Solving the Digital Dilemma: A Strategic Approach

Why Digital Mammography? 

A complete discussion of the reasons for acquiring and implementing digital mammography is beyond the scope of this article; however, the following is a partial list of the clinical and business drivers for the technology:

Intentionally omitted from the preceding list is marketing, which has probably driven more sales of digital mammography to date than any other factor. Marketing is considered by the author to be of subordinate importance in the implementation decision for digital mammography, because a focus on marketing will almost inevitably serve to draw needed attention away from the strategic aspects of acquisition and implementation, and the need to comprehensively redesign workflow.

The results of the DMIST study1 have recently been released,2 and digital mammography has been proven to be superior to film screen mammography, in a limited spectrum of patients. Notwithstanding this less than complete validation, it is clear that digital mammography will continue to be implemented across the US and in Europe at an accelerating pace. Beyond the potential for improving interpretive quality that is inherent in digital technologies, and which derives from the ability to manipulate digital data, digital mammography “completes the loop” with respect to the digital transformation of imaging, and presents significant opportunities to improve the overall economics of breast-imaging. It is also clear, as will become apparent below, that maintaining both analog and digital mammography platforms in one delivery system is untenable from both the clinical and the economic perspectives.

Challenges to Digital Implementation 

Each of the manufacturers of FDA-approved digital mammography equipment has developed its digital mammography technology from the perspective of replacing the analog or film screen mammography unit with a digital acquisition device, and adding a specialized viewer on which the interpreting physician could view the images. Unfortunately, this is a very shortsighted view, which ignores the natural implications and potential of any digital image, and has handicapped many implementations by early adopters. Think, for a moment, of the consumer digital camera and how it has changed photography. The operator not only controls electronic image magnification but, with minimal software, images can be cropped, colors can be enhanced or changed, titles added, people or objects added or deleted, and the result transmitted across the world to digital devices—including cell phones!

Clearly, the potential of digital mammography technology goes far beyond a simple one-for-one technology replacement of the analog image. The major nonclinical implication of digital mammography is not the replacement of film and processing; rather, it is in digital mammography’s potential to free the mammography delivery system from the tyranny that the mammography file jacket imposes on the system. This freedom can be attained, however, only if the implementation plan is strategic and encompasses the entire potential for the new technology. The implementation vision must reach out to contemplate all process improvements that may be achieved.

Consider, for a moment, the workflow in a film screen mammography facility, and one must immediately think of the patient file folder. It may accurately be stated that nothing in the breast facility happens without the folder. The folder contains the patient history, prior reports, and prior images. Following the progress of the folder through the facility, it is first “married” with new information at patient registration, it must be in the technologist’s hands as she prepares for imaging the patient, and it accompanies the new images to processing and then on to the viewer for the hanging of both current and prior images. The physician has the file as he or she reviews the films, and in most centers, including early adopters of digital mammography, the file is still the physical repository of any prior history of the patient, including mammography reports.

Transporting the file around within the facility not only carries a high direct cost,a but, by controlling the workflow of the system, the file also places defacto limits on the productivity of both the capital and human resources in the center. This is examined in more detail in the following paragraphs.

In the mammography delivery chain, it is the technologist who controls productivity. If she is doing anything other than delivering a mammogram, some other link in the delivery chain is being affected. Take, for example, the patient’s interaction with the technologist. If a file jacket is being utilized as the “infrastructure,” the technologist will check the enclosed films for positioning and exposure information, and review the patient history or even write down new patient information in the file. In an analog system, reviewing the films is necessary, but in a digital system, this review can be done on an acquisition workstation, if the appropriate acquisition workstation is available. That leaves the patient history and, if the center has implemented a comprehensive mammography reporting system, this information should also be available to the technologist without resorting to a physical file.

Unfortunately, unless the facility has digitized the prior film images, the physical file will still need to travel within the facility for several years after the facility acquires digital mammography technology. First, of course, is the use by the technologist, but after she has completed the mammogram, the film priors must be hung on an alternator to allow comparison with current images by the interpreting physician. Thus, the file will first need to be moved to the mammography acquisition unit with the technologist, then to the reading room for hanging the prior images. Unless the center is utilizing a mammography reporting system, the interpreting physician will also have to rely on the file for patient historical data, and prior reports that may form an important part of the interpretation. After interpretation, the transcribed report will need to be placed in the file, along with the films, and the file returned to storage. We take this process for granted because it is all that we have known but, with digital mammography and adjunct technologies, it is no longer necessary.

