Switch to digital mammography demands intensive planning

Congratulations! You've just received budget approval for full-field digital mammography. But before you begin the process of purchasing and installing your new system, you have a bit of preliminary work to do. You have likely already installed PACS and perhaps have integrated with a radiology information system, reporting system, and electronic medical record.

Although many similarities in workflow exist between digital mammography and other soft-copy interpretation, issues unique to mammography can have a significant effect on efficiency. Consider that 30 seconds wasted in reading a 10-minute CT scan equates to 5% decreased efficiency, while the same time used in screening mammography can mean up to 50% time lost. It is therefore vital to spend adequate energy and resources planning and implementing a well-designed electronic workspace. Several issues must be addressed:

• Archiving. Will you use your current storage system or purchase a separate archive from your FFDM vendor? Digital file sizes are quite large, varying between 35 and 200 MB per uncompressed four-image study. Future applications such as tomosynthesis will create even more data to store. Will you be storing only the processed study (often called "for presentation images") or the raw data as well ("for processing images")? Will you be storing your computer-assisted diagnosis results? Where will the studies be acquired? Where will they be interpreted?

Storage and transmission requirements will vary depending on the answers to these questions. Archive configuration should be customized with input from internal IT support and your vendor. Some third-party vendors now offer to meet storage needs in an offsite archive, an intriguing option, especially for small operations. Whatever the mammography archive configuration, you also want access to all of your PACS images from your mammography workstation.

• Hardware. Archiving? Big deal! Who cares where or how the images are stored as long as they get to you quickly and in the proper format. What likely concerns you more is the efficiency of image interpretation, and that, as in PACS, is essentially a workstation issue.

Workstation hardware configuration is not a complex decision. If you are supplying your own hardware, you must meet or exceed the vendor's recommendations to have any standing in a performance dispute. Monitor choice is also relatively easy. No direct FDA approval is necessary for FFDM monitors, but the FDA requires that any monitor comply with a quality assurance program equal to that recommended by the FFDM manufacturer. Currently, all manufacturers recommend dual 5-megapixel gray-scale monitors. A third color monitor is becoming standard with newer workstations, and this should prove helpful in displaying ultrasound and MR images as well as reporting software and other textual data.

Workstation software is another issue, however. Experience in PACS design should allow mammography workstation vendors to build on prior knowledge, yet this has not always been the case. Early FFDM vendors have not lived up to the expectations that their older PACS cousins have created. Manufacturers and designers are just now realizing how important some conventional PACS features are to radiologists; current and future software versions are beginning to reflect a much improved design.

• Work list functionality. Even in PACS, work list functionality has not always been ideal. The rudimentary requirements of opening one or multiple cases, indicating more than one interpreter, automatically opening the next available case after closure of the last one, and sorting and filtering are a given. Even more vital in screening mammography than in conventional image interpretation is the ability to remove a case from the work list without marking it read. In most screening workflow environments, a significant number of patients must return for additional views. These cases shouldn't be left on the work list for a colleague to stumble across, yet neither should they be marked read.

Such cases must be easily assigned to a temporary work list and removed from the "to be read" list. When the additional images are acquired, a study can be reassigned to the work list. Compare this with your analog workflow. You would not simply leave the folder of a patient awaiting additional views in the original pile. You would use a separate bin for these cases until they are completed.

Hanging protocols. Like work list functionality, hanging protocols are critical and complex in mammography. DICOM standards now define basic parameters such as proper sizing, orientation, and gray-scale presentation. But screening mammograms typically require a progression of hanging protocols, often called reading protocols. These should allow you to view images as you are used to in the analog world: perhaps first an overview of all four images, followed by individual inspection of each using side-by-side or mirror image comparison with any of the old studies. Hanging protocols must accommodate nonstandard studies and additional views. Your workstation should be able to display images from any vendor in the proper configuration, although few meet this requirement at anything but a basic level.

Graphical user interface. In the analog world, you have a set of tools specific to mammography arranged as you wish on the desktop. You can use a tool (wax marker, for example) once or more than once at your preference. You can pick up a magnifying glass and view (zoom) one area, then easily move to another (pan).

Functionality should be no different in the digital world. Tools cannot require an icon click or pulldown menu, thus forcing you to remove your eyes (and concentration) from the images. Pan and zoom functions should be seamlessly linked, as these are often used together. Annotations added to the images must be saved at the user's preference and should easily toggle on or off during viewing.

This is an absolute requirement in communicating to the technologist what area of the breast needs additional views. All computer-aided detection annotations must have this feature as well.

• Integration. Optimal screening workflow requires viewing case after case without distraction. The capacity to easily retrieve pertinent patient information, such as family history or prior biopsy results, requires integration of workstation software with other information systems such as RIS and EMR. At the very least, you will need a RIS interface if you intend to read from your workstation's electronic work list. Integration with the reporting system would be helpful and could eliminate the need for data entry. Integration of this sort can put you within striking distance of paperless workflow, which may improve efficiency further.

• User interface devices. Virtually all current PACS workstations were designed around the conventional keyboard and mouse, simply because that is what was available. Digital mammography has an advantage here. Many vendors include their own customized mechanical user interface device with workstation purchase. While these devices are not yet perfected, they are a big step forward from the anachronism of the QWERTY keyboard and mouse.

• Ergonomic workspace. An efficient digital workflow should minimize unnecessary physical and mental activity. With the long hours spent in screening mammography, proper ergonomic positioning of chair and worktable is perhaps more important than with other imaging. The decreased luminance of soft-copy reading also requires careful attention to room lighting and glare reduction. This is particularly true during the transition from film to soft copy.

Some radiologists are choosing to digitize old studies for comparison. This can be costly and carries other potential disadvantages. For those who wish to compare current digital images with old films, careful attention to the physical workstation arrangements during transition is vital. Remember that, unlike CT comparison requirements, which decrease exponentially and fall off rapidly at six months, screening mammograms are typically compared with films dating back two years or more.

Make sure you adequately prepare for FFDM purchase and installation. Use PACS knowledge and experience in evaluating workflow but also be aware of requirements unique to digital mammography. You now, fortunately, have some help. The new 2007 Integrating the Healthcare Enterprise mammography profile works with other existing profiles to address most of the above workflow issues.

The best software design in the world will not be helpful if your own analog configuration is unsatisfactory. Converting to a filmless and paperless environment will quickly expose any flaws in your current workflow. Be prepared to make the necessary corrections and adjustments during FFDM implementation.

Dr. Weiss is clinical head of imaging informatics at Geisinger Medical Center in Geisinger, PA.