DOI: 10.1148/rg.27si075514
RadioGraphics 2007;27:S131-S145
© RSNA, 2007
Breast MR Imaging Artifacts: How to Recognize and Fix Them1
Jennifer A. Harvey, MD,
R. Edward Hendrick, PhD,
Jennifer M. Coll, MD,
Brandi T. Nicholson, MD,
Brian T. Burkholder, RT, and
Michael A. Cohen, MD
1 From the Department of Radiology, University of Virginia, PO Box 800170, Charlottesville, VA 22908 (J.A.H., J.M.C., B.T.N., B.T.B., M.A.C.); and Lynn Sage Comprehensive Breast Center, Northwestern University’s Feinberg Medical School and Northwestern Memorial Hospital, Chicago, Ill (R.E.H.). Presented as an education exhibit at the 2006 RSNA Annual Meeting. Received March 12, 2007; revision requested April 5 and received May 3; accepted May 10. J.A.H. is a researcher with GE Healthcare, Wyeth, and Organon; R.E.H. is with the speakers’ bureau of GE Healthcare and a member of the medical advisory boards of BioLucent and Koning; all remaining authors have no financial relationships to disclose.
Address correspondence to J.A.H. (e-mail: jah7w{at}virginia.edu).
 |
Abstract
|
|---|
Patient and technical factors may lead to unwanted artifacts at breast magnetic resonance (MR) imaging. Use of a properly functioning high-field-strength MR imaging system, a dedicated bilateral breast coil, and an optimal imaging protocol provides a solid framework for performing high-quality breast MR imaging. Problems related to breast positioning, selection of imaging volume, and phase-encoding direction can be overcome by training and providing feedback to MR imaging technologists. Common artifacts seen at breast MR imaging include motion, suboptimal fat suppression, metallic susceptibility, phase wrap, radiofrequency noise, and chemical shift. Once they are recognized, many of these artifacts can be corrected. Protocol monitoring and imaging-based feedback from the interpreting radiologist are essential for minimizing artifacts and optimizing breast MR imaging.
© RSNA, 2007
|
LEARNING OBJECTIVES FOR TEST 4
|
|---|
After reading this article and taking the test, the reader will be able to:
- Discuss the technical requirements for optimal breast MR imaging.
- Identify common artifacts seen at breast MR imaging.
- Describe how to provide imaging-based feedback to technologists to optimize breast MR imaging and reduce artifacts.
|
Introduction
|
|---|
The indications for and use of breast magnetic resonance (MR) imaging have increased over the past decade. Current potential indications for contrast material–enhanced breast MR imaging include (a) evaluation of the extent of known breast cancer in women who desire breast conservation, (b) detection of contralateral breast cancer in women with newly diagnosed breast cancer, (c) screening of women who are at high risk for developing breast cancer, (d) assessment of response to neoadjuvant therapy, (e) evaluation of chest wall invasion in patients with posterior carcinomas, and (f) detection of breast cancer in women with axillary metastasis and normal mammographic findings (1–4).
Breast MR imaging is technically demanding, requiring excellent fat saturation, high spatial resolution, and rapid performance of postcontrast sequences. Patient and technical factors can lead to breast MR imaging artifacts. These artifacts may degrade image quality and confound interpretation.
Breast MR imaging studies may be the only type of MR imaging studies that some radiologists interpret, particularly those who are dedicated exclusively to breast imaging. Technical factors that lead to suboptimal studies may not be recognized due to lack of experience with MR imaging in general. In this article, we review basic breast MR imaging technique and discuss and illustrate patient and technical factors that may result in artifacts. In addition, we provide practical suggestions for improving the quality of breast MR imaging.
|
Basic Breast MR Imaging Technique
|
|---|
A 1.5-T (or higher) magnet provides excellent signal for breast imaging. A breast MR imaging examination performed for potential cancer diagnosis requires the use of a dedicated breast coil, typically with four, seven, or eight channels (3). Bilateral imaging allows assessment of symmetry, which can be helpful in detecting subtle lesions or in determining the significance of the patchy enhancement that is commonly seen in premenopausal women (3,5). In addition, unilateral breast imaging is more susceptible to aliasing or "wraparound" artifacts (3), although this is not typically a problem with current software. Bilateral imaging does not have to result in lower spatial resolution—a function of matrix size, field of view (FOV), and section thickness—compared with unilateral imaging. With the advent of parallel imaging, bilateral imaging can now exceed the spatial resolution of most unilateral imaging studies performed just a few years ago (3).
The patient is positioned prone in the breast coil for the study. For contrast-enhanced breast MR imaging, both breasts may be evaluated simultaneously with a three-dimensional (3D) volume sequence. Fat suppression is used, since an enhancing cancer has high signal intensity similar to that of nonsuppressed fat. Gadolinium-based contrast material is administered intravenously.
Breast MR imaging protocols vary somewhat, particularly between the United States and Europe. In the United States, emphasis has been placed on lesion morphologic features, requiring high spatial resolution and fat suppression (6). In Europe, emphasis has more commonly been placed on lesion dynamic enhancement characteristics, requiring rapid performance of sequences (7). During the development of breast MR imaging, fat suppression was not compatible with the requirement of rapid imaging. Therefore, subtraction of pre- from postcontrast images was used to view the breasts without interference from the high-signal-intensity fat. Fortunately, parallel imaging sequences have been developed in the past few years that allow simultaneous fat suppression, high spatial resolution, and rapid imaging. Pulse sequences providing high spatial resolution for lesion morphologic features and high temporal resolution for dynamic enhancement characteristics can now be used together to help make decisions about lesion outcome (3,8).
