T and results in greater magnetic susceptibility than at 1.5 T.
Artifacts
Chemical Shift.—
Chemical shift refers to the resonance frequency variations that result from intrinsic magnetic shielding of different chemical species. Chemical shift artifact of the first kind is due to a difference between the resonance frequency of protons in water and that of protons in fat and is directly proportional to the strength of the main magnetic field. This difference causes a chemical shift misregistration artifact that is seen only along the frequency-encoding axis and the section-selective dimension (1,20,21). It is easily seen around the kidneys. Chemical shift artifact of the first kind appears as a band of low signal intensity toward the lower part of the frequency-encoding gradient field and as a band of high signal intensity toward the higher part of the frequency-encoding 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 imaging as at standard 1.5-T imaging (Fig 11). This enlarged artifact does not usually cause substantial problems at clinical body MR imaging at 3.0 T. However, it may be problematic in some cases, such as in the search for a subcapsular renal hematoma. In such cases, the receiver bandwidth may be increased to minimize the chemical shift artifact (Fig 12). Unfortunately, the improvement comes at the expense of the SNR. Other solutions include the use of saturation pulses and short inversion time inversion-recovery sequences to minimize the fat signal. Swapping the phase- and frequency-encoding directions or changing the polarity of the frequency-encoding gradient may displace the artifact and make it less apparent (Fig 13).
View larger version (68K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 11a. Chemical shift artifact of the first kind. Comparison of axial in-phase gradient-echo images acquired at 1.5 T (a) and at 3.0 T (b) with the same FOV, base resolution, and receiver bandwidth shows a significantly more apparent artifact at 3.0 T. The artifact appears as a low-signal-intensity band (black arrowheads) toward the higher part of the frequency-encoding gradient field and a high-signal-intensity band (white arrowheads) toward the lower part of the frequency-encoding gradient field. The difference in resonance frequency is directly proportional to the main magnetic field strength. Parameters at 1.5-T imaging were as follows: 180/4.2; bandwidth, 15 kHz; section thickness, 7 mm; matrix, 256 x 160. Parameters at 3.0-T imaging were the same except for the TE, which was 2.1 msec.
|
|
View larger version (64K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 11b. Chemical shift artifact of the first kind. Comparison of axial in-phase gradient-echo images acquired at 1.5 T (a) and at 3.0 T (b) with the same FOV, base resolution, and receiver bandwidth shows a significantly more apparent artifact at 3.0 T. The artifact appears as a low-signal-intensity band (black arrowheads) toward the higher part of the frequency-encoding gradient field and a high-signal-intensity band (white arrowheads) toward the lower part of the frequency-encoding gradient field. The difference in resonance frequency is directly proportional to the main magnetic field strength. Parameters at 1.5-T imaging were as follows: 180/4.2; bandwidth, 15 kHz; section thickness, 7 mm; matrix, 256 x 160. Parameters at 3.0-T imaging were the same except for the TE, which was 2.1 msec.
|
|
View larger version (64K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 12a. Increasing the bandwidth to lessen chemical shift artifact of the first kind. Comparison of axial in-phase gradient-echo images acquired at 3.0 T (180/2.1; section thickness, 7 mm; matrix, 256 x 160) with bandwidths of 15 kHz (a) and 32 kHz (b) demonstrates a less apparent artifact in b, with a decrease in the width of the low-signal-intensity bands (black arrowheads) and high-signal-intensity bands (white arrowheads) at the kidney margins as well as the low-signal-intensity band at the liver margin (black arrow).
|
|
View larger version (65K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 12b. Increasing the bandwidth to lessen chemical shift artifact of the first kind. Comparison of axial in-phase gradient-echo images acquired at 3.0 T (180/2.1; section thickness, 7 mm; matrix, 256 x 160) with bandwidths of 15 kHz (a) and 32 kHz (b) demonstrates a less apparent artifact in b, with a decrease in the width of the low-signal-intensity bands (black arrowheads) and high-signal-intensity bands (white arrowheads) at the kidney margins as well as the low-signal-intensity band at the liver margin (black arrow).
