PACS Administrators: Training Tips from the Trenches
Blink Technology Speeds Anomaly Detection
New Service Manages Unstructured Data Up to 500 Terabytes
All in One – Increasing workflow efficiency in mammography

PACS Administrators: Training Tips from the Trenches

By Cat Vasko

PACS administrators face a plethora of challenges when bringing a new system online, but the most daunting challenge of all can be training an array of health care professionals—from technologists to radiologists to referring physicians—to use the system properly. Axis Imaging News asked three PACS administrators from across the country about the hurdles faced during training and how they were overcome.

Shaun Hill, RT(R)
PACS Administrator,
High Point Regional Health System, Piedmont Triad, NC

When we moved from our legacy vendor to our new PACS, our focus became quality-driven. In other words, we were making the technologists more responsible for producing higher-quality work by ensuring that data fields were complete, images were properly oriented, studies were sent under the correct patient ID and accession number, etc. The new PACS uses a quality assurance station, whereas our legacy system did not. This meant there would be a lot more hands involved in the process than before. Before we had a “send it and keep going” workflow. Now, we need to check each study and verify it before changing its status to “reviewed.” This flags the radiologists that studies are ready for them to read.

Since we knew this new model would be challenging and difficult to do alone, we used a “super user” approach, whereby about 15 people had advanced training on the new workflow and were apprised of new challenges and expectations. Each super user was responsible for a pod of about four to five end users. This [pod structure] was intended to give end users a dedicated individual to work with and ensure that they were ready and had a source of help.

This approach worked for the most part, but not without a few problems. First, we found that even though the super users were hand picked by our leadership in radiology for ability, aptitude, and attention to detail, some were weaker than expected. This meant that there were entire pods that were weak, thus requiring extra help from the administrator. We also found that many end users had developed a comfort level in working directly with the administrator and were reluctant to change. We had to begin forcing the issue, because we were getting bogged down with constant questions.

One lesson learned was to provide periodic training for super users, not just a one-time boot camp. I believe we also underestimated how much help people were going to need. Our go-live was really a hot swap. We had one workflow one day, and another workflow the next. People are bound to have a lot of questions and anxiety.

Alana Praytor
Senior Clinical Applications Analyst (PACS),
Norman Regional Hospital, Norman, Okla

The training of all shifts in the diagnostic imaging department was probably our biggest challenge. Not only was it necessary to train on weekends and at night, but we also had to address the differences in workflow that generally occur between these shifts. Weekend and night shifts often have a morphed workflow when compared to the day shifts, and they tend to operate with less staff, often without radiologists in-house. We also found that the staff had a range of computer skills, so we had to offer them the opportunity to learn very basic computer skills. You can’t assume everyone knows what “drag and drop” or “perform a right mouse click” means.

For referring physicians, getting their focused attention long enough to instruct them in a new system was a significant challenge. They never seem to have time to stop and learn something new, and we found that we had to be available whenever and wherever they were agreeable to the training. The physicians were given the opportunity to select times that would best meet their needs and schedules, whether it was after hours, in the office, in the operating room, in the doctor’s dictation room—whatever it took to make them as comfortable as possible. Many physicians find computers intimidating, and they often have a hard time admitting that they need help. Physicians are trained to be confident in their abilities and do not like to appear incapable while trying to use a device their children mastered years ago.

As with the referring physicians, getting a radiologist’s undivided attention was probably the biggest challenge when trying to provide their training. Today’s radiologists are very busy, and you often find yourself training them on an as-needed basis. They tend to learn the minimum that will allow them to use the system and then never expand their horizons beyond that path. If they come across something they need help with, they just pick up the phone and call for assistance.

Initially, we also ran into difficulties getting the radiologists to commit to the challenge of going completely digital—in other words, no films. They would struggle with the new technology as long as their workload was light, but when they found themselves getting behind, they would revert back to the films that were still being produced. Hard copies were within their comfort zone, as they had years of experience hanging films and had become very efficient with that process.