There certainly is variability in breast MR imaging protocols, even within the United States. At our institution, we perform breast MR imaging using a Siemens Sonata imaging unit (Siemens, Malvern, Pa) with a 1.5-T magnet. The current protocol at our institution consists of an axial T2-weighted sequence, an axial gradient-echo (GRE) T1-weighted sequence, five axial GRE T1-weighted sequences at 1-minute intervals beginning 30 seconds after the intravenous injection of contrast material, and a sagittal T1-weighted sequence. All sequences are performed with fat suppression. We have recently added an axial non-fat-suppressed T1-weighted sequence prior to contrast material administration, which can be helpful in confirming the presence of fat in the hilum of a lymph node. The technical parameters of breast MR imaging at our institution are presented in the Table. The success of imaging with these parameters will vary depending on the manufacturer and model of the magnet and breast coil used. An alternative to performing an additional sagittal sequence after dynamic imaging is to increase the dynamic imaging time slightly (up to 1
minutes), obtaining isotropic voxels and creating reconstructed images in the sagittal plane. After the examination is performed, the technologist also subtracts the precontrast sequence from each of the five postcontrast T1-weighted sequences and constructs maximum-intensity-projection images from each of the five subtracted series. The total number of images for a routine breast MR imaging examination ranges from 1800 to 2200. Because of the large number of images, all studies are reviewed on a soft-copy review workstation.
|
Patient Factors
|
|---|
Patient factors contribute to the complexity of breast MR imaging. 
Excellent positioning is key in breast MR imaging, just as it is in mammography. Patient motion and metallic artifact may degrade image quality.
Positioning
Each breast should be centered from superior to inferior within the aperture of the dedicated breast coil (Fig 1). Areas of high signal intensity will result where breast tissue is adjacent to coil elements (Fig 1). The breasts should be pulled away from the chest wall into the two holes in the bilateral breast coil. As little breast tissue as possible should remain above the coil apertures. Improper positioning can result in compression of breast tissue between the chest wall and the center partition of the breast coil (Fig 2).
View larger version (59K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1. Improper positioning of the breasts in the breast coil. Sagittal T1-weighted scout MR image (repetition time msec/echo time msec = 20/5) demonstrates signal intensity changes at the superior breast (arrowhead) and inclusion of the upper abdomen (arrow), findings that indicate that the patient is centered too cephalad in the coil. The approximate location of the coil aperture is circled.
|
|
View larger version (61K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2a. Improper positioning. (a) Axial T1-weighted MR image (4/1) shows how, if the breasts are not pulled down into the coil, portions of the breast tissue remain above the coil (arrow) between the chest wall and the center partition of the coil. Because this area is compressed by the patient’s body weight due to prone positioning, the contrast enhancement of the compressed area may be poor. (b) On an axial T1-weighted MR image (4/1) obtained with improved patient positioning, no breast tissue remains above the coil.
|
|
View larger version (80K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2b. Improper positioning. (a) Axial T1-weighted MR image (4/1) shows how, if the breasts are not pulled down into the coil, portions of the breast tissue remain above the coil (arrow) between the chest wall and the center partition of the coil. Because this area is compressed by the patient’s body weight due to prone positioning, the contrast enhancement of the compressed area may be poor. (b) On an axial T1-weighted MR image (4/1) obtained with improved patient positioning, no breast tissue remains above the coil.
|
|
MR imaging technologists may not be accustomed to positioning the breasts and may feel awkward performing this task. Training with a mammography technologist can help MR imaging technologists become more comfortable with positioning of the breasts. An imaging nurse or patient care assistant can also be trained to properly perform this task. A female staff member should be present during positioning. Of note, preserving patient modesty is difficult when moving the patient into the prone position with the breasts uncovered, a maneuver that can make the patient feel very vulnerable. Blocking the observation window with blinds or holding up a sheet to block the view affords some privacy.
At present, breast coils are "one size fits all." The two holes in the coil may be too far apart for petite women, whereas the coil may be too small for women with large breasts (Fig 3). Ideally, breast coils of various sizes will be made available, so that breast size can be matched to coil size. This option would improve both patient comfort and image signal-to-noise ratios (SNRs) (3).
View larger version (113K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3. Difficulty of positioning in a woman with large breasts. Sagittal T1-weighted MR image (26/6) of the right breast demonstrates that positioning in the breast coil can be difficult for women with large breasts, which tend to overfill the coil. Deformities are seen where the coil support touches the breast (arrowheads), and signal intensity changes are seen where breast tissue is in proximity to coil elements (arrows).
|
|
Motion
Motion may affect an entire study or only a few images. If patient motion occurs during imaging, the entire series will be affected. However, motion due to pulsation of blood vessels may affect only a few sections (Fig 4). Motion always propagates in the phase-encoding direction, regardless of the direction of the motion. Even minimal motion will cause misregistration on subtraction images (Fig 5) that may limit their usefulness.