|
|
View larger version (63K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 13a. Avoiding chemical shift artifact of the first kind while maintaining a constant bandwidth. Axial in-phase gradient-echo images acquired at 3.0 T (180/ 2.1; section thickness, 7 mm; matrix, 256 x 160; bandwidth, 15 kHz) show chemical shift artifacts at the margins of the kidneys (arrowheads) and liver (arrow). The artifacts are in the usual positions and are most obvious in a, an image acquired with the frequency-encoding gradient applied along the transverse axis. In b, an image obtained by applying the frequency-encoding gradient in the anteroposterior direction, the position of the artifacts has shifted and allows evaluation of the regions that were obscured in a. In c, an image obtained with a fat saturation technique and with the same orientation of the frequency-encoding gradient as in a, the artifacts are much less obtrusive.
|
|
View larger version (53K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 13b. Avoiding chemical shift artifact of the first kind while maintaining a constant bandwidth. Axial in-phase gradient-echo images acquired at 3.0 T (180/ 2.1; section thickness, 7 mm; matrix, 256 x 160; bandwidth, 15 kHz) show chemical shift artifacts at the margins of the kidneys (arrowheads) and liver (arrow). The artifacts are in the usual positions and are most obvious in a, an image acquired with the frequency-encoding gradient applied along the transverse axis. In b, an image obtained by applying the frequency-encoding gradient in the anteroposterior direction, the position of the artifacts has shifted and allows evaluation of the regions that were obscured in a. In c, an image obtained with a fat saturation technique and with the same orientation of the frequency-encoding gradient as in a, the artifacts are much less obtrusive.
|
|
View larger version (61K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 13c. Avoiding chemical shift artifact of the first kind while maintaining a constant bandwidth. Axial in-phase gradient-echo images acquired at 3.0 T (180/ 2.1; section thickness, 7 mm; matrix, 256 x 160; bandwidth, 15 kHz) show chemical shift artifacts at the margins of the kidneys (arrowheads) and liver (arrow). The artifacts are in the usual positions and are most obvious in a, an image acquired with the frequency-encoding gradient applied along the transverse axis. In b, an image obtained by applying the frequency-encoding gradient in the anteroposterior direction, the position of the artifacts has shifted and allows evaluation of the regions that were obscured in a. In c, an image obtained with a fat saturation technique and with the same orientation of the frequency-encoding gradient as in a, the artifacts are much less obtrusive.
|
|
Chemical shift artifact of the second kind is not limited to the frequency-encoding axis but may be seen in all the pixels along a fat-water interface. It is based on an intravoxel phase cancellation effect in voxels in which both fat and water are present (Fig 14) (1,20,21). The size of the artifact does not increase with the main magnetic field strength; instead, it is defined by the spatial resolution of the MR imaging sequence. At gradient-echo in-phase and opposed-phase imaging, the TE must be adjusted because the resonance frequency difference at 3.0 T is twice that at 1.5 T. At 3.0-T MR imaging, both fat and water protons are in phase at roughly 2.2, 4.4, and 6.6 msec (and so on) and out of phase at roughly 1.1, 3.3, and 5.5 msec (and so on). At 1.5-T imaging, fat and water are out of phase at 2.2 msec and in phase at 4.4 msec. In short, by doubling the field strength, we halve the TE for in-phase and opposed-phase imaging. Because problems arise at imaging with interleaved sequences with TE values of less than 2.2 msec, the standard in-phase and out-of-phase TE values used at 3.0-T imaging are 2.2 and 5.5 msec, respectively. Such increases in TE result in greater T2* dephasing and an associated increase in magnetic susceptibility. The increased difference in resonance frequency between water and fat at 3.0 T also may be advantageous because it allows a better separation of the fat and water peaks at MR spectroscopy and better or faster fat suppression.
View larger version (88K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 14a. Chemical shift artifact of the second kind. Comparison of in-phase (a, c) and out-of-phase (b, d) gradient-echo images acquired at 1.5 T (a, b) and at 3.0 T (c, d) demonstrates an increased chemical shift artifact in c and d, a change associated with the use of longer TEs at 3.0 T. To avoid an unacceptable reduction in image quality at very short TEs (such as 2.2 msec and 1.1 msec, which would have been exactly half the TEs used at in-phase and out-of-phase 1.5-T imaging), TEs of 2.3 msec and 5.8 msec were used for in-phase and out-of-phase 3.0-T imaging, respectively. In d, note the increased susceptibility artifacts (arrowheads), which represent siderotic nodules in the liver and Gamna-Gandy bodies in the spleen. These features are more prominent in d than in a or b because of the increased field strength and more visible than in c because of the increased TE.
