The established technique for breast imaging, both screening and diagnostic, is film-screen mammography. Using film-screen mammography it is possible to detect about 90% of breast cancers and find these at an early enough stage to reduce mortality by approximately 50%. Although the sensitivity of film-screen mammography is high, its specificity is low. In most reported series, the likelihood that a lesion found by mammography and sent to biopsy will be malignant is only 20 to 35%. Additionally, clinically palpable lesions often are not sufficiently characterized by mammography to eliminate the need for additional workup.

Widely accepted ancillary breast imaging techniqes include sonography and magnetic resonance imaging (MRI). Sonography has the ability to demonstrate margins and internal texture, often more fully than mammography. Most importantly, this makes it possible to diagnose simple cysts within the breast. In many patients sonography also makes it possible to increase or decrease suspicion that a lesion is malignant and to more accurately map the extent of tumor within the breast than is possible with mammography.

MRI is well established as the most accurate imaging modality for assessing implant complications. Studies also have demonstrated that it is worthwhile to hunt for the primary tumor when metastatic disease, consistent with metastatic breast cancer, is present and the site of the primary cancer is not seen with mammography or apparent on physical examination. In other special instances, such as differentiating scar from tumor recurrence after breast conservation or mapping the extent of cancer within the breast, MRI can be particularly helpful.

Even with this imaging triad, evaluation of breast disease by the radiologist can sometimes be limited. Dense tissue obscures some lesions on mammography. Screening fails to identify about 10% of breast cancers. Even when using sonography and MRI it is not always possible to differentiate benign lesions from those that are malignant. In an effort to confront these and other issues in breast imaging, new imaging technologies are in development to improve detection and characterization of disease in the breast. The most important of these is isotope imaging of the breast using either sestamibi scanning or positive emission tomography (PET), and mammography with information captured digitally rather than with film-screen.

Digital mammography

Film-screen or analog imaging makes it possible to record a continuous representation of spatial and intensity variations in the x-ray pattern. In contrast, a digital image is acquired as a composite of square elements or pixels. At each pixel, which is usually 0.05 to 0.1 mm on each side, the brightness is determined by the average of the brightness over that area. For full-field digital breast imaging, the format ranges from 4000-4800 ¥ 5000-6000 pixels per image.

The digital image is formed when a detector absorbs the x-rays and converts them to an electrical signal corresponding to each pixel. The intensity of the x-ray patterns is limited to a finite set of gray levels. The information in the image can be stored in a computer memory. The image can be printed on film and read from a viewbox or displayed on a monitor, allowing manipulation of the window and level of the image, as well as allowing other alterations in the display of the information such as magnification.

Many of the potential advantages of digital mammography are not unique to breast imaging but also apply to other areas in radiology. When images are digitally acquired and displayed, film is eliminated. This eliminates the expense, time and effort, and quality control programs necessary for film processing. When images are stored electronically rather than on film, the need for film storage also is eliminated. Additionally, the elimination of film processing makes the images available for interpretation more quickly. This advantage has led to the widespread use of spot digital mammography in stereotactic breast biopsy procedures, the only currently commercially available digital mammography technology.

The lack of storage of the mammographic information on film also eliminates issues of possession of the original image that exist with film-screen mammography. Because the original image information is digital, not analog, the quality of each reproduced image is the same as the original, and the problems of loss of information that occur with copies of film-screen images do not occur.

Digital mammography also offers advantages that are possible with linkage to computer aided diagnosis (CAD) systems. Because the digital information is available in a format that is usable by CAD systems, it can easily be fed into such systems for analysis. This makes possible automatic double reading of mammograms with the second reading done by the computer. Although CAD systems currently are not available for clinical use, CAD technology can be readily interfaced with digital information to enable easy integration of CAD systems, if and when the systems become available for this application.

When mammograms are available in a digital format, it is possible to transmit images via telephone. The use of teleradiology enables a radiologist to operate multiple screening or diagnostic centers from a single site, monitoring studies in real time. This can help decrease the cost of delivering mammography services. Additionally, expert radiologists can manage patients at distant sites and offer consultation with second opinions more readily.

Despite these considerable advantages, some problems also exist in using full breast digital imaging systems. When available, these will be considerably more expensive than film-screen imaging systems. At currently estimated prices, they may be three to six times more expensive than traditional mammography. This high price is a result of the high resolution necessary and the large amount of information required for each image. Breast imaging requires very high resolution to maintain diagnostic efficacy. The level of resolution available in film-screen mammography is 20 line-pairs per millimeter. Evidence suggests that with digital systems, 10 line-pairs per millimeter may be adequate resolution. Even with this decrease in spatial resolution, the information to be stored for each image is very large. To accommodate this, large volumes of computer memory will be required to store the information contained in each mammogram.

Other technical difficulties have arisen in the design of these systems. Available monitors for image interpretation have limited display capabilities; monitors that could display all information in the digital image would be extremely expensive. A variety of configurations for the actual mammography unit have been proposed. In some of these, patient positioning may present some difficulties. In others, a long exposure time is required and, therefore, patient motion may become a problem.

