FOR CR to be successfully used as a distributed radiology solution, it first is necessary to establish a network of imagersthat is, workstations, review monitors, and wet- or dry-process printers deployed at strategic locations throughout the enterprise. It is this network that makes it possible for an image acquired in, for instance, the intensive care unit to be viewed electronically a moment or two later by a radiologist in another part of the campus, and, by the referring physician in a still farther corner of the institution and simultaneously by the attending clinician back in the ICU, who also might wish to then produce a paper copy for the patient's relatives as a way of updating them on the progress of treatment.

"A well-planned and properly implemented network of CR imagers increases the quality of care in environments that previously were film-dependent," asserts Kiley Rodgers, RT, PACS/IT administrator at Rush-Presbyterian-St Luke's Medical Center in Chicago, which implemented such a network about 3 years ago. "The problem for environments that do not have an imager network is that there can be a wide dispersion of care providers across a physically large campus, yet they all are bound to a single sheet of film that only one of them can see at any one time. Absent an imager network, you'll have situations where, for example, a clinician located on the 12th floor will have to come downstairs to the 1st-floor radiology department and lose perhaps 30 minutes of valuable clinical time looking around for a sheet of film, only to discover that the object of his quest has been misplaced or that someone else in another part of the institution has already checked it out and has it in their possession. Meanwhile, during the entirety of that trip downstairs and the ensuing fruitless search, that physician is kept from providing care to his patients.

"When CR images are instead accessible from anywhere in the institution, this kind of time-loss doesn't happen. That same clinician on the 12th floor doesn't have to come downstairs anymore, unless he or she wants to talk to a radiologist in person."

Rodgers adds that a network of imagers, properly arrayed, should translate into institutional cost savings by reducing the need for film stock and for personnel engaged in film archiving, retrieval, and mounting.

"Economically speaking, a CR imager network accommodates very cost-effective and highly efficient distribution of images across an enterprise and, if desired, outward to other institutions," says Rodgers. "Such a network can be implemented in conjunction with, or as a digital forerunner to, a PACS. An imager network is obviously most at home in the environment of a mid-sized to large institution, but it also can be advantageously employed even in smaller facilities, including orthopedic clinics and freestanding imaging centers."

The cost of installing and operating a network of imagers depends on variables such as the number of nodes and the types of devices attached to them, as well as whether an electronic infrastructure to support the network exists (and, if it exists, to what extent it must be upgraded or expanded), Rodgers indicates.

"Costs will include the purchase of the CR system, the processors, the workstations and review monitors, the printers, the network fiber-optic backbone, the Category V cabling to connect the equipment to that backbone, the hubs, the routers and the switches, plus the costs of manpower to plan, install, and manage the network as well as the costs to train the users," he says. "The average price tag of all this for a medium-sized institution is around $300,000, although it can be significantly higher if the network is poorly conceived and a lot of fixes have to be undertaken in order to get it functioning properly."

KNOW YOUR CUSTOMER

To ensure the network is well designed, Rodgers suggests that there be a clear vision as to what the institution wants and needs to accomplish with it.

"Don't start building the imager network until you've developed guiding ideals for it and have given considerable thought to how those ideals will be turned into reality in other words, you need to know where you're going and then have some sense of how you're going to get there," he says.

In the case of Rush-Presbyterian, the vision revolved around making access to images as fast and convenient as possible. Accordingly, the number of imagers each department or service-line area was assigned depended on how many plain-film radiographic images it was historically producing or utilizing or both, Rodgers tells.

"For our inpatient radiology department, we installed a CR system and linked to it via the network three individual ID stationseach of which has preview capabilityand one QA device," he says. "Then, in orthopedics, where we have an even larger volume of imaging demand, we installed two CR systems and to each of those connected three ID stations that have preview capability and one QA device. Meanwhile, in pediatrics, we installed only one review station and no CR system, since plain-film imaging demand originating in that department traditionally has been comparatively low."