Training physicians requires patience, persistence, and kid gloves. Remember, you are literally changing their world. Change is indeed good, but difficult.

David Brown
Imaging Informatics Manager,
Cedars-Sinai Medical Center,Los Angeles

Our biggest PACS training success story is who we made our departmental PACS training coordinator. There should be one person who is made responsible for the training of radiologists, super users, and attending physicians. Choose this person wisely; he or she will make all the difference in a successful PACS implementation. Additionally, always listen to your training coordinator; that person is on the front lines and understands the needs of the users, particularly the radiologists. On more than one occasion, my training coordinator has been my “reality check” with regard to the readiness of the department to go live with a new software upgrade.

A major PACS training hurdle is getting the users invested in the need to train adequately before the system goes live. The first time that you go live on PACS, it is difficult from the perspective that no one quite knows how they will be impacted. There is more willingness to set aside time for PACS training because of this uncertainty factor, but there also is more anxiety and confusion. You’ll need to focus this energy on training.

The best approach to an initial PACS go-live is to provide no safety cushion. Film alternators need to be replaced by PACS workstations over a weekend. This forces everyone to focus more intently on how their work environment is going to change when PACS is introduced. Buy-in for training will be significantly increased.

Include enough training hours from your vendor during contract negotiation. Applications support will be needed prior to go-live, during go-live, and about a month after go-live. Also, do not train general staff too distantly from the planned go-live. If general staff is trained 2 months ahead, they will have forgotten everything by go-live.

Train your residents and fellows. They will be a valuable resource post go-live to radiologists, who usually do not have as much time to devote to training.

Another PACS training hurdle occurs when performing a significant upgrade to the system. Our upgrade to the “next-generation PACS” included both hardware and software. Our users assumed that they already knew PACS, and they didn’t commit the necessary time for learning the new system. These users discovered that they were not able to accomplish their work efficiently after the go-live. We effectively used our residents and fellows in this scenario as well. We also had adequate vendor applications support.

Get radiologists to trust the system. Encourage them to use tools, such as cine displays, that allow them to read more efficiently. When radiologists first go live on PACS, display protocols usually are set up to emulate film. Images are displayed in tile mode. Radiologists embrace PACS more fully as they become comfortable with scrolling through images interactively. And this encourages them to explore other available tools.

Cat Vasko is associate editor of  Axis Imaging News. For more information, contact .

Blink Technology Speeds Anomaly Detection

A digital blink comparator helps radiologists detect 12% to 15% more lesions than the “series and stack” method

By Matthew J. Kuhn, MD

The legacy of a century-long use of conventional film in radiology is that traditional hanging protocols still form the backbone of PACS, despite the promise of more sophisticated methods of lesion detection and evaluation. For example, astronomers have long realized that the “series and stack” methods of comparing their specialized images are time-consuming and inaccurate. The dwarf planet Pluto, in fact, could not be discovered until a “blink comparator” method was used to compare old images of the sky with newer images in a way that rapidly alternated the old images in front of the new ones.

This technique, now digital, can readily be applied to PACS, allowing for dramatic improvement in accuracy and efficiency. Even subtle changes between two images will easily emerge and immediately stand out to a radiologist using the digital blink comparator technique.

Matthew J. Kuhn, MD

The human eye and brain are more sensitive to motion than to searching for tiny details in a static image. Movement draws our attention to the object in question. A scene with moving components sends a stream of changing information to the brain, keeping it alert and focused.

Currently, there is no identifiable stopping point to a radiologist’s search technique. The use of a digital blink comparator will quickly and accurately show all of the abnormalities on an image at once; doing so will allow the radiologist to rapidly move on to the next set of images.

In addition, with current PACS techniques, the radiologist’s eyes and neck are constantly moving back and forth over a wide area consisting of multiple monitors. This produces repetitive eye and neck strain, which can be physically demanding and painful over the course of a work session. On the contrary, with the digital blink comparator, only one monitor is needed and the reader’s eyes and neck are, therefore, pointed straight ahead and no longer are strained glancing back from one image to another.