View larger version (94K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4. Motion artifact. Axial fat-saturated T2-weighted MR image (5210/123) reveals considerable motion blurring along the phase-encoding direction (left to right) due to vessel pulsation. This blurring was present on only a few sections in the series.
|
|
View larger version (91K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5a. Misregistration resulting from patient motion. (a) Axial fat-saturated T1-weighted MR image (4/2) obtained following the intravenous administration of gadolinium-based contrast material demonstrates a small spiculated mass (arrow). The mass proved to be invasive ductal carcinoma at core needle biopsy. (b) Subtracted image of the same lesion shows good detail. (c) Axial fat-saturated T1-weighted MR image (4/2) obtained in a different patient following the intravenous administration of gadolinium-based contrast material demonstrates a small, lobular enhancing mass with internal nonenhancing septa (arrow), findings that are consistent with a fibroadenoma. (d) On a subtracted image, the lesion (arrow) is poorly delineated due to misregistration from slight patient motion between pre- and postcontrast sequences (cf c). It is not apparent on this image that the lesion is a fibroadenoma.
|
|
View larger version (91K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5b. Misregistration resulting from patient motion. (a) Axial fat-saturated T1-weighted MR image (4/2) obtained following the intravenous administration of gadolinium-based contrast material demonstrates a small spiculated mass (arrow). The mass proved to be invasive ductal carcinoma at core needle biopsy. (b) Subtracted image of the same lesion shows good detail. (c) Axial fat-saturated T1-weighted MR image (4/2) obtained in a different patient following the intravenous administration of gadolinium-based contrast material demonstrates a small, lobular enhancing mass with internal nonenhancing septa (arrow), findings that are consistent with a fibroadenoma. (d) On a subtracted image, the lesion (arrow) is poorly delineated due to misregistration from slight patient motion between pre- and postcontrast sequences (cf c). It is not apparent on this image that the lesion is a fibroadenoma.
|
|
View larger version (119K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5c. Misregistration resulting from patient motion. (a) Axial fat-saturated T1-weighted MR image (4/2) obtained following the intravenous administration of gadolinium-based contrast material demonstrates a small spiculated mass (arrow). The mass proved to be invasive ductal carcinoma at core needle biopsy. (b) Subtracted image of the same lesion shows good detail. (c) Axial fat-saturated T1-weighted MR image (4/2) obtained in a different patient following the intravenous administration of gadolinium-based contrast material demonstrates a small, lobular enhancing mass with internal nonenhancing septa (arrow), findings that are consistent with a fibroadenoma. (d) On a subtracted image, the lesion (arrow) is poorly delineated due to misregistration from slight patient motion between pre- and postcontrast sequences (cf c). It is not apparent on this image that the lesion is a fibroadenoma.
|
|
View larger version (89K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5d. Misregistration resulting from patient motion. (a) Axial fat-saturated T1-weighted MR image (4/2) obtained following the intravenous administration of gadolinium-based contrast material demonstrates a small spiculated mass (arrow). The mass proved to be invasive ductal carcinoma at core needle biopsy. (b) Subtracted image of the same lesion shows good detail. (c) Axial fat-saturated T1-weighted MR image (4/2) obtained in a different patient following the intravenous administration of gadolinium-based contrast material demonstrates a small, lobular enhancing mass with internal nonenhancing septa (arrow), findings that are consistent with a fibroadenoma. (d) On a subtracted image, the lesion (arrow) is poorly delineated due to misregistration from slight patient motion between pre- and postcontrast sequences (cf c). It is not apparent on this image that the lesion is a fibroadenoma.
|
|
To decrease patient motion, technologists should counsel the patient regarding the importance of holding still during the examination. Sedation for patients with claustrophobia or for pain management may be necessary in some cases to increase patient comfort. Some patients may become uncomfortable and fidgety due to positioning of the arms above the head. Ensuring that the patient is comfortable prior to starting the study by propping up the head and arms with pillows may result in less patient motion.
Metallic Artifact
Ferromagnetic metals (iron, nickel, cobalt) cause severe inhomogeneity in the magnetic field. Metallic objects, including biopsy clips, jewelry, snaps on clothing, or imaging equipment on or near the patient can disturb the main magnetic field, resulting in metallic artifacts. Even nonferromagnetic metals, such as titanium, cause some artifact, resulting in a local signal intensity void in the vicinity of the metal, often with a surrounding area of high signal intensity and image distortion. The degree of distortion is determined by the shape as well as the composition of the metallic object (9). Surgical clips and some percutaneous biopsy clips may cause considerable distortion (Fig 6). If extensive, the artifact could preclude diagnosis of local recurrence at the lumpectomy site. Conversely, a small signal intensity void from percutaneous biopsy clips may help confirm the biopsy location (Fig 6). Metallic susceptibility artifact can be seen at MR imaging, even when no metal is perceptible at mammography, due to the deposition of tiny metal fragments from core needle biopsy or from electrocautery devices used during surgery (Fig 7) (10,11).
View larger version (68K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6a. Metallic artifact in a woman who had undergone prior lumpectomy of the right breast. (a) Axial fat-saturated GRE T2-weighted MR image (5210/123) shows surgical clips at the lumpectomy site creating a local signal intensity void and distortion (arrow). Detection of a small local recurrence may be difficult due to the artifact. (b) Axial fat-saturated T1-weighted MR image (5/1) obtained after intravenous contrast material administration demonstrates an irregular enhancing mass with a small signal intensity void centrally (arrow). Invasive ductal carcinoma had recently been diagnosed at this site with ultrasonography (US)–guided core needle biopsy. The signal intensity void is due to a titanium clip that was placed within the cancer after biopsy. (c) Axial fat-saturated T1-weighted MR image (4/2) of the left breast shows a large signal intensity void (arrow) with distortion (arrowhead) in the lateral breast due to prior percutaneous placement of a stainless steel clip after US-guided core needle biopsy of a benign mass.