The challenges of reviewing prior analog exams on conventional alternators, along with current digital images on softcopy workstations, has been exhaustively discussed in the literature and will not be covered in any detail here. It is important to note, however, that the following objections have been raised, and most commentators believe that these issues contribute to increased time to interpret a digital mammogram.

From a more anecdotal, but nonetheless important, perspective, breast radiologists who review current images on softcopy workstations, while comparing priors that are hung on an alternator, report their belief that they are able to “equalize” the images in their minds to the extent that they do not think that the clinical value of the comparative process is diminished. When further questioned, however, virtually all of these radiologists complain of feeling more tired after interpreting a large number of digital mammograms, than after interpreting a comparable number of film screen studies. The time penalty ascribed to the awkward comparative process is generally considered to be about one-third of the total time of the interpretation.b

In any event, until the physical file is eliminated, it rules the workflow of mammography. Although digital mammography presents the center with an opportunity to rethink the entire delivery system, it has most often been implemented in the United States as a modality upgrade only. The balance of this paper will address the requisites for the strategic implementation of digital mammography and its potential to integrate all of breast imaging.

Physicians moving from analog to digital mammography find that reviewing digital mammograms is quite different than reviewing their analog counterparts, and that it takes some time to become accustomed to the new paradigm. Beyond the simple fact that there are several options for enhancement of the images (eg, window and level, contrast enhancement, etc.), the images just look different, with characteristics all their own. Higher resolution images also require panning to view the image at full resolution. The adjustment period will vary between physicians, however, it is not unusual to expect 3 to 6 months for experienced breast radiologists who interpret mammograms daily to reach full efficiency interpreting in the digital environment. For physicians who read at lower volumes or who rotate between general radiology tasks and mammography, the time required to become fully competent at digital interpretation will be extended.

In discussions with interpreting physicians who are currently using digital mammography, it is common to hear generalized complaints involving the workstation. This focus on the workstation can be attributed to two factors:

It is very important to view the workstation in its proper perspective. The workstation is the source of functionality, but it can achieve no better results than are allowed by the infrastructure that provides its support. For this reason, the author strongly recommends that infrastructure be the first consideration in planning for a digital mammography implementation.

Infrastructure First! 

Infrastructure can be a rather intimidating word, but it shouldn’t be, as we live with, and depend on, infrastructure on a daily basis. In film screen mammography, the infrastructure includes such things as the processor, alternator, and film library. It also includes the mammography reporting system, if one is being utilized. Importantly, because of the tangible physical nature of the infrastructure for film screen mammography, it includes one other element that we often do not consider, but without which the system of film screen mammography would not work. That element is people. By this I do not mean people performing the specific tasks that go into producing and interpreting a mammogram. Rather, I am referring to the people and tasks that are involved in communicating or transporting images and files around within the breast center, hanging those files, making certain that records for one patient are not put in the wrong file jacket, and filing the jacket in the proper location.

Basic digital mammography throughput, up to the point where the physician has a set of images to review, has the potential to be significantly faster than its analog counterpart. This is true even for the CR-M (computed radiography) version of digital mammography that at the time of this writing has not yet been approved by the FDA. The overall time to position and image the patient, and to perform quality assurance on the images, is dramatically reduced from that required in a film-based system because of the elimination of the time for the technologist to leave the room and to process the images. The author has previously reported on a Swedish breast center that has been able, through the use of digital mammography and an electronic patient record, to reduce the time for a screening mammogram to 5 to 6 minutes, and to increase the number of screening mammograms delivered per 8-hour day to over 70—for one digital mammography unit.3 This level of productivity is achieved by a single technologist. The technologist’s tasks, however, are limited to the patient care and technology functions for which she is trained. She has access to the complete electronic historical records for patients, and is able to confirm and correct elements of those histories with each patient, but she never touches a “file,” as we conventionally use that term. This information, as well as both current and prior images, is made available to both the technologists and the physicians utilizing picture archiving and communication system (PACS) technology.