RSNA Presentation Time

This technique will be presented at RSNA on November 26, 2006, in room 5404CD at 11:45 am. Also, for more information and a demonstration of the blink comparator, visit

The digital blink comparator method takes a matched pair or stack of old and new medical images and, after displaying them side by side in conventional format, blinks each paired image in front of the other one at a rate of 0.3 per second. The radiologist’s eye easily accommodates for misregistrations. Even chest radiographs taken in inspiration and expiration can be successfully blinked to bring out additional lesions. Mammograms, nuclear medicine studies, bone images, and CT and MR examinations, among others, have been tested and been found applicable to this technique.

The use of a digital blink comparator was first reported to improve diagnostic accuracy in neuroradiology in a scientific presentation at the annual meeting of the American Society of Neuroradiology, Oak Brook, Ill, in April 2006. In a series of 50 pairs of MR and CT images viewed in both stack and blink modes, two blinded readers detected 12.8% and 15.6% more lesions in the blink mode compared to the traditional side-by-side comparison method. At this year’s annual meeting of the Radiological Society of North America (RSNA), Oak Brook, Ill, a report will be presented with an even larger series of patients for whom the digital blink comparator method was used for general diagnostic radiology and found improvements in lesion detection rates of more than 20%. (See “RSNA Presentation Time” above for more information.) The use of blink technology was also found to allow a radiologist to more easily and efficiently detect subtle changes in size and configuration of known lesions over time.

Matthew J. Kuhn, MD, is chief of the Division of Neuroradiology and clinical professor of radiology, neurology, and neurosurgery at Southern Illinois University School of Medicine, Springfield. He also is the developer of the blink radiology imaging technology.

New Service Manages Unstructured Data Up to 500 Terabytes

Aiming to address the growing problem of information life cycle management, EMC Corp, Hopkinton, Mass, has established new software and consulting services. The Infoscape software enables customers to search data and classify it based on business value, manage the retention of unstructured information, and facilitate automatic policy-based data migration across a multi-tiered storage infrastructure. The product also can be purchased with an Information Management Strategy consulting service, designed to help companies discover, classify, and manage enterprise information.

“It takes a look at the unstructured data or file system data that users have in their environment and that reside on file servers or file shares, and it does things like identify that information, where it lives, and who owns it,” explains Sheila Childs, director of software product marketing at EMC. “And then it does things like crack a file open to understand the content, classify it, and then group it into categories that the user finds relevant. And once it’s done that, it allows the user to look at the IT infrastructure that’s in place, the back-end storage devices or the tools for delivering security, and it allows the user to map the information that he or she has discovered and classified to that infrastructure.”

Increased efficiency and lower costs over the life cycle of information are two potential benefits to the software service. “It’s customizable, and one of the things that it will help the health care industry with is being able to identify information at risk related to their patients,” Childs says. “So, if their specific images or documents contain patients’ insurance information or health care history, it allows the user to identify information that contains sensitive data and protect it, quarantine it, move it, whatever the user needs to do.”

The software shouldn’t interfere with intraoperability, according to Childs. “What Infoscape would provide,” she says, “through some APIs [application programming interfaces] that we are writing and some plug-in capabilities, is based on some requirements or policies from other applications. It would search for, identify, classify, and serve up the information required.”

EMC also is offering a consulting service for long-term optimization of Infoscape’s capabilities. “The consulting service is based on our methodology of a plan-build-manage strategy,” Childs explains. “The consultants identify the business problems, the IT users’ problems, how they are going to classify their data, what the services are that they want to deliver. The ‘build’ and the ‘manage’ pieces are really designed to facilitate the deployment, and to stay with the customer long-term to ensure that things are running smoothly. As new expertise is required, having experts or consultants who can come in and assist is probably very long-lived.”

Childs notes that consulting can be useful for companies lacking a sense of how they want to classify and manage information. “This whole idea is relatively new, and I don’t think there are too many facilities that have a handle on what they want to do,” Childs says. “But I think that the product itself will probably never go out of date. The way that we’ve designed it is, as long as there’s information to be managed, there will be a way to manage it.”

