The Reality of Digital Mammography: How Digital Are You?
Digital mammograms are now being delivered at more than 50% of the mammography sites in the nation. While the technology was validated in the US National Cancer Institute’s Digital Mammographic Imaging Screening Trial,1 the small statistical advantage of digital over analog mammography was found primarily in younger, premenopausal patients with radiographically dense breast tissue.
Gerald R. Kolb, JD
Continuing research has extolled the virtues of digital mammography, but the truth is that analog mammography is still very good. The digital image has presented breast imagers with an incremental improvement in the quality of the image that they interpret. From an image perspective, digital mammmography has been an evolutionary (rather than a revolutionary) change. It is fair to comment, however, that the major impetus for acquiring digital mammography came from marketing pressure. From the perspective of process and workflow, however, digital mammography is nothing short of revolutionary, but only if you are open to, budget for, and integrate digital mammography for its abilities to enhance the delivery system, as well as for its direct image value. Digital mammography can be a game changer, and it can add value to your organization that goes far beyond the incremental increase in reimbursement. Digital mammography, however, can accomplish this result only if you reengineer your entire delivery system to become a digital system and become a truly digital breast center. The Digital Image There is value in the digital image. We all know about the interpretive enhancement provided through window and leveling functions, electronic magnification, and other interpretation tools. While these are helpful adjuncts in the interpretive process, step back for a moment and view that image with new eyes. It is digital, and it can be annotated with arrows or ovals to identify areas of interest. Little electronic notes can further describe these, and some systems can be configured to allow the addition of voice clips to patient files. With this level of information available on a recalled patient, for example, the technologist can move directly to the additional imaging required, without needing to go over the file first with the physician. It should even be possible to indicate relative level of complexity for recalled patients (for use by schedulers in allocating time for the diagnostic visit). Annotation can work both ways, and it can work in real time. The traditional work pattern in a diagnostic mammogram is for the technologist to acquire the images, then leave the patient in the mammography room and go to the reading area to discuss with the physician what should be done next. At that time, the technologist can also communicate any ancillary issues to the physician (for example, the patient is complaining of a new pain in her left breast). Beyond the obvious fact that, in a busy breast center, the technologist might have to wait in line to talk with the physician, the patient has been left (half naked) to fend for herself until the technologist returns. With a proper digital system, the patient’s images can appear at the top of the physician’s worklist as soon as they have been acquired. The technologist can also add annotations (for example, the patient is quite frail and very sensitive to compression) that might assist the physician in ordering additional studies. Both patient care and throughput are improved by keeping as much of the professional interaction between the physicians and technologists in an electronic format as possible. Patient-care improvement is an important point that cannot be overemphasized. The previous example illustrates the ability to keep the technologist with the patient during a diagnostic mammogram; in screening, digital imaging also allows significant reductions in the time required to acquire the images. In fact, depending on the equipment being used, the actual time from initial positioning through acquiring the last image ranges from 5 to 8 minutes. A very short visit might make the patient (and the technologist) feel rushed, so many successful centers assign two technologists to each screening room. In this scenario, the technologist is with a single patient for closer to 20 minutes, but only has 10 minutes of room time. The additional time is spent discussing the patient’s history and any new problems. Note that additional workstations and space should be provided for obtaining this information. One of the most important aspects of the digital image is that it can be viewed anywhere, at any time, by anyone with the appropriate security access. This characteristic, together with the digital format of other information in the system, can change the delivery system in ways that are limited only by our creativity. File Tyranny In the analog world, the patient’s file (containing the written patient record, medical reports, and copies of images) has dictated her movement through the system, and thereby how all work proceeds within a breast center. When you analyze the work process in the analog center, you quickly realize that the patient can only advance from one room to another, and the technologist and physician can only accomplish their tasks, when they are in physical possession of the file. A secondary task of every employee becomes making sure that paperwork documenting contact with the file (patient) is properly placed in the file; otherwise, there will be no record that the patient registered, was provided with the privacy policy, had her history reviewed, or whatever else the task entailed. If it didn’t get into the file, it wasn’t done, legally speaking. The need for a file does not go away with digital mammography, but that file is transformed into an electronic medical record (EMR). In breast imaging, our EMR is more complex because it contains digital images in addition to the text-based material that is common to nonimaging EMRs. The implications of an EMR for the organization are immense. In fact, I would submit that the value of the well-integrated EMR is limited only by privacy regulations and the imaginations of those who use the data that these records