| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 All authors: Department of Radiology, University of Michigan Health System, B1 D520 UH, 1500 E Medical Center Dr., Ann Arbor, MI 48109-0030.
Received July 20, 2005;
accepted after revision April 24, 2006.
Address correspondence to C. E. Blane
(cblane{at}umich.edu).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Breast phantoms were created using gelatin doped with gadolinium. Four commercially available tissue marker clips were evaluated. Clinical MR evaluation of all phantoms with 1.5-T gradient-recalled echo sequences was performed. Images were evaluated for size and character of the visible artifacts and graphically appreciable fat saturation inhomogeneities. Quantitative measurement of the local inhomogeneity in 3D parts per million maps was obtained as a function of distance from each tissue marker.
RESULTS. All tissue marker clips caused signal void artifacts on non-fat-suppressed images that measured 2-6 times the clip diameter. The degree of fat suppression inhomogeneity was minor but clinically appreciable. The local clip-induced field inhomogeneity varied from 0.25 to greater than 4.0 PPM for the four clips. At 0.25 PPM, the zonal diameter of frequency shift varied from 6 mm to 44 mm.
CONCLUSION. Artifacts caused by tissue marker clips could limit the sensitivity of MRI for detection and follow-up of breast cancer. The local effects on field inhomogeneity will affect local fat suppression and make spectroscopy data less reliable. These effects, though small, are measurable and vary among the different clips evaluated.
Keywords: artifact • breast • breast cancer • MRI • tissue marker clips
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Imaging-guided biopsy of nonpalpable breast abnormalities has become an indispensable part of routine breast care. When a nonpalpable breast abnormality is detected, imaging assistance for biopsy guidance is required. Often, imaging-guided core needle biopsy is used to sample abnormalities detected on mammography or sonography. Biopsy may remove enough of the tissue that the lesion is no longer detectable on imaging. Therefore, it has become standard practice to place a tissue marker clip at the site of percutaneous core needle biopsy for future definitive surgical reference for cases in which the lesion proves to be malignant or atypical and excision is necessary [2-5, 7-10]. Because the radiologist does not know the pathologic outcome at time of core needle biopsy, clips will be left after benign and malignant biopsies.
With the introduction of neoadjuvant chemotherapy, larger masses may decrease in size so that they are no longer evident clinically or on imaging studies before surgical resection. To ensure that the original site of cancer can be localized with imaging for lumpectomy after the neoadjuvant chemotherapy, marker clips are placed at the center of the malignancy. Because breast MRI may be superior to mammography for identification of malignancy and accuracy in prediction of the extent of disease, some clinicians have used MRI for evaluation of the tumor response with neoadjuvant chemotherapy [11].
Because MRI may have a role in the evaluation and management of breast cancer patients, compatibility of marker clips is an important consideration. Shellock [12] has previously shown clips to be safe without risk of movement or dislodgement when undergoing MRI. The Shellock study has also shown a lack of substantial heating, even under excessive radiofrequency exposure, further documenting the safety of the use of metallic marker clips with breast MRI [12]. The commonly used nonferromagnetic or weakly ferromagnetic metal used in tissue marker clips may create an MR artifact. The artifacts are dependent on the magnetic susceptibility of the material; the quality, shape, orientation, and position of the implant; and the parameters used for imaging [12].
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Four commonly used commercially available tissue marker clips were deployed into the phantoms by a radiologist using a standard technique with the manufacturer's introducer [13-16]. The devices studied were MammoMark Biopsy Site Marker by Artemis Medical, clip A; Gel Mark UltraCor by SenoRx, Inc., clip B; MicorMark II by Ethicon Endo-Surgery, clip C; and UltraClip Tissue Marker by INRAD, clip D. The clips were obtained in November 2003 in the original manufacturers' packaging. The fifth sample contained a 2.5-mm steel filament fashioned from a section of a standard metallic office staple made of nonnickel steel wire composed mainly of iron and carbon that was deployed into a phantom using an 18-gauge spinal needle, clip E. Clip E was studied to show artifact that would be caused by a clip of similar size if it were manufactured out of standard ferromagnetic material. This also simulates the possible artifact produced by a retained foreign body. A sixth phantom was poured from the same batch of gelatin and was the same size and shape as the other phantoms. No tissue marker clip was deployed into the control phantom.
The greatest diameter of the clip in its deployed state in the phantom was measured on the digital mammographic image using a standard digital measurement technique. All clips measured between 2 and 3 mm in greatest diameter. The steel filament also measured between 2 and 3 mm in length.
Mammography
Two-view mammograms of the phantoms were obtained (GE DMR, GE Healthcare)
after clip deployment to ensure correct placement, similar to postbiopsy
mammograms for clip placement after core needle biopsy
(Fig. 1).