|
|
View larger version (128K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6b. Metallic artifact in a woman who had undergone prior lumpectomy of the right breast. (a) Axial fat-saturated GRE T2-weighted MR image (5210/123) shows surgical clips at the lumpectomy site creating a local signal intensity void and distortion (arrow). Detection of a small local recurrence may be difficult due to the artifact. (b) Axial fat-saturated T1-weighted MR image (5/1) obtained after intravenous contrast material administration demonstrates an irregular enhancing mass with a small signal intensity void centrally (arrow). Invasive ductal carcinoma had recently been diagnosed at this site with ultrasonography (US)–guided core needle biopsy. The signal intensity void is due to a titanium clip that was placed within the cancer after biopsy. (c) Axial fat-saturated T1-weighted MR image (4/2) of the left breast shows a large signal intensity void (arrow) with distortion (arrowhead) in the lateral breast due to prior percutaneous placement of a stainless steel clip after US-guided core needle biopsy of a benign mass.
|
|
View larger version (132K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6c. Metallic artifact in a woman who had undergone prior lumpectomy of the right breast. (a) Axial fat-saturated GRE T2-weighted MR image (5210/123) shows surgical clips at the lumpectomy site creating a local signal intensity void and distortion (arrow). Detection of a small local recurrence may be difficult due to the artifact. (b) Axial fat-saturated T1-weighted MR image (5/1) obtained after intravenous contrast material administration demonstrates an irregular enhancing mass with a small signal intensity void centrally (arrow). Invasive ductal carcinoma had recently been diagnosed at this site with ultrasonography (US)–guided core needle biopsy. The signal intensity void is due to a titanium clip that was placed within the cancer after biopsy. (c) Axial fat-saturated T1-weighted MR image (4/2) of the left breast shows a large signal intensity void (arrow) with distortion (arrowhead) in the lateral breast due to prior percutaneous placement of a stainless steel clip after US-guided core needle biopsy of a benign mass.
|
|
View larger version (125K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7. Metallic artifact in a 46-year-old woman who had undergone breast reduction surgery 3 years earlier. Sagittal T1-weighted MR image (26/6) of the right breast shows multiple areas of magnetic susceptibility (arrows) in postoperative areas, including the skin of the inferior breast, inferior breast tissue, and the areola. Mammography did not show any evidence of metal fragments in the breast.
|
|
If possible, all metallic objects should be removed from the patient and the immediate environment. Metallic markers placed during core needle biopsy are available in titanium and will cause less distortion than markers made of stainless steel. Magnet susceptibility is greater at higher field strengths. Thus, use of titanium markers may become more important as use of 3-T magnets for breast MR imaging becomes more common.
|
Technical Factors
|
|---|
Technical factors, including selection of the FOV, coil type, fat suppression technique, and phase-encoding direction, can influence image quality and create unwelcome artifacts. Phase wrap, chemical shift, radiofrequency (RF) interference, and inadequate shimming of the magnet can also result in technically suboptimal images or artifacts. Fortunately, many of these problems can be corrected once they are identified.
FOV and Anatomic Coverage
The FOV determines what is seen on an MR image by defining the size of the anatomic region displayed on the image. The FOV should be adjusted to adequately cover the area being examined while maximizing spatial resolution. A smaller FOV for a given matrix size allows smaller pixels, thereby yielding improved in-plane spatial resolution. For axial bilateral imaging, we typically use a FOV ranging from 280 to 320 mm, with a typical matrix of 512 x 256.
The FOV divided by matrix size determines pixel size. Thus, an oversized FOV increases pixel size and includes extraneous organs in the examination (Fig 8a). Conversely, an undersized FOV will reduce signal and may result in inadequate coverage of breast tissue (Fig 8b). In addition, wraparound artifact can occur when the FOV is too small. In some instances, the FOV must be larger than desired. A common reason for use of an oversized FOV is the inability of the patient to place the arms over the head due to shoulder pain from arthritis, prior injury, or prior surgery. In such cases, the FOV must be enlarged to avoid phase wrap artifact from the arms. Optimally, the FOV should be limited to the area of the breasts while ensuring adequate coverage of the area of interest. If image wrap occurs (usually in the phase-encoding direction), the FOV may need to be enlarged.
View larger version (100K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 8a. Incorrectly sized FOV. (a) Axial T1-weighted MR image (4/2) of both breasts shows the results of using an oversized FOV. (b) Axial T1-weighted MR image (15/4) of the left breast shows the results of using an undersized FOV. The lateral portion of the breast is not visualized, and there is phase wrap artifact from the opposite breast overlying the area of interest (arrows).
|
|
View larger version (132K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 8b. Incorrectly sized FOV. (a) Axial T1-weighted MR image (4/2) of both breasts shows the results of using an oversized FOV. (b) Axial T1-weighted MR image (15/4) of the left breast shows the results of using an undersized FOV. The lateral portion of the breast is not visualized, and there is phase wrap artifact from the opposite breast overlying the area of interest (arrows).
|
|
Anatomic coverage can also be too limited or too broad in the section-select direction. Inadequate coverage resulting from too few sections may lead to incomplete examination of the breast tissue (Fig 9). The FOV should extend from about the level of the clavicle to just inferior to the inframammary fold. This will include the middle and lower axilla for evaluation of axillary adenopathy.
View larger version (85K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 9a. Undersized FOV in the section-select direction. (a) Axial T1-weighted MR image (4/2) representing the first section of the imaging sequence shows fibroglandular breast tissue (arrow), a finding that indicates that the superior portion of the breast was not included in the sequence. (b) On an axial T1-weighted MR image (4/2) from a repeat sequence that was obtained more superior on the chest, no fibroglandular tissue is visualized, a finding that indicates that the entire breast mound is included.
|
|
View larger version (78K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 9b. Undersized FOV in the section-select direction. (a) Axial T1-weighted MR image (4/2) representing the first section of the imaging sequence shows fibroglandular breast tissue (arrow), a finding that indicates that the superior portion of the breast was not included in the sequence. (b) On an axial T1-weighted MR image (4/2) from a repeat sequence that was obtained more superior on the chest, no fibroglandular tissue is visualized, a finding that indicates that the entire breast mound is included.
|
|
Care must be taken to use the appropriate FOV for adequate anatomic coverage while maximizing spatial resolution. Adequate anatomic section coverage must be achieved by using the optimal combination of parameters while maintaining temporal resolution.