contain. Defining a System A digital system has two basic segments and several different components. Success in implementing a digital system will be directly proportional to the level of integration of the system’s segments and components with each other. The segments are the human resources and the technology resources. In a truly integrated system, each segment is utterly dependent upon the other. To put this in proper perspective, if the technology goes down, you can usually finish with the current patient, but you might not be able to go much further. System reliability is the reason that it is very important both to purchase robust, reliable equipment and to maintain that equipment under comprehensive service agreements. Going bare is not an option, especially with regard to the key IT. You must also invest in human resources. Well-integrated systems will reduce FTEs, or at least reallocate staff time more productively, but using the advanced technologies involved in the information backbone will require both training and patience. Do not assume that having an email account qualifies one to navigate an information system successfully. The learning curve affects everyone to one extent or another, and it is critical, in planning the rollout of a system, to plan for learning delays. This is particularly true for physicians as, if they are dedicated breast imagers, they might have no experience interpreting in a digital environment. Many physicians find that the learning curve covers approximately six months, but that they never get back to their best analog speeds. This is the natural result of having options to control image presentation, however, and should also result in some decreases in the recall rate. The principal accomplishment of the technology segment of the system is that it enables information, including images, to be viewed anywhere, at any time, by anyone who is appropriately authorized to do so. Technology allows us to remove the reliance on the physical file that has dictated our workflow for so long; there is no more delivering mammography files across town to be read. The technology segment of the system consists of five basic components: the acquisition unit, the PACS, the workstation, the RIS, and the mammography reporting system. The Acquisition Unit The acquisition unit includes both the mammography unit and the technologist’s workstation. The workstation should have a two-megapixel monitor, as well as an enhanced feature set that allows the workstation to communicate, through the PACS, with the interpreting workstation. The uprated workstation is required because the one-megapixel monitor that is standard on many units will not satisfactorily display motion artifact, hair, or other technical image defects. The acquisition-unit decision is the easiest decision you will make in transitioning to digital mammography. Choosing CR or DR mammography is still largely a matter of preference. Do not, however, make this decision based on talk about future mammography technology. All acquisition technologies currently approved by the FDA are first-generation technologies and will be replaced by future developments. The replacements will be, at best, forklift upgrades. In this kind of environment, it is probably prudent to project no more than a five-year usable life for new digital-mammography equipment. The PACS The PACS is the heart of the information system because it is the repository of all of the images and because it controls access to, and storage of, the images through established protocols. If something has to do with the image, it will involve the PACS. Your PACS must be robust, scalable, and redundant. Third parties can provide redundancy, but the PACS must be reliable, as it will control all activity within the center. Scalability is critical for a PACS because the amount of information that it will have to administer is staggering. Each screening mammogram will require up to 500MB of information for interpretation by the physician, depending on the resolution of the acquisition unit. Diagnostic cases can involve considerably more information. A good way to visualize a PACS is as a large filing cabinet for images. Unlike a conventional filing cabinet, however, the PACS is fully indexed and is able to deliver images in the order and configuration in which they are requested. This ability allows individual preferences to be established for each physician. Because the system is electronic, the information is available on demand, anywhere, providing flexibility that has been unknown in breast imaging. Take, for example, breast programs with more than one site: The PACS can decouple the physician from the screening location. Providing the freedom to move screening mammograms between physicians allows more effective work allocation. Today, most PACS also allow access, again with proper security, via Internet. Surgeons and other referring physicians with a need to see the patient’s images can be provided with easy access to those images with a few mouse clicks. They will not have diagnostic-quality images without five-megapixel monitors, but they will be able to identify the tumor location and extent of disease, as well as to see the annotations made by the interpreting physician. Within an organization that has an existing PACS, there is a tendency to add on to that PACS. I recommend considering a breast-specific PACS even in these situations. Generally, a breast-specific PACS can be acquired for less capital than that necessary to upgrade an institutional PACS to handle the increased volume. In addition, isolating the breast images will avoid the political controversy that invariably occurs when the slowing of the institutional PACS is noted. It does not matter that the hospital might have recently added ultrafast CT, with its huge image files; the slowdown will be blamed on mammography. The breast PACS might, however, use the institutional PACS for long-term storage, which will eliminate some cost. The breast PACS will also incorporate all breast modalities directly, and can access the institutional PACS for any studies that may be stored there. While a PACS must handle large amounts of data, the cost of storage media has dropped dramatically