|
|
View larger version (98K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 14b. Chemical shift artifact of the second kind. Comparison of in-phase (a, c) and out-of-phase (b, d) gradient-echo images acquired at 1.5 T (a, b) and at 3.0 T (c, d) demonstrates an increased chemical shift artifact in c and d, a change associated with the use of longer TEs at 3.0 T. To avoid an unacceptable reduction in image quality at very short TEs (such as 2.2 msec and 1.1 msec, which would have been exactly half the TEs used at in-phase and out-of-phase 1.5-T imaging), TEs of 2.3 msec and 5.8 msec were used for in-phase and out-of-phase 3.0-T imaging, respectively. In d, note the increased susceptibility artifacts (arrowheads), which represent siderotic nodules in the liver and Gamna-Gandy bodies in the spleen. These features are more prominent in d than in a or b because of the increased field strength and more visible than in c because of the increased TE.
|
|
View larger version (112K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 14c. Chemical shift artifact of the second kind. Comparison of in-phase (a, c) and out-of-phase (b, d) gradient-echo images acquired at 1.5 T (a, b) and at 3.0 T (c, d) demonstrates an increased chemical shift artifact in c and d, a change associated with the use of longer TEs at 3.0 T. To avoid an unacceptable reduction in image quality at very short TEs (such as 2.2 msec and 1.1 msec, which would have been exactly half the TEs used at in-phase and out-of-phase 1.5-T imaging), TEs of 2.3 msec and 5.8 msec were used for in-phase and out-of-phase 3.0-T imaging, respectively. In d, note the increased susceptibility artifacts (arrowheads), which represent siderotic nodules in the liver and Gamna-Gandy bodies in the spleen. These features are more prominent in d than in a or b because of the increased field strength and more visible than in c because of the increased TE.
|
|
View larger version (110K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 14d. Chemical shift artifact of the second kind. Comparison of in-phase (a, c) and out-of-phase (b, d) gradient-echo images acquired at 1.5 T (a, b) and at 3.0 T (c, d) demonstrates an increased chemical shift artifact in c and d, a change associated with the use of longer TEs at 3.0 T. To avoid an unacceptable reduction in image quality at very short TEs (such as 2.2 msec and 1.1 msec, which would have been exactly half the TEs used at in-phase and out-of-phase 1.5-T imaging), TEs of 2.3 msec and 5.8 msec were used for in-phase and out-of-phase 3.0-T imaging, respectively. In d, note the increased susceptibility artifacts (arrowheads), which represent siderotic nodules in the liver and Gamna-Gandy bodies in the spleen. These features are more prominent in d than in a or b because of the increased field strength and more visible than in c because of the increased TE.
|
|
Magnetic Susceptibility.—
Susceptibility is the ratio of the internal magnetization induced in the tissue to the magnetization of the external magnetic field. As long as the susceptibility of the tissues imaged is relatively unchanged across the field of view, the magnetic field remains uniform. However, any drastic changes in susceptibility result in distortion of the magnetic field.
The most common alterations in magnetic susceptibility within the body occur at air-tissue interfaces, which cause signal loss due to rapid T2* dephasing. Metallic objects also distort the surrounding magnetic field and produce susceptibility artifacts in the adjacent soft tissues on images. Paramagnetic objects exhibit weak magnetization and increase the local magnetic field, thereby causing artifacts from an induced reduction in local T2*. The latter may be problematic at first-pass contrast-enhanced MR imaging (eg, MR angiography) or when calculating arterial input functions for tumor perfusion assessments (23,28).
At higher field strengths, the magnetic field is more inhomogeneous and more sensitive to T2* dephasing characteristics. The effects may be desirable, as in the increased detectability of blood products or brachytherapy seeds (Fig 15), or undesirable, as in postsurgical or postprocedural imaging in patients with intra-abdominal air and metallic objects that cause significant image distortions (Fig 16). Means for reducing susceptibility artifacts include shim coils to decrease local field inhomogenities, and fast SE sequences that include a 180° inversion pulse to reduce T2* dephasing.
Top
Abstract
Introduction