Commit to Digital Implementation 

Following the commitment to begin with an infrastructure platform, the decision of how to implement digital mammography must be made. The author recommends that breast centers make a total commitment to digital, rather than a phased or piecemeal adoption of digital mammography over time.

The principle business purpose of moving to digital mammography is in its potential to redefine the workflow paradigm, and to open up opportunities in information management that have not heretofore been available to breast centers. To accomplish this task, the change to digital mammography must be complete, with no residual analog mammography. To retain any film screen mammography capacity will require the maintenance of two separate image infrastructures. Not only is the latter course very expensive from a financial standpoint, it is also very costly in terms of making the cultural changes that are requisite to the implementation of digital mammography. After all factors are examined, there are simply no compelling reasons to maintain parallel analog and digital systems.

The second tier of commitment to digital mammography is integration of the mammography information system with the PACS. This element overlaps the infrastructure aspects of a strategic implementation, but doesn’t come into play until the physicians are interpreting current digital mammograms. It should be obvious, however, that the filmless/paperless “loop” cannot be completed until all information necessary to the patient encounter is available electronically to whomever needs the information—where it is needed, when it is needed, and how it is needed. The mammography information system is a key element in the breast-imaging delivery system, and the more thoroughly integrated they are into the PACS, the more productive the delivery system can become. As mentioned previously, every second of interpretive time per mammogram carries a large annual economic impact.

Prepare for Digital 

Infrastructure decisions (PACS and CAD) and the purchase of the actual digital mammography acquisition technology are relatively easy to make. Unfortunately, the making of these decisions, and even wise purchasing, will not guarantee success. Infrastructure facilitates enhanced workflow, but it does not create enhanced workflow. People use infrastructure to improve workflow and it is thus the human element that is the final determinant of success in the strategic implementation of digital mammography.

With digital mammography, we are asking highly trained clinical personnel to change the way they have been delivering and interpreting mammograms for years. Change is not commonly perceived as a friend to clinical medicine by those charged with the delivery of that medicine. This reluctance to change is normal, predictable, and it is an integral part of what makes mammography the important screening and diagnostic tool that it has become. Good mammography has value because it is reproducible. The breast is positioned in standard views, and exposure must be carefully controlled to optimize the presentation for interpretation. There is very little room for individual variation and, in fact, much of the MQSA regulatory process has been instituted to reduce variability.

Against the context of the implementation of digital mammography, it must be emphasized that there are no changes that will reduce quality—it is only the idea of change that creates a barrier. Workflow will change substantially for technologists, with some changes that fairly radically affect the way they deliver care and interact both with the patient and with their peers. For example, when the need to leave the mammography room to process the mammograms is eliminated, virtually all of the technologist’s time is involved in direct contact with the patient. She will no longer have the informal “breaks” during processing when she can interact with her fellow technologists, and this can add to job stress if alternatives are not found. In a slightly different vein, the digital breast radiologist has many more tools to use for interpretation, and will need to digitally “work up” many mammograms before reporting his or her interpretation. Beyond interpretation, recording findings and reporting results will also change, as facilitated by the mammography information system and electronic medical record. These are changes to both the routine and to the way that technologists and radiologists will perform their respective tasks.

The responsibility of administration during the preparation phase is to examine every task and every process from the perspective of how each will need to be accomplished in the optimal digital environment. This is not the time to compromise, but it is the time to collaborate. Internal champions need to be identified and developed, and the team needs to be lead toward acceptance of the changes that will be required. This process is not easy. Experience shows that technologists who should be excited to be involved with cutting edge technology, often are afraid of the change that will be required of them. This typically appears as passive–aggressive conduct, with involved parties offering vocal support, but also raising a continual stream of small barriers, often couched in terms of “patient care.” Remember that this behavior is a reaction to change and the technologist’s conditioning to oppose change. Rather than attempt to “mandate” or “enforce” the necessary changes that need to occur, technologists must be empowered to both embrace change, and to participate in the development of new processes. This is not an easy task, and it is one that requires good leadership.