It is reasonable for breast imagers to expect that full field digital imaging will be available commercially within the next few years. These systems will have the potential to eliminate the need for film-based images, film processing, and storage. Linkage with computer aided diagnostic technology also will likely be available in the future. It is unknown whether this technology will result in improved diagnosis of breast disease.

Positive emission tomography (PET)

PET imaging produces an image based on quantitative metabolic activity mapped with a dual-gamma (positron-emitting) tracer. The most commonly used radiopharmaceutical has been a glucose analogue, 1-deoxy-2-fluoro (18F)-D-glucose (FDG). Glycolysis can be measured by the accumulation of FDG in cells and this level is increased in cancer cells compared to normal tissue. A role for PET using FDG as a tracer has been suggested in breast imaging to differentiate benign from malignant lesions and to stage breast cancers. It also has been suggested that FDG uptake can be correlated with histologic grade, tumor angiogenesis, and chemotherapeutic effects in breast cancer.

Experience with the use of PET to assess disease confined to the breast has been limited. Tumors of less than 1 cm in diameter are difficult to resolve, and some benign entities show levels of FDG accumulation similar to that found in carcinoma. Also, scanning can be falsely negative in obese patients. As with digital technology, the equipment needed to produce images is expensive. PET scanners can cost $1-2 million and a cyclotron is need to produce the isotope used for scanning.

PET technology has been found to be most useful in assessing disease outside the breast, particularly axillary nodal metastases and bone metastases (figure 1). PET has a negative predictive value of excluding axillary metastases of 95%. It also is highly reliable in determining the presence of lytic bone metastases. However, it has not been found to be very useful in assessing sclerotic lesions. Determining the response to treatment of locally advanced breast disease is another valuable application of PET scanning, as well as determining response of systemic disease to treatment.

Due to its high cost and the limited availability of the tracer isotope, widespread use of PET scanning is unlikely. However, where this technology is available, it will likely be most useful in breast cancer patients for staging large tumors and assessing response of systemic disease to treatment.

Sestamibi scanning

A variety of scintimammographic or single-gamma emitting radiopharmaceuticals have been tested to determine their usefulness in breast cancer imaging. Recently, the most frequently reported tracer applied to breast imaging has been Tc-99m-sestamibi. This technique has been reported to have a high positive predictive value, sensitivity,

and specificity for the diagnosis of breast cancer, but efficacy has largely been limited to the diagnosis of large, often palpable cancers (figure 2). Tumors of less than 1 cm or medially located lesions are poorly imaged with this technology. Reported sensitivities have ranged from 97% for cancers of greater than 2 cm to 26% for those of less than 0.5 cm. In addition, false positive results have been reported in a variety of benign entities, including normal breast tissue.

Improvements in the detection of lesions with sestamibi scanning will require decreased detector size and increased maneuverability of the detector, permitting reduction in the amount of tissue between the lesion and detector. Added maneuverability may increase the ability of this technique to characterize medial breast lesions. However, even with improved technology, the usefulness of sestamibi scanning in the diagnosis of breast disease is questioned by many imagers. It is uncertain how this technology might decrease the time, expense, and accuracy of diagnosis presently possible with mammography, sonography, MRI, and percutaneous needle biopsy techniques.

The current application of these three new technologies is as yet unestablished. Full breast digital mammography, when widely available, will offer the breast imager the advantages that digital radiography has demonstrated for other body parts. However, it will be expensive to install and will require the radiologist to adjust to new technologies when interpreting images. Whether there will be any improvement in diagnosis is as yet unestablished. Isotope imaging with PET is expensive, and the radiopharmaceutical has only limited availability. Of all settings for which PET scanning is available, it seems to be most worthwhile in screening the axilla and the skeleton for metastatic disease, and in follow-up studies for determining response to treatment in women with widely metastatic breast carcinoma. Tc-99m-sestamibi scanning of the breast has only been shown to be effective in relatively large breast cancers, and its role in the diagnostic armamentarium of breast imagers remains unestablished. AR

References

1. Williams MB, Pisano ED, Schnall MD, Fajardo LL: Future directions in imaging of breast disease. Radiology 206:297-300, 1998.

2. Feig SA, Yaffe MJ: Digital mammography, computer-aided diagnosis and telemammography. Radiol Clin North Am 33:1205-1230, 1995.

3. Oshida M, Uno K, Suzuki M, et al: Predicting the prognoses of breast carcinoma patients with positron emission tomography using 2-deoxy-2-fluoro (18F)-D-glucose. Cancer 82:2227-2234, 1998.

4. Palmedo H, Biersack JH, Lastoria S, et al: Scintimammography with technetium-99m methoxyisobutylisonitrile: Results of a prospective European multicentre trial. Eur J Nucl Med 25:375-385, 1998.

5. Khalkhali I, Cutrone JA, Mena IG, et al: Scintimammography: The complementary role of Tc-99m sestamibi prone breast imaging for the diagnosis of breast carcinoma. Radiology 196:412-426, 1995.

Dr. Dershaw is the Director of the Breast Imaging Section in the Department of Radiology at Memorial Sloan-Kettering Cancer Center in New York City.