| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Breast Imaging |
1 From the Department of Medical Imaging, University of Toronto, Sunnybrook and Women’s College Health Sciences Centre, 2075 Bayview Ave, MG178, Toronto, Ontario, Canada M4N 3M5. From the 2001 RSNA scientific assembly. Received August 1, 2002; revision requested September 23; revision received December 6; accepted January 15, 2003. Supported by a program project grant from the Terry Fox Foundation. Address correspondence to R.A.J. (e-mail: roberta.jong@sw.ca).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Twenty-two women scheduled for biopsy because they were suspected of having abnormalities at breast imaging underwent imaging with contrast material–enhanced digital mammography. Six sequential images of the affected breast were obtained, with a contrast agent injected intravenously between the time the first and second images were obtained. Image processing included registration and logarithmic subtraction. Lesions were evaluated for the presence, morphology, and kinetics of enhancement. Lesion type, size, and pathologic findings were correlated with the findings at contrast-enhanced digital mammography.
RESULTS: At contrast-enhanced digital mammography, enhancement was observed in eight of 10 patients with biopsy-proved cancers. In one case of ductal carcinoma in situ and one case of invasive ductal carcinoma, enhancement was not observed. No enhancement was seen in seven of 12 cases in which lesions were suspected of being malignant at initial imaging but were benign. Morphology generally correlated with the pathologic diagnosis. The kinetics of lesion enhancement showed similarity to that seen with gadolinium-enhanced magnetic resonance imaging but was not consistent.
CONCLUSION: The results of this preliminary study suggest that contrast-enhanced digital mammography potentially may be useful in identification of lesions in the mammographically dense breast. Further investigation of contrast-enhanced digital mammography as a diagnostic tool for breast cancer is warranted.
© RSNA, 2003
Index terms: Breast neoplasms, diagnosis • Breast neoplasms, radiography, 00.11 • Breast neoplasms, US, 00.1298 • Contrast media • Radiography, digital, 00.1215
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Chang et al (2) and Sibala et al (3) showed that contrast material–enhanced computed tomography (CT) could depict cancers in dense breasts, where conventional mammography was limited because of the lack of intrinsic tissue contrast. In the mid-1980s, Watt et al (4), Ackerman et al (5), and Watt et al (6) performed digital subtraction angiography (DSA) of the breast by using an x-ray image intensifier system. Benign and malignant lesions were differentiated according to the strength of enhancement.
During the past decade, considerable attention has been focused on contrast-enhanced magnetic resonance (MR) imaging of the breast with gadopentetate dimeglumine as a paramagnetic contrast agent. It has been demonstrated that the region of three-dimensional enhancement in MR imaging correlates well with histologic assessment of tumormargins (7). During the past few years, researchers in several studies have explored the use of morphology (8) or kinetics (9–12) of gadopentetate dimeglumine uptake. Either qualitative (9,10) or quantitative (11) methods can be used to aid in the differentiation of benign from malignant breast tumors. Breast MR imaging is finding application in several areas, including validation of the diagnosis (10), monitoring of neoadjuvant therapy (13), guidance for breast intervention (14), and assessment of the extent of disease for presurgical planning (15). Warner et al (16) and others (17) have also demonstrated that MR imaging can be of value in the surveillance of women for hereditary breast cancer and can be an effective screening tool in this small population of women who exhibit a very high lifetime risk. Findings of this body of work show that breast lesions can be detected at MR imaging on the basis of tumor angiogenesis by using a low-molecular-weight imaging agent capable of extravascular enhancement. As such, it is logical to probe whether a similar detection capability would be possible with diffusible radiographic contrast media used in conjunction with digital mammography.