One of the prerequisites of a strategic digital mammography implementation is that all patient information must be in the mammography information system before the patient walks into the clinical area of the breast center. Technologists can confirm critical data with the patient, but they should not be required to serve as clerks, charged with the responsibility of entering the patient history into the system or completing in incomplete form. Preparation for workflow paradigm change should begin long before implementation, with workflow plan alterations incorporated that ensure the entry of relevant information at the time of registration or immediately on patient check-in, depending on local circumstances and staffing. Again, many technologists have been performing the data entry function and consider it as a part of patient care. It must be stressed to the technical staff that their review of patient data will still be important, but that they will just not be responsible for the entry of that data. Review is a clinical function; data entry is a clerical function. Histories can be scanned into the PACS, or entered by a clerical staff member, but this function should not be a part of the technologist’s job description.

The author has found that an important ingredient in technologist acceptance of workflow change is the creation of a collaborative environment for that change, and administrative recognition of the important contribution that the technical staff will make to the success of the new technology. Much of the groundwork for this can be done well in advance of implementation, and can, in part, take the form of enhanced education for technologists. An example can be found in the fact that, with digital mammography and automatic exposure control, there is less emphasis on exposure, but this relief only shifts the emphasis to positioning. Because the quality assurance feedback is almost instantaneous, physicians become (justifiably) less forgiving of positioning that is less than optimal. Institutions or practices implementing digital mammography are strongly encouraged to retain the services of one of the recognized experts in positioning for an in-house positioning seminar and workshop for all technologists. Activities such as this go a long way toward establishing institutional or program commitment to, and recommendation of, the fact that the technologist is an honored and respected member of the delivery team. The positioning seminar will also have the important affect of reducing unnecessary recalls and otherwise improving quality.

The implementation of digital mammography tends to emphasize the difference between general radiologists who also read mammograms, and dedicated breast radiologists. The interpretive tools that are available with digital mammography provide more interpretive power for the breast radiologist, but they also involve more time, which is compounded in the case of its use by radiologists who have not developed confidence in their ability to find cancers or to recognize the normal breast. This comment is not meant in any way to be disrespectful. It is simply factual. Digital mammography can dramatically increase throughput, but system throughput is ultimately dependent on the ability of the radiologist to interpret and report mammograms efficiently and accurately.

Centers implementing digital mammography should carefully review mammography audit data for the individual radiologists interpreting mammograms. While the author does not believe that recall rates should be viewed in isolation (recall rates should be reviewed in the context of cancer detection rate, false negative rate, and percentage of minimal cancers), recall rates impact workflow significantly. The impact of excessive recall rates (>8–12% with a low cancer detection rate) is to move patients unnecessarily into diagnostic mammography, which is very costly because of the requirement of direct supervision, as well as a longer required scheduling block, even with digital mammography. It is, therefore, very important that all screening mammograms be interpreted by highly qualified breast radiologists, to meet both clinical and economic objectives.

Appropriate use of the mammography audit can be very important to improvement of the cancer detection rate, and the overall performance of the radiologists interpreting mammograms.5, 6 One must always remember that the final arbiter of quality of the mammogram is the interpreting physician. Digital mammography is not “business as usual” and the author strongly recommends that an integral part of the preparation for the implementation of digital mammography is the attendance of interpreting physicians at CME courses that feature a discussion of digital mammography as well as interactive interpretive skills development. There are currently no interactive digital interpretive skills seminars, but analog programs of this type will be beneficial as they reinforce the physician’s skills through the structured examination of a large number of mammograms with a variety of malignant and benign lesions.c Confidence in reporting benign conditions is particularly important, as digital mammography will provide more tools (eg, window and leveling, digital magnification, etc.) that can help the radiologist refine the screening interpretation. Ultimately, this should translate into decreased recall rates. Continuing to prepare for digital mammography, radiologists need to become very familiar with the mammography information system, and facile with direct entry of information into the system (touch, keyboard, or voice). Interpretive throughput cannot be optimized if a data entry specialist or medical assistant is required to perform the data entry function from a paper document. This lesson has been proven time and time again across the nation as health care systems struggle with implementation of the electronic medical record.