MRI
Scans were obtained on a 1.5-T clinical MRI scanner (Signa LX, GE
Healthcare) using a quadrature transmit-receive head coil. The five phantoms
were scanned individually but positioned identically in the head coil such
that the control phantom could be used as a phase reference. Each phantom was
scanned with a threeplane locator and four 3D spoiled gradient-echo (SPGR)
series. Two series used parameters commonly used in dynamic contrast-enhanced
MRI: 3D SPGR; TR/TE, 8.0/3.3; flip angle, 30°; field of view, 20 cm;
bandwidth, 31 kHz; slice thickness, 3.5 mm; 32 slices; matrix, 256 x
256; and number of excitations (NEX), 1 with fat suppression (series 1) and
without fat suppression (series 2). Field inhomogeneity attributable to the
clip inclusions was quantified via field-mapping. Settings for these two
series differed only in TE (TE, 7 in series 3 vs TE, 9 in series 4) with other
parameters fixed: 3D SPGR; TR, 22; flip angle, 20°; field of view, 20 cm;
bandwidth, 15.6 kHz; slice thickness, 3 mm; 32 slices; matrix, 256 x
256, and NEX, 1. All 3D series were acquired in the coronal plane to
illustrate artifacts parallel and perpendicular to the external magnetic field
(B0) in one image.
Processing Methods
Two types of artifacts resultant to metallic breast marker clips were
studied. The first was signal loss due to signal dephasing on gradient-echo
images. The second was field inhomogeneity due to paramagnetic or
susceptibility properties of the clip that affect fat suppression performance
(i.e., water signal erroneously suppressed or fat signal unsuppressed) or the
quality of localized spectroscopy. The first of these was measured directly
from series 2 images as the maximal spatial extent of the signal loss
zone.
|
where I signifies the complex-valued image with its subscripts signifying
the conditions under which the image was acquired, the asterisk signifies the
complex conjugate, and 
is conversion from phase to PPM for a
2-millisecond TE difference on a 1.5-T system. Note: the argument of the
arctan function has the inherent inhomogeneity of the magnet removed by
correction with the control phantom. Moreover, this analysis not only isolates
field disturbances of each clip but also expresses the disturbance in PPM
units, which are independent of the MRI system field strength.
The 3D PPM maps were visually inspected to select the slice that contained the clip and the greatest inhomogeneity range. The spatial extent at which the absolute value of the inhomogeneity exceeded 0.25, 0.5, 1.0, 2.0, and 3.0 PPM along vertical (i.e., parallel to B0) and horizontal (i.e., perpendicular to B0) lines through the clip was measured from the PPM maps.
The MR images were evaluated and the maximal diameter of the signal void artifact was recorded on an Advantage workstation (GE Healthcare) using standard digital measurement techniques (Fig. 1). Qualitative clinical evaluation was also performed. Three Mammography Quality Standards Act of 1992 (MQSA)-qualified, fellowship-trained breast imagers independently evaluated the paired series (with and without fat saturation) for each sample (Fig. 2). They reviewed the images on the workstation and were allowed to adjust the windows and levels of the images for optimum evaluation. Using the non-fat-saturated series as a control, the reviewers decided by consensus whether there was clinically appreciable signal inhomogeneity on the fat-saturated images. Each reviewer was blinded as to which clip was being evaluated. The reviewers were also blinded to the answers of the other reviewers. One of the authors acted as an arbitrator in a single case in which opinion was not unanimous.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
The frequency shift, both negative and positive, induced by the local field inhomogeneities was measured parallel and perpendicular to B0 (Table 2). The zones of equivalent frequency shifts were mapped, and their diameters were measured for each of the samples. All clips caused frequency shift, with clips D and E exhibiting the largest diameter zones of significant frequency shift (at least 0.25 PPM) and a diameter of 44 mm. Clip A had the smallest diameter of frequency shift, which measured 6 mm. Clip E, the ferromagnetic metallic filament, caused the largest diameter of measurable frequency shift, 120 mm.
|
The breast radiologists determined that there were no clinically appreciable additional artifacts on the fat-saturated sequences for clips A, B, and C. There were clinically important additional artifacts for clips D and E (Fig. 2).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
When evaluating breast MRI after the placement of tissue marker clips, the most obvious artifact to be considered is the single void seen on all sequences. This appears as a region of signal void with surrounding increased signal that is greater in size than the actual clip itself [17] and has been referred to as the black hole artifact [21]. No clinically useful information is obtainable within the entire volume of the signal void, which could potentially have a negative effect on lesion detection if a small enhancing lesion was present at or adjacent to the previously placed clip. This could potentially limit MR assessment of the extent of a breast cancer, the size of the lesion, and the assessment of the surgical margins. Matsuura et al. [22] have shown that the signal void artifact diameters of implanted devices of the same size and shape are greater for stainless steel than for titanium, with little difference between pure titanium and titanium alloy. Standard ferromagnetic materials cause even greater artifact [22]. Our results do not adequately quantify the inherent properties of the materials given the difference in composition, shape, and mass of the markers.