Skarpathiotakis et al (18) previously carried out computer modelling and experimental studies to determine how to optimize the acquisition and processing of contrast-enhanced digital mammographic images and to understand the attainable concentration of contrast medium detectable with this technique. The purpose of our current clinical pilot study was to investigate the potential of using full-field digital mammography in conjunction with iodine as contrast material to facilitate the detection and the characterization of lesions in the breast.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Contrast-enhanced digital mammography was performed in 26 patients who fulfilled these criteria and who agreed to undergo the contrast-enhanced digital mammographic examination prior to their biopsy. These women had lesions that were suspected of being malignant and that were initially depicted at conventional mammography or ultrasonography (US). Three early contrast-enhanced digital mammographic studies were not technically successful. One patient did not undergo the stereotactic core-needle biopsy because of the thinness of her breast and was lost to follow-up. This article includes the remaining 22 women, 21 of whom underwent biopsies and one of whom had a cyst that was confirmed at US. Women ranged in age from 40 to 74 years. The amount of breast density was not recorded. In these patients, there were 16 mass lesions, three foci of calcifications, two masses with calcifications, and one area of architectural distortion. Lesions were analyzed by the radiologist (R.A.J. or R.S.S.) for the presence, morphology, and kinetics of enhancement. In particular, the appearance of rimlike enhancement that is sometimes seen in cancers on gadolinium-enhanced MR images was evaluated. Important morphologic features of the lesion were the shape and margin characteristics. The pattern of enhancement over time was assessed with the criteria reported for the kinetics of the time–signal intensity curves for gadolinium-enhanced MR imaging (9). The enhancement of lesions on MR images has been classified into three types. A type 1 curve shows gradually increasing enhancement and is more commonly seen with benign lesions. The type 2 curve, which shows a rapid increase in enhancement and then a plateau, is more indeterminate. The type 3 curve shows a rapid increase in enhancement with a rapid washout of the lesion and is more commonly seen in malignant lesions. Pathologic findings were obtained in 21 women. Lesion size was determined by means of pathologic analysis in nine cases in which the lesion was excised. For those cases with lesions in which no surgical biopsy was performed, the US report measurement was obtained in nine cases and a mammographic report measurement was obtained in one case. In two cases of microcalcifications and one of architectural distortion, no accurate size measurement was obtained.
Contrast Agent
The contrast agent (iohexol, Omnipaque 300; Nycomed, Roskilde, Denmark) was a nonionic solution containing 300 mg of iodine per milliliter, which is commonly used for CT. In our study, we injected 100 mL of the agent by hand over a period of approximately 1 minute.
Instrumentation
During this study, two digital mammographic systems were used for image acquisition. The first eight examinations were performed with a prototype system (Senoscan; Fischer Imaging, Denver, Colo). This was an imaging system in which a collimated fan beam of x rays passed through the breast and was recorded by a long narrow detector. Because of the fan geometry, the system did not require an antiscatter grid. This prototype system was equipped with a tungsten anode. For image acquisition, an imaging time of approximately 4 seconds was required, although the exposure time for a given pixel on the image was less than 0.25 second. The more recent images were acquired with a production system (Senographe 2000D; GE Medical Systems, Milwaukee, Wis). This system is based on another mammographic unit (DMR; GE Medical Systems) with a cesium iodide–amorphous silicon flat-plate detector to acquire the image in digital form. The x-ray tube has both molybdenum and rhodium targets and, in addition to the usual choice of molybdenum or rhodium filtration, the unit was modified to include a copper filter.
Technique
Typically, the contrast-enhanced digital mammographic procedure was performed in approximately 11–15 minutes. This included 3 minutes for placement of the intravenous catheter, 1 minute for obtaining the mask image, 1 minute or more for completion of the injection, and up to 9 minutes for acquisition of postinjection images. The breast was lightly compressed in the craniocaudal projection for the duration of the examination, with enough pressure to limit anatomic motion but not enough to significantly reduce blood flow. The craniocaudal view was used because it was easier for the patient to remain motionless in this position. In cases 12–22, imaging was performed with the production system, and an initial reference or scout mammogram was obtained with normal mammographic parameters, typically 26–32 kV, and the appropriate target and filter. The purpose of obtaining this image was to provide information regarding the soft-tissue anatomy and to provide anatomic reference points for the subsequent images that were to be obtained with identical conditions of positioning and compression. Our experience determined that this was not necessary, and it has since been eliminated to reduce the radiation exposure to the patient.
The exposure factors were then adjusted to produce an x-ray beam containing as high a fraction of x rays above the k absorption edge of iodine (33 keV) as possible. With the prototype system, exposures were obtained at 45 kV, with a tungsten target, filtered with 0.13 mm of holmium and 2 mm of aluminum, whereas with the production system, the molybdenum target was used with a varying kilovoltage of 45–49 kV and a molybdenum anode with added copper and aluminum filtration.
A single mask image was then produced. Immediately following this exposure, the women received an injection of 100 mL of iohexol that was administered in the antecubital vein of the arm contralateral to the breast of concern. Immediately after completion of the injection (time, 
1 minute), the first postcontrast image was obtained. A second postcontrast image was obtained 2 minutes later at 3 minutes, and then subsequent images were obtained at 5, 7, and 10 minutes; a total of six images for each patient in addition to the scout image were produced. The milliampere-second setting for each patient was chosen according to the thickness and composition of the breast, but once selected, it was kept constant for the mask and postcontrast images. Figure 1 summarizes the timing sequence. The total radiation dose for the six images was approximately equivalent to that from a single screen-film mammographic image.