It should be clear from this section that there is a great deal of work that can and needs to be done before the first digital mammogram is delivered to a patient. This paper can only touch on preparation work because of the differences in existing situations within various breast centers. The key point that the author would like to leave the reader from this section is that assessment, planning, and preparation with respect to workflow and the human element are crucial to a successful strategic implementation of digital mammography.

Select the Digital Mammography Technology 

Some readers may be wondering why the choice of acquisition technology has not been addressed previously in this paper. The reason is really quite simple: this is the point in the process where the choice of digital technology fits in the strategic process of implementation. Develop the infrastructure, begin developing a digital library, prepare the staff, and then determine which vendor’s digital mammography technology will best suit the needs of the breast center.

Other than the obvious logic that the foundation should be built first, the pace of development of digital mammography has been quite rapid, and continues at a significant pace. The capital investment required for digital acquisition modalities, as well as the other required system elements, is substantial, and breast centers should delay the choice of acquisition technology until they are ready for full implementation.

A sampling of upcoming digital mammography technology developments follows. This article is equally applicable to these new technologies.

It Is Not Perfect! 

Those contemplating purchase of digital mammography should understand that it is not perfect. When the FDA was initially presented with digital mammography systems for approval, it approved the entire system, rather than the individual components (eg, acquisition, workstation, storage and image management). Consequently, no provision was made, for example, for the uniform display of images from different manufacturers on a single workstation, or the reproducibility of digital annotations that are made by the physician on the image. For this reason, the FDA requires that digital mammography sites have the capability to print mammography images from digital for patients or others who might require access to the images.

Unfortunately, printed images do not help the digital breast imager from a workflow perspective and, while these images can be re-digitized into the PACS for softcopy presentation, this is considered a retrograde process with inevitable deterioration of the image. The images can also be burned into CD or DVD media, but this does not answer the question that is presented by the lack of uniform presentation protocols.

Help is on the way. In 2004, Margarita Zuley, MD, and her colleagues at the Elizabeth Wende Breast Clinic (EWBC) in Rochester, New York, one of the first centers with access to all of the digital mammography equipment, found that cross-platform interpretation was difficult at best, and nearly impossible at times. The EWBC worked closely with the development of a PACS-based workstation solution, but found that many of the challenges were based on a lack of uniformity in communication and other protocols governing the images and how they were handled by the information infrastructure. With the realization that the challenges that they were facing were universal and required a universal solution, the EWBC enlisted the help of the IHE (Integrating the Healthcare Enterprise) initiative.8 As this paper is being written, several IHE standards have been promulgated in draft form, and are being circulated for discussion. Testing and final adoption of these standards is expected within the next year.

The IHE develops integration standards for utilization by manufacturers in the development of imaging technologies and infrastructure. These standards are voluntary, and they become effective mandates to industry only when purchasers of equipment are willing to include IHE compliance in their requests for proposals when contemplating purchase.9 Considerable consumer pressure is building for solutions, and it is anticipated that the IHE efforts will be successful.

But… We have Already Implemented Digital Mammography 

The ideas that have been presented in this paper carry the same value for facilities that have already acquired digital mammography—just more urgency. It is never too late to assess, plan, and reengineer existing systems. There are also some immediate “fixes” that can be implemented to achieve more rapid results than may be required for programs that have the luxury of time to stage an implementation.

Take, for example, the problem of interpretation using film prior and softcopy current mammograms. With the appropriate infrastructure, a worklist can be developed from the schedule, and the prior film studies digitized into the PACS from that list 1 to 2 days before the date of the patient’s visit. Although this process carries some additional expense (the cost of labor to pull and process the images), that cost will be far less than the cost of the additional time required by an ineffective interpretive process and the throughput impact of needing to keep the file active and traveling through the breast center.

Conclusions 

Changing from analog to digital mammography is a much more challenging process than a simple modality change. The decision to implement digital mammography should be viewed as a strategic opportunity to create a far more effective clinical and economic delivery system for breast-imaging. Important choices that will be required span infrastructure, acquisition units, and workstations, but to affect change in process also requires a significant investment of time and effort in developing the human resources involved in the breast health delivery system. It is only through a program that addresses all of these facets that clinical excellence and economic success can both be attained.