Use of Breast Coil
The breast coil must be plugged in and selected for use by the technologist. If the breast coil is not selected, images will be generated with use of the body coil as the receiving coil, resulting in "grainy" images due to the low SNR (Fig 10). Sequences performed with the body coil should be repeated using the breast coil, which may entail a second appointment if contrast material has already been administered. A review of initial precontrast images by the technologist can ensure that the breast coil has been selected.
View larger version (132K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 10a. Use of body coil versus breast coil. (a) Axial T2-weighted MR image (4630/123) obtained with the body coil as the receiver coil is quite noisy because the technologist inadvertently failed to select the breast coil for use. (b) Axial T2-weighted MR image (4630/123) obtained after selecting the breast coil demonstrates considerably decreased noise and improved SNR.
|
|
View larger version (103K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 10b. Use of body coil versus breast coil. (a) Axial T2-weighted MR image (4630/123) obtained with the body coil as the receiver coil is quite noisy because the technologist inadvertently failed to select the breast coil for use. (b) Axial T2-weighted MR image (4630/123) obtained after selecting the breast coil demonstrates considerably decreased noise and improved SNR.
|
|
Fat Saturation
The high signal of fat interferes with the detection of enhancing lesions. Fat suppression is important for the detection of breast cancer. The chemical shift of water and the methyl group, which is the main component of fat, is 3.5 ppm. Under normal conditions, the MR imaging unit software automatically identifies the water peak as the highest signal peak and fat is suppressed by applying saturation pulses at a frequency of 3.5 ppm (224 Hz at 1.5 T) below the water peak. With breasts composed primarily of fat, however, the imaging unit sometimes incorrectly "zeroes in" on the fat peak, misidentifying it as the water peak, and applies fat-suppression pulses at the incorrect frequency. This phenomenon results in fat being uniformly bright, just when good fat suppression is needed the most. At our institution, fat suppression is carefully reviewed and, if necessary, the center frequency is adjusted by the technologist to ensure uniform fat suppression.
Difficulties with lack of fat saturation are more common in breasts with a high percentage of fat or when breast implants are present (Fig 11). In such cases, the contrast-enhanced sequences must be repeated; otherwise, enhancing masses may not be identified. If fat saturation is difficult, the patient must be instructed to remain very still during the examination so that the subtraction images will be interpretable.
View larger version (62K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 11. Lack of fat saturation in a patient with silicone breast implants. Axial T1-weighted MR image (28/5) demonstrates no suppression of the fat signal intensity, leaving fatty tissues bright (arrows).
|
|
Inhomogeneous fat saturation is a common problem and may be difficult to correct (Fig 12a). Inhomogeneous fat saturation can be due to field inhomogeneity, which may be improved by shimming the magnet. Shimming of the static magnetic field is achieved with additional smaller current-carrying coils designed specifically to correct magnetic field inhomogeneities in each direction (x, y, and z). Poor shimming results in distorted images and compromises uniform fat saturation. The magnet can be reshimmed by a service engineer. The technologist can also shim the magnet at the console by adjusting the shim fields using the software on the MR imaging unit with the patient in position, so that shimming is customized to the patient being imaged. However, this feature is available only on some MR imaging units (eg, Siemens). More frequent shimming may be needed if magnetic field homogeneity is being affected by conditions outside the imaging room, such as repositioning of nearby construction equipment or materials. Shimming can be a particular problem in mobile MR imaging units.
View larger version (107K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 12a. Inhomogeneous fat saturation. (a) Axial T1-weighted MR image (4/2) shows inhomogeneous fat saturation (arrow). This finding is more common in breasts that are predominantly fatty. (b) Sagittal T1-weighted MR image (26/6) of the right breast obtained in a different patient shows signal flare (arrow) where the breast is near a breast coil element.
|
|
View larger version (106K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 12b. Inhomogeneous fat saturation. (a) Axial T1-weighted MR image (4/2) shows inhomogeneous fat saturation (arrow). This finding is more common in breasts that are predominantly fatty. (b) Sagittal T1-weighted MR image (26/6) of the right breast obtained in a different patient shows signal flare (arrow) where the breast is near a breast coil element.
|
|
An artifact with the appearance of nonuniform fat suppression can occur when some breast tissue is very close to coil elements. This artifact is not related to fat suppression; it is known as signal flaring and can be exacerbated by poor fat suppression (Fig 12b). Placing additional padding between the breast tissue and coil elements can correct some signal nonuniformities caused by signal flaring.
Most breast MR imaging is performed with 1.5-T magnets. Uniform fat suppression is more difficult to achieve at higher field strengths (such as 3 T) because of the difficulty in maintaining a uniform magnetic field. This problem may be a barrier to the use of 3-T magnets for breast MR imaging, since uniform fat saturation is important for cancer detection. However, the separation of the signal peaks for fat and water is greater at 3 T (440 Hz); thus, correct identification of the fat peak is easier with higher-field-strength magnets.
