In the 1980s, the media exposed some facilities and some radiologists who were providing poor-quality mammography and were failing to inform women who were found to have a cancer that they had a problem. As a result, the diagnosis of their cancers was delayed. These stories led to a congressional investigation that ultimately resulted in the passage of the Mammography Quality Standards Act (MQSA). The MQSA was a mandate to produce basic standards for mammography, including equipment; qualifications for the radiologists, technologists, and physicists involved; quality control; and quality assurance programs. The Food and Drug Administration (FDA) was assigned to oversee a national program to try to insure that mammography facilities adhered to basic quality standards.

In an effort to reduce the chance that a significant finding would not be brought to the patient’s attention, causing her diagnosis to be delayed, the media exposés urged women to obtain their mammography reports before they left the mammography screening facility. Other facilities used the immediate interpretation of mammograms as a marketing tool, and many women began to expect that their screening study should be interpreted while they waited. It is now apparent that this is a very expensive and a somewhat risky practice. We found that when trying to provide immediate interpretations, patients began to “back up” in the waiting room. Some who were in a rush to leave would pressure the technologists for the reading, and the technologists would pressure the radiologists. This forced the radiologist to rush the interpretation, raising the risk that a lesion would be overlooked. We had performed a study to evaluate the benefit of double reading. The study included over 6,000 women among whom we detected 39 cancers. Our first reader read slowly and methodically and took well over an hour to review 40 mammograms. The mammograms were then read again by a second reader, who reviewed all the studies in a few minutes. The main reader found 36 of the cancers. The “quick” (second) reader found 3 malignancies that had been missed by the main reader. However, the quick reader missed 8 cancers that were found by the main reader. Our study points out the fact that two readers are better than one, but if there is only one reader, that single reader should read slowly and methodically and not be rushed because a patient is waiting for the report. As a result of our study, we now provide a second reading at no charge to the patient, but we have stopped providing an immediate reading because of the danger in rushing the reading.

Providing immediate interpretations of screening mammograms also virtually precludes the practice of double reading. Since the latter is not reimbursed, the only way to provide a double read is to interpret studies efficiently so that the second reading takes a minimum amount of time. Efficient double reading is facilitated by having the mammograms interpreted in batches. For these reasons it is probably not a good idea to read screening mammograms while the patient waits. If mammograms are not interpreted “on line” and the patient is not provided with her report before leaving the facility, then, by law, there needs to be a system in place to be sure that the report is generated and sent to the patient within 30 days of the study. Although it can be debated as to whose responsibility it is to try to insure that a patient complies with follow-up recommendations, we believe it is a good idea to have a system that tries to monitor whether women comply with those recommendations. This too must be organized in an efficient manner since it can rapidly become very labor intensive and expensive, and this, too, is not covered by reimbursement.

Interpreting screening studies in batches can reduce the cost by maximizing the radiologist’s productivity. The disadvantage of the patient’s not having an immediate report is more than offset by the reduced expense and the added advantage that reading mammograms on an alternator permits double reading. As has been shown in the interpretation of chest x-rays (1), barium enemas, and other imaging studies, there is a psychovisual phenomenon that guarantees that all radiologists will periodically fail to perceive significant abnormalities. Mammographic interpretation is no exception. Every radiologist, no matter how skilled and no matter how hard he or she concentrates on looking for cancer, is guaranteed to periodically overlook cancers that are visible on mammograms. Because different readers overlook different findings, having more than one reader interpret a mammogram reduces the error rate. Interpreting mammograms on an alternator permits multiple reviewers with little increase in cost. In our own study, double reading in this fashion increased our breast cancer detection rate by 7%. Bird (2) found that a second reviewer increased the cancer detection rate by 5%, and Tabar et al (3) have reported as high as a 15% improvement in cancer detection with double reading. This increased yield from double reading has also been documented by Thurfjell et al (4). Double reading can be accomplished using digital mammography once the workstations allow random lists to be generated.

Mammography is invaluable in the detection of anomalies (benign and malignant lesions) in the breast (detection), but it is less suited for the differentiation of those anomalies into benign or malignant (diagnostic) categories. As breast imaging techniques have evolved, a better understanding of the underlying histopathologic changes that account for some of the mammographic findings has developed. Nevertheless, the problem remains that the diagnosis (in the true sense of the word) of many lesions, based on imaging criteria, often defies the high levels of certainty required if a tissue diagnosis is to be avoided. In view of the important consequences of early breast cancer detection and the high level of safety in obtaining a tissue diagnosis, the accuracy of a noninvasive diagnostic study that is used to avoid a biopsy should approach 100%.

Despite uncertainties, criteria have been established to assist in diagnosis. Before the instigation of an invasive procedure, it is important to analyze and categorize anomalies in the breast so that appropriate action can be taken. Mammography and appropriate diagnostic techniques, such as ultrasound for differentiating cysts from solids, should be used to maximize the diagnosis of early cancers and reduce the need to biopsy benign lesions whenever possible.

The observer should avoid the urge to immediately view mammograms too closely. It is preferable to sit back and view the images from a distance, bringing foveal vision to bear on the entire image. Some observers find the use of a minifying lens beneficial to reduce the size of the images. This aids in appreciating areas of asymmetry.

The images should then be viewed from a distance of 1 to 1½ feet to begin evaluating some of the tissue details. Finally, the observer should evaluate the mammograms so as to view the subtle details. Although all structures visible on the mammogram can be seen by the normal, unaided eye, a magnifying glass is helpful for making small detail
more obvious. One of the advantages of digital images is that they can be electronically enlarged (the same as photographic enlargement), obviating the need for a magnifying lens. Until monitors are available that can display all of the digitally acquired imaging data, when reading digital images the observer should remember that, at some point, the images is sufficiently displayed as to provide viewing of all of the data that were acquired since some systems acquire data at a higher resolution than can be displayed over the entire breast on the monitor.

The images should be systematically reviewed with particular attention to the deep tissues, especially near the inframammary fold and high in the axilla (places where the rare tumor may be overlooked). Because most cancers develop at the interface between the edge of the parenchyma and the subcutaneous and retromammary fat (9), these tissues
should be evaluated for architectural distortion, masses, or calcifications. Digital systems facilitate the evaluation of these zones by providing equalization algorithms that enhance these tissues, which are often “burned out” on film/screen mammograms (Fig. 13-5).