Fat-suppression techniques are considered essential for the interpretation of breast MRI. They provide images in which enhancing lesions appear bright against a dark background [1]. Fat suppression can be achieved actively with a specific fat-suppression pulse applied or passively with postprocessing image subtraction. Because of the potential for patient motion with passive techniques, active fat suppression is the procedure of choice. However, active fat suppression is highly sensitive to field inhomogeneity [1]. In theory, local field inhomogeneities caused by the tissue marker clips could preclude satisfactory fat suppression and cause the typical banding artifact [23]. This effect is observed when local frequency shifts reach a 2-3 PPM threshold, leading to inadequate suppression of fat signal or partial suppression of water signal seen as a banding artifact. This banding artifact threshold was not met by clip A. Although clips B and C marginally reached the frequency shift threshold to cause banding artifact, there was no clinically appreciable fat-saturation inhomogeneity. Clips D and E had the largest diameters of frequency shift. They also caused clinically appreciable banding artifact on the fat-saturated images that would likely limit evaluation for malignancy in vivo (Fig. 2).
MR spectroscopy for evaluating choline levels in malignant tumors is a promising field for the evaluation of breast lesions and could improve the specificity of breast MRI in differentiating benign from malignant lesions [24-26]. Local signal disturbance of even very low magnitudes can disrupt the spectroscopic data, rendering it nondiagnostic in the immediate vicinity. Acceptable spectroscopic resolution is approximately 0.1 PPM (e.g., choline and creatine resonances are separated by 0.2 PPM) [27]. A local frequency inhomogeneity of 0.25 PPM or more greatly hinders metabolite peak identification and signal-to-noise ratio. Given that tissue marker clips are usually placed directly within the zone of interest, any local frequency shift will likely compromise spectroscopic analysis.
Although signal artifact can be reduced by reducing voxel dimensions, this leads to an increase in scanning time and a decrease in signal-to-noise ratio throughout the image. In addition, inadvertent suppression of water signal caused by its proximity to the clips and the fat-suppression pulses is not remedied by altering the acquired slice orientation or image resolution.
The utility of tissue marker clips has been shown, but consideration as to the potential need for follow-up MRI should be made when placing the clips. In particular, clips should be used that meet the clinical requirements of stability and visibility on mammography and sonography but that cause the lowest possible degree of visible MR artifact. If breast MRI becomes a screening technique, the image degradation caused by metallic marking clips has the potential to prevent early cancer detection for a group of patients who have retained tissue marker clips or foreign bodies within the breast tissue. Kriege et al. [28] and Warner et al. [29] have recently shown the potential use of annual MR screening in women at high risk, including women harboring the breast cancer genes, BRCA1 and BRCA2.
MRI was found to be more sensitive than mammography but less specific, generating numerous false-positive biopsies and additional short-term examinations. Future MR screening of patients who have benign core biopsy results could be adversely affected in the immediate region around the clip due to artifacts. Small artifacts should present minimal problems because artifacts of less than 5 mm would obscure only very small, minimal breast cancer (< 1 cm). In fact, some degree of artifact is essential for localization of the marker on both diagnostic MRI and for lesion localization for MR-guided procedures. A signal void of less than 1 cm at the site of a core biopsy-proven cancer is unlikely to affect the surgical outcome regarding the extent of lumpectomy because standard lumpectomy removes the primary tumor and a surrounding rim of normal tissue that is often greater than the size of the MR artifact.
There are limitations of our study. Because we studied only a single clip from each manufacturer, we are not in a position to endorse or not endorse a particular product. Manufacturers change products for the better routinely, and the product we tested may or may not reflect the current product. For example, INRAD has two different microclips with composition variation in its product line, only one of which was included in the study. We do not know the range of artifacts produced by a series of clips from the same company. Our breast phantom is not breast tissue, but its qualities should be the same for each clip studied. We purposely blinded reviewers to provide a uniform test situation and to avoid any bias due to product recognition.
A complex interaction of factors affects the degree of quantifiable artifact. In general, clips composed of titanium or nonferromagnetic material with a smaller diameter and less mass should produce less gross image degradation. However, other factors such as visibility, safety, ease of use, and stability of the marker within the breast must, of course, also be considered when evaluating the optimum marker clip. The type and number of retained clips within a breast or the presence of metallic foreign bodies within the breast should be documented and reviewed when determining candidates for breast MRI, when prescribing the study, and when evaluating the images. Clip manufacturers are urged to continue to consider MR compatibility and the potential to degrade MR images when developing future products.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  J. A. Harvey, R. E. Hendrick, J. M. Coll, B. T. Nicholson, B. T. Burkholder, and M. A. Cohen Breast MR Imaging Artifacts: How to Recognize and Fix Them RadioGraphics, October 1, 2007; 27(suppl_1): S131 - S145. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