The vision for the CR imager network at Rush-Presbyterian included a configuration scheme that would allow for the continuation of the radiology department's preferred method of reading images, which is organ-based, Rodgers reports.

"We route specific image types to specific imagers," he says. "For example, all chest images automatically route to those reading stations that have been dedicated to chest work, while all bone images are routed to just those reading stations dedicated to bones, and so forth.

"But we also have the ability to make these organ-dedicated imagers function as slaves to a particular nearby CR for purposes of viewing or printing whatever types of images that host CR happens to acquire. We've provided this capability as a safeguard in the event the network fails. Because of it, we're able to take those imagers and press them into the service of whatever needs exist during the time the network is down. We also have the ability to physically tie them into an emergency mini-network employing a portable hub. So, we've got lots of flexibility here."

When transforming the vision for an imager network into an actual blueprint for implementation, Rodgers says it helps to bear in mind that the setup will not be a deployment primarily for the use of the radiology department, but, rather, something that can impact the ability of work to be accomplished across the enterprise.

"Even though an imager network is a radiology-specific system, it's a mistake to think of it as such instead of as a global system," he says. "Thinking of it in terms of being exclusively a radiology-specific system can lull implementers into making the mistake of buying equipment that only meets the needs of radiologists and not the needs of the technologists, the attending clinicians, and others."

For that reason, treat each department and the users therein as a network customer, Rodgers recommends.

"Know who the customer is and what the customer's needs are," he says. "The customer whose needs should be considered as foundational is the technologist." The reason for taking that view, he explains, is that the network represents a set of tools the technologist will use to acquire images, manipulate them, and exercise sufficient quality controls upon them so they can quickly, efficiently, and satisfactorily be passed along to the next group of customers (which can be the radiologist and either or both the attending clinician and the referring physician).

"Without considering the technologist as your base customer, then there will be a vastly greater likelihood that radiologists and other customers will be unhappy with the system, because the technologists will very probably be unable to do all the things necessary to forward a quality product to those other customers," Rodgers says.

Input to help formulate the imager network plan should be solicited from the information systems (IS) department, since those are the professionals in possession of the nuts-and-bolts expertise required to link all the acquisition devices and the imagers into a functioning, stable network, Rodgers advises. Meanwhile, a must-source of input will be the chief financial officer, who is in a position to usher network planners through the intricacies of the budgetary process. Then there is the chief executive officerhis or her input is necessary to supply the big-picture view that can properly shape both the imager network vision and its implementation outlines, Rodgers says.

At some point early in the conceptualization process, someone within the enterprise needs to be designated as the person responsible for advancing the emerging plan and, eventually, overseeing its implementation. Rodgers suggests that the most appropriate individual will be whoever possesses previous experience with imager networks. That most likely will be a PACS administrator.

"If you've already got a PACS administrator on your payroll, then great; if not, then hire oneand do it before you sign off on the plan for implementation," Rodgers urges.

PLAN, THEN REASSESS

After customer needs are identified, the next step is to determine what will be required in the way of equipment, infrastructure, and personnel to properly satisfy those demands. Here, input from vendors can prove invaluable, Rodgers conveys.

"Bring the vendors to the table early on to produce an unbiased analysis of product requirements for your institution's needs," he says. "This perspective can be crucial because vendors will likely notice needs and potential trouble areas you might have overlooked, either as a result of inexperience with this type of deployment or from an overly optimistic view of what you can get by with at a minimum.

"Also ask the vendors for specific recommendations about equipment, deployment, installation, configuration, and connectivity. Each product they recommend will have its pluses and minuses in each of those aspects, so you'll want to insist on having a range of product choices to consider: in this way, you'll be more likely to identify the products best suited for your application and budget."

Once an imager network plan is developed, months may pass before funding requests for it are approved and implementation begins. Thus, it usually is a good idea to review the plan before proceeding with network setup, Rodgers recommends.