|
Kinetics
The system was calibrated by using a test object consisting of a slab of breast tissue–equivalent plastic that contained wells filled with varying concentrations of iodine-based contrast medium (Fig 2). These were then used to determine projected iodine concentrations in units of milligrams per square centimeter. The uptake and washout kinetics of the contrast agent were then computed from the subtracted patient images. A region of interest was chosen by the previously mentioned radiologist at the site of the lesion on the image obtained at 1 minute, and the mean pixel value was computed for the same region of interest in the entire set of subtracted images. The region of interest was the brightest area of early enhancement in the lesion, and its size varied with the size of the lesion and ranged from 2,500 to 22,500 pixels (25–225 mm2). In addition, the same procedure was performed for a rectangular region of interest of similar size that was chosen by the radiologist at a site of normal breast tissue adjacent to the lesion. All measurements were then converted to the projected iodine concentration. The projected iodine concentration for the lesion site and the normal breast tissue site were plotted versus time.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
1 minute) followed by rapid washout (9). The case in Figure 3 shows that the contrast agent reaches peak signal intensity slightly later before it begins to wash out. The other malignancies were irregular and poorly defined masses (Fig 5). Inhomogeneous enhancement with linear areas of enhancement was observed in one case of ductal carcinoma in situ with invasive ductal carcinoma (Fig 6). The kinetic curves of three invasive cancers showed a decrease over time, with another four showing a plateau and one showing a continued rise.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
When Watt et al (4), Ackerman et al (5), and Watt et al (6) performed DSA of the breast, they observed a rapid and strong increase in iodine content followed by a washout on the subtracted images of malignant lesions, whereas much less or no enhancement was observed on images of benign lesions. This pattern was also found in some of our patients.
Heywang-Köbrunner (21), Heywang et al (22), and Kaiser and Zeitler (23) reported the enhancement kinetics and characterization of breast lesions by using contrast-enhanced MR imaging. They showed that enhancement with malignant tumors was rapid, whereas that with benign tumors was much slower. In our work, findings showed that the enhancement kinetics were not sufficiently consistent to allow reliable differentiation of benign from malignant lesions. It is generally believed that for good specificity in breast MR imaging, both morphology and kinetics should be considered. Our results also support this conclusion for contrast-enhanced digital mammography.
In our pilot study, enhancement was observed in 89% (eight of nine) of the invasive cancers. There was no enhancement in seven (58%) of the 12 benign lesions that were initially considered worrisome at mammography or US. In the three cases with positive findings at US and negative findings at mammography and in which no enhancement was seen, the lesions were confirmed to be benign. The morphology of the lesions was generally consistent with the benign and malignant features seen at other imaging modalities. However, there appeared to be less frequent enhancement in benign lesions compared with the experience at contrast-enhanced MR imaging (22–26). The kinetic curves did not consistently demonstrate distinctly different patterns for benign and malignant lesions. In some cases, however, kinetic curves were similar to those that have been reported for MR imaging.
In the presentation of the kinetics data, we have included curves for uptake, both in a region of interest corresponding to a lesion suspected of being malignant and also in an area corresponding to normal tissue. We note that in some cases on images acquired with the prototype system, the kinetic curve in the normal area appeared to assume a negative value of iodine. This was most likely caused by drifts in the output of the x-ray tube between the mask image and postcontrast images or possibly caused by a drift in radiation sensitivity of the x-ray detector with this prototype system. This system is no longer operational; however, such drifts would not have been atypical with that machine. We were able to perform a simple experiment with the production system, where a uniform slab of plastic attenuator material was imaged in a sequence simulating that used in contrast-enhanced digital mammography. It was observed that in some cases the subtraction images became negative because of small drifts. However, the effect was sufficiently small (ie, typically corresponding to a projected iodine concentration less than 0.1 mg/cm2) to be below the threshold of detectability of iodine. In any case, such effects could be eliminated by including a small plastic reference object adjacent to the breast and by scaling all acquired images to hold the signal from this object constant.
The current implementation of this technique is restricted to a unilateral study of the breast, as only one breast can be monitored during the injection of contrast medium. Although a second injection is possible to image the second breast, a delay of at least 30–60 minutes would be needed between injections to minimize background enhancement in overlying tissue from the initial injection. The use of the craniocaudal position does not allow as much breast tissue to be visualized as does the mediolateral oblique view. Because the images are projections through the entire breast, large lesions yielded stronger enhancement than did small ones, and as such, size affects quantitative interpretation of the enhancement curves. In this study, acquisition of the first image was delayed as a result of our use of manual injection, which required approximately 1 minute for injection of the entire bolus of contrast medium.