Review of the location of water and fat peaks and of the center frequency that is automatically selected by the system software may improve fat saturation. Shimming improves magnetic field homogeneity and may result in improved uniformity of fat saturation. Development of different-sized breast coils to accommodate both larger and smaller breasts could reduce focal field inhomogeneity and improve image SNR.
Phase Encoding
Artifacts from blood flow as well as cardiac, respiratory, and patient motion all propagate in the phase-encoding direction, regardless of the direction of the motion. Random motion results in blurring of moving tissues but also causes a structured noise pattern, resulting in "ghosting" of brighter moving tissues in the phase-encoding direction (12). For breast MR imaging, the correct choice of the phase-encoding direction is left to right for axial images and superior to inferior for sagittal images. The smearing from respiratory or cardiac motion then occurs across the lateral chest rather than the breasts. If the phase-encoding direction is incorrectly chosen to be anterior to posterior with either axial or sagittal sequences, smearing from cardiac or respiratory motion can obscure large amounts of breast tissue (Fig 13).
View larger version (121K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 13. Smearing from incorrect choice of phase-encoding direction. Sagittal T2-weighted MR image (4267/106) of the left breast shows smearing of the image at the lower portion of the breast (arrows), which resulted from phase encoding in the anteroposterior direction. Note that this image was part of a nondiagnostic study and that fat suppression was not used.
|
|
Periodic motion, such as vascular pulsation, results in ghosting occurring at regular intervals in the phase-encoding direction (Fig 14). Ghosting is a form of phase-encoding artifact (13). Reducing the repetition time may help reduce ghosting so that it is not propagated throughout the entire image, but doing so is often impractical in breast MR imaging.
View larger version (154K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 14. Ghosting from vascular pulsation. Sagittal fat-suppressed T2-weighted MR image (9128/60) of the right breast obtained in a patient with a silicone implant (arrow) shows periodic bright artifacts (arrowheads) due to ghosting of high-signal-intensity blood in a pulsating subclavian vein.
|
|
Even when phase encoding is performed in the correct direction, pseudolesions can be created (Fig 15). If a potential lesion is detected that may be due to image wrap in the phase-encoding direction, it is often helpful to review all postcontrast and subtracted series to see if the enhancing area appears in a consistent location on all series. This review will typically resolve the issue. In the rare instance in which review does not resolve the status of the lesion, a contrast-enhanced series can be repeated with the phase- and frequency-encoding directions transposed. If the lesion persists, it is not due to phase-encoding ghosting artifact.
View larger version (101K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 15. Ghosting from cardiac motion in a 51-year-old woman who had undergone prior left mastectomy and right lumpectomy. On an axial T1-weighted MR image (4/2), one of the right serratus anterior muscles (arrow) has higher signal intensity than the surrounding muscle tissue. This hyperintensity could be mistaken for possible involvement with cancer but is actually due to ghosting along the phase-encoding direction from cardiac motion.
|
|
Wraparound
Wraparound artifact occurs when not all of the signal-producing tissue is within the FOV (12). The signal from excited tissue outside the FOV becomes superimposed on structures within the FOV through misregistration during Fourier transform reconstruction (Fig 16). Wraparound is also known as aliasing or phase wrap because it occurs primarily in the phase-encoding direction. Wraparound of normal structures or of the coil itself can occur (Fig 17).
View larger version (66K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 17a. Wraparound artifact due to a breast coil. (a) Axial fat-suppressed T1-weighted MR image (4/2) shows medium-signal-intensity artifact (single arrows) on the upper part of the image. Because the patient is prone, this artifact represents part of the breast coil. This phenomenon can occur with collected moisture in the coil structure. Dotted line and double arrow indicate the center partition of the coil. (b) Axial fat-suppressed T1-weighted MR image (4/2) obtained in a different patient shows the same type of artifact (single arrows) overlying the breast tissue due to larger breast size. The artifact is the same distance from the center partition of the coil (dotted line and double arrow) as it is in a, and will, in fact, occur at the same distance from the center partition in all cases.
|
|
View larger version (95K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 17b. Wraparound artifact due to a breast coil. (a) Axial fat-suppressed T1-weighted MR image (4/2) shows medium-signal-intensity artifact (single arrows) on the upper part of the image. Because the patient is prone, this artifact represents part of the breast coil. This phenomenon can occur with collected moisture in the coil structure. Dotted line and double arrow indicate the center partition of the coil. (b) Axial fat-suppressed T1-weighted MR image (4/2) obtained in a different patient shows the same type of artifact (single arrows) overlying the breast tissue due to larger breast size. The artifact is the same distance from the center partition of the coil (dotted line and double arrow) as it is in a, and will, in fact, occur at the same distance from the center partition in all cases.
|
|
The sequence used for the dynamic portion of the MR imaging examination is a 3D (volume) sequence. In 3D imaging, there is a second phase-encoding direction—the section-select direction—and aliasing can occur in this direction if the coverage is not adequate. This aliasing can be reduced by the addition of section oversampling or by increasing the number of sections. Both techniques add signal, but they also increase imaging time and can reduce the temporal resolution of the dynamic portion of the examination. Three-dimensional wraparound appears as ghosting of other sections into the section of interest and is more likely to occur near the end sections of the 3D volume (Fig 18). Wraparound artifact can be reduced by increasing the number of sampling points in the phase-encoding direction or by enlarging the FOV.