over the past few years, and storage is not generally a cost issue. Network bandwidth is, however. The individual components of the system are relatively fast, and require a dedicated network with a 1GB switch to perform optimally. When the need to transfer information between sites arises, bandwidth becomes a serious concern. Generally speaking, a capacity in the range of 10MB to 50MB per second—well beyond that of a T1 line—will be needed to eliminate slowing as both images and information move around a distributed network. I urge thorough planning in this area before any investments are made because bandwidth expense can be a major operating item. The Workstation The interpreting workstation serves as the operating interface with the PACS, RIS, and reporting system. In a very real sense, it provides the physician with a cockpit from which all aspects of the clinical process are at literally at his or her fingertips. Through the interpreting workstation, the physician should also be able to access, via easy-to-build bridges, images that reside on other PACS. As with the PACS, there might be controversy regarding which interpretive workstation to select. A fundamental reality of the new mammographic environment is that centers are receiving an increasing number of prior images in digital format, and many of them might have originated on different vendors’ equipment. Given this factor, and the first-generation nature of the acquisition modalities, I would encourage the selection of a PACS workstation that has been carefully optimized for mammography. These units will be DICOM compatible, but be sure that the PACS vendor is participating in the Integrating the Healthcare Enterprise initiative. This program is designing advanced standards that will ensure that the devices manufactured by member companies can communicate with each other. Integration is, again, the heart of success in the creation of a true system. The RIS The RIS interfaces with the PACS and manages people, events, and information. The RIS is a large database that is designed to be the repository of patient demographic information, medical reports, physician information, and anything else (other than images) that might be required in providing clinical care and documenting that care in a medical record. The RIS sounds complex because it is. The RIS is often as expensive as the PACS, but much of that investment (and the PACS investment) will be wasted if the RIS is not successfully integrated with the PACS. Integration of PACS and RIS should be bidirectional and in real time. When clinical activities are occurring, the technologists and the physicians will be drawing information from both PACS and RIS; there should be no duplication of effort, nor should there be operator-initiated shifting from system to system. Seamless access in clinical activities translates directly into time savings. These savings might come in small increments, but those are high-cost increments with large volume multipliers. Consider that a radiologist interpreting screening mammograms at a rate of two per minute controls approximately $4,500 per hour in global revenue (for mammograms reimbursed at $150). Increasing that productivity by 10 seconds per interpretation increases the revenue contribution by almost $450 (10%). A RIS works well for general radiology, but it is deficient when it comes to breast imaging. The culprit is the generality of the RIS. Most of the tasks of the RIS (such as scheduling, registration, and administrative tracking) are common to all of radiology. The RIS is also integrated with the PACS, providing seamless access. In breast imaging, however, we have specific reporting requirements mandated by the Mammography Quality Standards Act of 1992 (MQSA), and there is a small (but deep) information set to be acquired and maintained for each imaging contact. Because we see most of our patients every year, we also have clinical-management opportunities that can be far more focused than those of a general radiology practice. This opportunity has led to the development of the mammography reporting system. The Mammography Reporting System The reporting system provides focus to the RIS. It should be integrated with the PACS and RIS, allowing information specific to breast imaging to be incorporated in the RIS easily, as well as linked to the images in the PACS. A mammography reporting system effectively automates much of the process, including MQSA reporting and recordkeeping, along with the increasingly important mammography audit. These systems also allow automatic fax transmission of reports and automatic analysis of risk factors, and they are generally designed to give clinicians a breast-specific user interface. As with the RIS, a good mammography reporting system is expensive, but it is vital to enhancing overall productivity in the breast enterprise. Human Resources While much of the staff-related investment in digital mammography will be in training, you will need a new person on the breast-center team: a PACS administrator. This is a full-time position, and this person will be critical to the success of the implementation. Even institutional breast centers have found it vital to have a PACS administrator within the department. When recruiting for this position, it is important to remember that the person will be managing the entire information system and will be instrumental in its success. Hire wisely. Going digital involves much more than just substituting digital equipment for analog equipment. The venture is exciting, it is rewarding, and it is expensive. Much of the expense is front-end loaded, which means that you will only receive the full benefit by increasing procedural volumes and, thereby, decreasing incremental cost. You must plan strategically, and the first step in that planning must be a commitment to providing excellence in breast care in a growing program. Without this level of commitment, it will be difficult to justify the capital investment necessary to build a sustainable program. Gerald R. Kolb, JD, is chief knowledge officer, Solis Women’s Health, Austin, Texas, a privately owned outpatient imaging company with 13 breast centers and 15 mammography centers in six states.