However, experience from MR imaging studies of the breast suggests that a high temporal resolution is desirable to resolve early enhancement of some lesions and to provide an adequate number of temporal samples to conduct a full assessment of contrast media kinetics. The choice of an optimal temporal resolution depends on the way that enhancement data are to be used, and considerable variation appears in the literature. Kuhl and Schild (9) recommended that a 1-minute temporal resolution is adequate, which is consistent with findings of this study. However, Boetes et al (27) showed that differentiation of benign and malignant lesions can be improved by repetitive imaging of a lesion every 2.3 seconds. Because the volume of injected medium for breast MR imaging is smaller (15 mL) than used in this study, it is possible that in MR imaging the tumor region may receive a more sharply defined bolus, and this could influence the measurement of lesion kinetics. Earlier data could be obtained with this technique by administering the contrast medium more quickly with a power injector. Now that the technique has demonstrated the ability to show cancers, we plan to recruit women with dense breasts and mammographically occult or subtle findings to evaluate the possible additional benefit over regular mammography.
Lewin et al (28) have discussed a dual-energy approach to contrast digital mammography and showed images similar to those presented here. In their technique, two images are produced in rapid sequence, one containing x rays predominantly below the k edge of iodine (33.2 keV) and one at higher energy. The iodine signal is isolated with performance of a weighted subtraction of the two images. This procedure eliminates the need to produce a mask image and thereby minimizes the effects of motion between the two images. Because two images must be obtained at each time point, presumably the radiation dose is higher than it was with our method if a multipoint study of kinetics is performed.
Another possible area for improvement is the elimination of background uptake from overlying and underlying tissues in the breast. With even a low level of uptake in these superimposed and adjacent tissues, the projected signal of the entire thickness of the breast could reduce the conspicuity of a lesion and affect the quantitative measurements. This problem with overlying tissue does not occur with breast MR imaging, which produces tomographic images. With contrast-enhanced digital mammography, the problem could be overcome with use of a tomographic technique such as tomosynthesis (29) or the application of tuned-aperture CT in mammography (30).
The results of this preliminary study suggest that contrast-enhanced digital mammography potentially may be useful in the identification of lesions in the mammographically dense breast. As in MR imaging, other applications may be in the identification of the extent of disease or in the detection of an otherwise occult carcinoma that has manifested with axillary metastases. This information may aid in the diagnosis and guidance of core-needle biopsy or excision of these lesions. Furthermore, with the increasing availability of digital mammography, contrast-enhanced digital mammography will become accessible and relatively inexpensive compared with current MR imaging technology. These results encourage further investigation of contrast-enhanced digital mammography as a diagnostic tool for breast cancer.
|
FOOTNOTES |
|---|
Abbreviation: DSA = digital subtraction angiography
Author contributions: Guarantor of integrity of entire study, R.A.J.; study concepts, R.A.J., M.J.Y., R.S.S., D.B.P.; study design, R.A.J., M.J.Y., R.S.S.; literature research, R.A.J., M.J.Y., R.S.S., M.S., D.B.P.; clinical studies, R.A.J., R.S.S., N.M.D., A.G.; experimental studies, M.J.Y., M.S.; data acquisition, N.M.D., M.S., A.G.; data analysis/interpretation, R.A.J., M.J.Y., M.S., R.S.S.; manuscript preparation and definition of intellectual content, R.A.J., M.J.Y., M.S., D.B.P.; manuscript editing and revision/review, R.A.J., M.J.Y., D.B.P.; manuscript final version approval, R.A.J., M.J.Y.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  C. Dromain, C. Balleyguier, S. Muller, M.-C. Mathieu, F. Rochard, P. Opolon, and R. Sigal Evaluation of tumor angiogenesis of breast carcinoma using contrast-enhanced digital mammography. Am. J. Roentgenol., November 1, 2006; 187(5): W528 - W537. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K C YOUNG Recent developments in digital mammography Imaging, June 1, 2006; 18(2): 68 - 74. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. H. BRUSIN Digital Mammography: An Update Radiol. Technol., January 1, 2006; 77(3): 226M - 234M. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Koomen, E. D. Pisano, C. Kuzmiak, D. Pavic, and R. McLelland Future Directions in Breast Imaging J. Clin. Oncol., March 10, 2005; 23(8): 1674 - 1677. [Full Text] [PDF]
|
|
|
|
 |
|
 E. D. Pisano and M. J. Yaffe Digital Mammography Radiology, February 1, 2005; 234(2): 353 - 362. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Mahesh AAPM/RSNA Physics Tutorial for Residents: Digital Mammography: An Overview RadioGraphics, November 1, 2004; 24(6): 1747 - 1760. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
) and an area of normal tissue (
) adjacent to the mass. Curve for carcinoma shows early enhancement with a decrease over time, while the curve for normal tissue continues to increase at 10 minutes.