View larger version (165K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 18. Three-dimensional phase wrap artifact. Sagittal T2-weighted MR image (4/2) of the left breast obtained with a 3D volume sequence demonstrates 3D phase wrap artifact (arrows) in the section-select direction. The artifact appears as a ghostlike outline of signal-producing breast tissues.
|
|
Zebra Artifact
Zebra artifact, also known as moiré artifact, is a type of phase interference that occurs when signal-producing tissue outside the selected FOV wraps into the FOV and when poor magnet shimming causes a phase shift between the tissue within and that outside the selected FOV. The resultant black-white banding is due to rapid phase shifting as a function of position, causing signals from the two regions to be out of phase and therefore to cancel each other, yielding a black band; and to be in phase and therefore to combine, yielding a white band (Fig 19). With breast MR imaging, this banding can occur when the breast coil is not selected and the body coil is used. If zebra artifact is observed on images, the affected sequences should be repeated after taking measures to eliminate the possibility of image wrap or phase shift. Image wrap can be minimized by either (a) enlarging the selected FOV to include all signal-producing tissues; or (b) applying phase oversampling ("no phase wrap" being one version), which allows the acquisition of more phase samples than are represented on the reconstructed image. Reshimming of the magnet can also reduce phase shift.
View larger version (122K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 19. Zebra artifact. Axial T1-weighted MR image (5/2) obtained with a body coil shows concentric rings of high and low signal intensity (arrowhead), a finding that is consistent with zebra (moiré) artifact. Poor shimming caused inhomogeneous fat saturation; the fat appears dark on the lateral portions of the right breast and bright on the left breast (open arrow). Image wrap (solid arrows) occurred in the phase-encoding direction because signal-producing tissues were excluded from the selected FOV. The zebra artifact is the result of a phase shift across the image, likely due in this case to both poor shimming and image wrap of signal-producing tissues. The phase difference between tissues included in the selected FOV and tissues wrapping into the image cause the light-dark banding.
|
|
RF Interference
RF interference occurs when there is a source of RF signal in the imaging room or there is a leak in the RF shield surrounding the imaging unit so that RF signals penetrate the shield (Figs 20, 21). Sources of RF signal in the imaging room include radios or televisions, faulty fluorescent lights, and electronic monitoring equipment. Imaging rooms have an RF shield that completely surrounds the imaging unit. Anything that might cause a leak in the RF shield, such as not completely closing the imaging room door or construction changes to the shield, can permit RF signals to penetrate the imaging room and be picked up by the RF receiver coils during imaging. The result can be a noisier overall background for MR images or RF bands appearing at discrete frequencies on the images. The latter appear as bright-dark bands at a fixed location along the frequency-encoding direction but propagating across the image in the phase-encoding direction.
View larger version (84K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 20. RF interference artifact. (20) Axial T2-weighted MR image (5210/123) demonstrates multiple high-signal-intensity bands (arrows) in the phase-encoding direction due to external RF noise. The door to the imaging room had not been completely shut, allowing external RF noise to leak into the room during imaging.
|
|
View larger version (88K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 21a. (21a) Axial T1-weighted MR image (4/2) demonstrates a linear high-signal-intensity band overlying the right breast (arrows) due to RF noise occurring at a particular frequency within the bandwidth of the acquired image. The RF artifact propagates in the phase-encoding direction, which is left to right on this image. (21b) Photograph shows damage to the copper seal (arrows) on the door to the imaging room. This damage proved to be the cause of the RF leak.
|
|
View larger version (150K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 21b. (21a) Axial T1-weighted MR image (4/2) demonstrates a linear high-signal-intensity band overlying the right breast (arrows) due to RF noise occurring at a particular frequency within the bandwidth of the acquired image. The RF artifact propagates in the phase-encoding direction, which is left to right on this image. (21b) Photograph shows damage to the copper seal (arrows) on the door to the imaging room. This damage proved to be the cause of the RF leak.
|
|
With older MR imaging units, RF feedthrough from the transmit coils to the receiver coils sometimes manifested as zipper artifact (Fig 22). This artifact occurred in the center of the image (in the frequency-encoding direction) and propagated as alternating light and dark bands along the phase-encoding direction. The key to recognizing this artifact is knowing that it occurs in the center of the image. Zipper artifact is uncommon with newer MR imaging units, which make use of a frequency offset that does not place the center RF frequency at the center of the image.
View larger version (99K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 22. Zipper artifact. Axial fat-suppressed T2-weighted MR image (9000/67) shows a band of small high-signal-intensity foci (arrowheads). The band was located in the center of the original image (measuring from top to bottom), a finding that is characteristic of zipper artifact. The high-signal-intensity area in the left breast (arrow) represents a silicone implant.
|
|
RF interference can be minimized by paying close attention to background noise in the image. If a high level of image noise persists, the RF shield should be checked. If increased RF interference is suspected to be due to external causes such as ongoing construction or new equipment placed in or near the MR imaging suite, the source should be determined by systematically imaging a phantom with the device in question turned on and turned off. Ongoing quality control of the MR imaging unit with a standard phantom (eg, the American College of Radiology [ACR] MRI Accreditation phantom) and a standard imaging protocol will help identify changes in system SNR. It should be noted that the ACR MRI Accreditation phantom will not fit in breast coils; consequently, system quality control must be achieved with the head or body coil. Part of the ACR Breast MRI Accreditation Program, which is under development, is the creation of a suitable phantom for evaluating the performance of breast MR imaging coils and pulse sequences.