—C. Vasko

All in One

Increasing workflow efficiency in mammography helps obtain all of the necessary information the first time around

By Judith Gunn Bronson, MS

The situation is all too common. The radiologist sees something questionable on a mammogram. What is it? Is it significant? From the images, it is not possible to tell. So, the patient must be recalled—at a cost of considerable inconvenience and anxiety for her, and more work for the mammography center—for the acquisition of a different view or another study, such as ultrasonography.

Suppose technologists could look at the films and spot at least some of these problematic areas. And suppose a radiologist could look at the images and decide whether to obtain the additional views or studies before the patient leaves. And suppose this could be done without requiring a radiologist at every screening site.

That is the goal of a telemammography system under development by a team at the University of Pittsburgh and the Magee-Womens Hospital of the University of Pittsburgh Medical Center. A further goal is to achieve the substantial reduction in patient recalls without any need for high-capacity transmission lines, satellite hook-ups, or exotic computer equipment. Early simulations suggest this goal is achievable.1

How It Works

The Pittsburgh system employs readily available computer and imaging equipment at both the remote (screening) and central (review) sites. Equipment at both sites has multi-threading capabilities to make sure that the various tasks in a review are completed in an orderly and timely manner.

When the mammography technologist sees something suspect on a film, the review process starts. The tech digitizes the current and any available prior images with a high-resolution (50-µm) Lumiscan laser digitizer from Kodak’s Health Group, Rochester, NY, and applies a computer-aided detection program. (Presumably, the process will be facilitated when digital mammography equipment is installed.) The images automatically are cropped to remove all areas that are not tissue, then compressed at 75:1 with a wavelet-based JPEG 2000 method and encrypted. Then, using the Scanjet 5470c from Hewlett-Packard, Palo Alto, Calif, relevant documents are scanned.

Next, the technologist uses software to create two messages. First, a generic image depicting both breasts viewed from the front and side is marked with an “X” at the site of the questionable finding. Second, to ensure that all necessary information reaches the radiologist, the technologist enters demographic data and uses pull-down menus to construct a message that specifies:

  • which breast is problematic;
  • which view showed the questionable area;
  • what the technologist saw or felt that is of concern (eg, calcification, asymmetry, or lump);
  • whether there is a change from a prior examination (if available) and, if so, whether the change is slight, moderate, or “remarkable”;
  • whether other changes from a previous examination are apparent; and
  • what additional procedure might be required.

Any additional information the technologist believes to be important can be entered in a free-text window.

This message, any documents, and the images are transmitted to the central site via telephone line or a local area network. At the review site, the images and message are unencrypted, decompressed, and displayed on a specially designed three-monitor workstation, including a 2084 x 2560 monitor, for immediate review by a radiologist. One set of images is displayed with the CAD overlay. The workstation enables manual lookup table adjustments, magnification, quadrant viewing, and multiple display formats to allow the radiologist to view the images in optimal form. If the radiologist agrees that additional workup is required, he or she marks the exact area requiring attention, and this mark appears on the image on the technologist’s computer monitor in near real time.

Although this process might sound time consuming, even in the early trials of the system, the entire procedure—from the beginning of film digitization to display of the images at the review site—routinely required less than 7 minutes. The compression of the images did not reduce their utility, as radiologists were not able to discriminate the compressed from the uncompressed images, even when the two were displayed side by side.

Clinical simulations of more than 2,400 cases suggest that use of this system will reduce patient recall rates by as much as 50% and help in the utilization of technologists as physician extenders. Availability of prior images proved particularly useful in reducing recalls.

Judith Gunn Bronson, MS, is a contributing writer for Axis Imaging News. For more information, contact .


  1. Leader JK, Hakim CM, Ganott MA, et al. A multisite telemammography system for remote management of screening mammography: an assessment of technical, operational, and clinical issues. J Dig Imaging. 2006;19:216–225. Available at: Accessed October 24, 2006.