"What sounded good 6 months earlier when the plan was written may not be the case today in light of the shifting operational composition of the enterprise or changing financial and market considerations," he explains. "Taking a second look at your plan will help you avoid the trouble and expense of having to reverse-engineer things after implementation has gotten under way."

One such problem that can arise after the plan is developed is the suddenly emergent or belatedly verbalized customer need.

"As you begin to pull together the network, you may find people calling attention to needs that were overlooked," says Rodgers. "You might be able to address the problem with a reallocation of imagers you've already purchased. But this approach requires you to shortchange the needs of others in departments that you're taking imagers away from. That's not a good solution. Neither is telling the customer with the now-identified need that it's not going to be possible this time around to accommodate him or her.

"The way this is going to have to be addressed is by simply acquiring more imager devices. However, this also means you'll end up going back to the decision-makers with a request for more money to be able to make those additional purchases. This is why it's so important to make sure you do thorough planning, so there are no surprises and no need to ask for additional capital. The CFO and the CEO will raise hackles if they begin to get the feeling that this undertaking is suddenly taking a lot of sharp twists and turns."

Another unwelcome surprise is the need to modify the physical plant in order to accommodate the network.

"You might run into things related to environmental parameters or the physical layout of the institution," Rodgers says. "A typical problem here is a room too hot or too small to accommodate a review station. These problems are resolved through discussions with your facilities group. For example, if the room is too hot, you need to discuss with the group ways to reroute venting or perhaps install an air-conditioning duct. But, again, it's best to identify issues such as these before the imager network plan is finalized, since having to address them after the fact only serves to slow down the implementation and requires a return trip to the financial well for extra funding."

Still more unexpected trouble can surface when attempts are made to integrate the devices of the network.

"If your imager network will be tied into PACS and the radiology or hospital information systems, you're in most cases going to be looking at a multi-vendor system with lots of non-DICOM-3 legacy devices, which means an abundance of issues regarding connectivity to be resolved," Rodgers explains. "Bottom line: don't expect your imager network to be plug-and-play. Even if you're not tying into PACS, RIS, and HIS, and the imager network components are all DICOM-3-compliant, you may be surprised by what doesn't work when you connect it all upand it can be things as minor as sending across an examination description, or a referring physician name. To address the issues of connectivity, your best bet is to obtain from your vendors their commitment  work with one another in partnership to overcome whatever obstacles are preventing full integration."

PHASED APPROACH

Rodgers believes it is logistically less complicated and economically less burdensome to implement an imager network in phases rather than in one fell swoop.

"We took a phased approach, starting with inpatient departments, because we felt it would be much more easily managed in small bites," he says. "We made it our goal thatbarring unforeseen problemswe expected to have, first of all, the emergency room fully digitized in about 30 days and the remainder of the departments within 6 months. This would be followed by 3 months of getting all the wrinkles ironed out of the implementation up to that point, and then we would commence the second 6-month phase to build and test the remainder of the network.

"We undertook this implementation by deploying devices and immediately configuring them, which allowed local customers to begin using the devices and gaining at least a measure of satisfaction from them, even though they wouldn't be fully networked for a matter of several more months. We were able to do this because we had in place a basic network infrastructure. If you have that, it only takes a couple of days to drop these devices into place and get them configured."

Rodgers observes that there are not any hard-and-fast rules about how long implementation of an imager network should take.

"It's not really possible to develop a formal timetable that says, OK, on day 5 we'll be installing cabling, and on day 10 we'll install a hub," he says. "Too many problems can crop up and make a shambles of your formal timetable. You can set general target dates for achieving various goals, but don't be alarmed if they're not met. As long as you see yourself moving in a forward direction and progress with implementation and testing is being made, then you should be satisfied that you're doing well.

"Still, expect a bumpy rideimplementations of this kind usually turn out that way. But, when it's completed, your imager network will no doubt quickly prove itself a tremendously worthwhile endeavor."

Rich Smith is a contributing writer for Decisions in Imaging Economics.