Chemical Shift
Chemical shift artifacts occur at fat-fluid interfaces due to differences in the resonant frequency of hydrogen in fat and hydrogen in water (224 Hz at 1.5 T). Spatial misregistration between the "fat image" and the "water image" occurs in the frequency-encoding direction due to the 224-Hz frequency shift between the two tissues. The number of pixels over which the chemical shift occurs depends solely on the bandwidth per pixel of the imaging sequence. For example, if the bandwidth per pixel is 22.4 Hz, the chemical shift will be 10 pixels (10 x 22.4 Hz per pixel = 224 Hz). This shift results in a bright or dark band perpendicular to the frequency-encoding direction where fat and water are adjacent to one another (Fig 23) (14). This kind of chemical shift artifact occurs on every MR image on which water and fat interfaces occur, which includes all breast images. Chemical shift artifact can be reduced by increasing the bandwidth per pixel of the imaging sequence.
View larger version (147K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 23a. Chemical shift artifact. GRE MR imaging was performed to demonstrate the effects of varying the echo time and bandwidth on chemical shift artifacts. (a) On a sagittal GRE MR image (200/7) of the right breast obtained at a typical intermediate bandwidth of 81.4 Hz/pixel, the bandwidth produces a water-fat displacement of 2.8 pixels in the horizontal (anteroposterior) direction. The echo time (7 mm) is such that, in GRE imaging, water and fat are out of phase, producing a black border at all water-fat interfaces. (b) Sagittal GRE MR image (200/7) obtained at a lower bandwidth (23.3 Hz/pixel) shows a water-fat displacement of 9.6 pixels. (c) On a sagittal GRE MR image (200/9) obtained at the same intermediate bandwidth as a, the pixel shift between water and fat is 2.8 pixels left to right. At this echo time (9 mm), water and fat are nearly in phase. The displacement of fat relative to water produces a bright band where water and fat overlap and a dark band where water and fat are shifted apart. (d) On a sagittal GRE MR image (200/9) obtained at a lower bandwidth of 15.8 Hz/pixel, fat and water are again nearly in phase, but the lower bandwidth causes a fat-water displacement of 14.1 pixels, creating wider bright bands where fat and water overlap and wider dark bands where fat and water are shifted apart (cf c).
|
|
View larger version (142K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 23b. Chemical shift artifact. GRE MR imaging was performed to demonstrate the effects of varying the echo time and bandwidth on chemical shift artifacts. (a) On a sagittal GRE MR image (200/7) of the right breast obtained at a typical intermediate bandwidth of 81.4 Hz/pixel, the bandwidth produces a water-fat displacement of 2.8 pixels in the horizontal (anteroposterior) direction. The echo time (7 mm) is such that, in GRE imaging, water and fat are out of phase, producing a black border at all water-fat interfaces. (b) Sagittal GRE MR image (200/7) obtained at a lower bandwidth (23.3 Hz/pixel) shows a water-fat displacement of 9.6 pixels. (c) On a sagittal GRE MR image (200/9) obtained at the same intermediate bandwidth as a, the pixel shift between water and fat is 2.8 pixels left to right. At this echo time (9 mm), water and fat are nearly in phase. The displacement of fat relative to water produces a bright band where water and fat overlap and a dark band where water and fat are shifted apart. (d) On a sagittal GRE MR image (200/9) obtained at a lower bandwidth of 15.8 Hz/pixel, fat and water are again nearly in phase, but the lower bandwidth causes a fat-water displacement of 14.1 pixels, creating wider bright bands where fat and water overlap and wider dark bands where fat and water are shifted apart (cf c).
|
|
View larger version (103K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 23c. Chemical shift artifact. GRE MR imaging was performed to demonstrate the effects of varying the echo time and bandwidth on chemical shift artifacts. (a) On a sagittal GRE MR image (200/7) of the right breast obtained at a typical intermediate bandwidth of 81.4 Hz/pixel, the bandwidth produces a water-fat displacement of 2.8 pixels in the horizontal (anteroposterior) direction. The echo time (7 mm) is such that, in GRE imaging, water and fat are out of phase, producing a black border at all water-fat interfaces. (b) Sagittal GRE MR image (200/7) obtained at a lower bandwidth (23.3 Hz/pixel) shows a water-fat displacement of 9.6 pixels. (c) On a sagittal GRE MR image (200/9) obtained at the same intermediate bandwidth as a, the pixel shift between water and fat is 2.8 pixels left to right. At this echo time (9 mm), water and fat are nearly in phase. The displacement of fat relative to water produces a bright band where water and fat overlap and a dark band where water and fat are shifted apart. (d) On a sagittal GRE MR image (200/9) obtained at a lower bandwidth of 15.8 Hz/pixel, fat and water are again nearly in phase, but the lower bandwidth causes a fat-water displacement of 14.1 pixels, creating wider bright bands where fat and water overlap and wider dark bands where fat and water are shifted apart (cf c).
|
|
View larger version (111K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 23d. Chemical shift artifact. GRE MR imaging was performed to demonstrate the effects of varying the echo time and bandwidth on chemical shift artifacts. (a) On a sagittal GRE MR image (200/7) of the right breast obtained at a typical intermediate bandwidth of 81.4 Hz/pixel, the bandwidth produces a water-fat displacement of 2.8 pixels in the horizontal (anteroposterior) direction. The echo time (7 mm) is such that, in GRE imaging, water and fat are out of phase, producing a black border at all water-fat interfaces. (b) Sagittal GRE MR image (200/7) obtained at a lower bandwidth (23.3 Hz/pixel) shows a water-fat displacement of 9.6 pixels. (c) On a sagittal GRE MR image (200/9) obtained at the same intermediate bandwidth as a, the pixel shift between water and fat is 2.8 pixels left to right. At this echo time (9 mm), water and fat are nearly in phase. The displacement of fat relative to water produces a bright band where water and fat overlap and a dark band where water and f | |
Top
Abstract
LEARNING OBJECTIVES FOR TEST...
