BEFORE A HOSPITAL can eliminate film-based imaging, there first must be a full distribution of digital imaging capability to those clinicians who historically have been users of filmand the most cost-efficient means of accomplishing this is an enterprise-wide deployment of computed radiography (CR), contends Katherine P. Andriole, PhD, an associate professor in the Department of Radiology at the University of California, San Francisco, and bioengineering at the University of California, Berkeley.

"The majority of the imaging work performed at the typical hospital is projection radiography," she says. "Such a facility will not be able to become completely filmless until it provides a digital answer for projection radiography. CR is an answer, in part, because it so closely mimics the operations of the film-screen environment."

UCSF has been using CR since 1992, when the enterprise first began establishing a filmless environment.

"CR was the first viable option available for going digital with projection radiography," Andriole says. "Our initial application for it was inpatient portable imaging. Most of our inpatients are severely ill, the type one would expect to encounter in an ICU. They aren't able to leave their beds, so we have to rely heavily on portable radiography. However, portable bedside exams are among the most difficult imaging situations to deal with. Consequently, we were experiencing a high incidence of unacceptable images, which required our technologists to go back and reshoot. If we were dealing with, say, a car-crash victim in serious condition and we wanted to find out if the patient had a fracture along the spine, we might be looking at having to do as many as four retakes to obtain just one image of diagnostically acceptable quality, simply because of the sheer difficulty involved in imaging a patient in this particular condition. But within 3 months after installing CR, we saw the number of exposures required for a diagnostic-quality c-spine image decline exponentially."

What helps makes CR superior to traditional analog x-ray is the modality's image detectora photo-stimulable phosphor plate contained within a removable cassette, Andriole explains.

"Once an image is captured on that plate, the cassette is taken to a nearby reader unit where the plate's content is translated via laser into a digital form," she says. "In turn, because it is digital, that image can then be transmitted over networks and viewed on computers or, alternatively, printed to film.

"This technology offers us several significant advantages. First, from a physics standpoint, CR has a very broad dynamic range, such that you can be off a little in your exposure technique without incurring any major consequences. In effect, you can slightly underexpose or overexpose the plate and still receive a diagnostic-quality image. This is possible because image data is processed electronically instead of chemically. Thus, during processing, there is an opportunity to electronically manipulate that image data and improve the output. Such an opportunity does not exist with chemical processing."

A BOON TO RADIOLOGY

CR today is deployed in various key locations around  UCSF's main campus institution600-bed Moffitt-Long Hospital (where the radiology department is headquartered)and at its outpatient-oriented facility two miles awayMt Zion Hospital. The pair are linked via a PACS network, which functions as the channel by which CR images are distributed to users. Since Moffitt-Long is a tertiary care center with a sizable service devoted to treating brain tumor cases, the most heavily utilized modalities are CT, MR, and PET.

"Between the two sites, our department conducts approximately 275,000 imaging examinations annually, about 60% of which are performed using cross-sectional modalities, while the remainder are done with projection radiographyand more than three-quarters of those are CR," says Andriole.

The departmentspecialty-section-based and staffed by 40 clinical faculty plus about 60 residents and fellows and roughly 40 technologistsis 85% filmless, but the goal is to achieve a fully electronic environment, Andriole indicates.

"At this time we continue to print film as needed for use in the operating rooms, none of which has yet been converted to digital," she says. "Also, some of our mammography service remains film-based. For the most part, however, our CR users now view images as soft-copythat is, they look at images on a view-station monitor."

The CR complement at Mt Zion currently consists of one system in the radiology department's diagnostic area and one in an orthopedic surgery clinic. Mt Zion is in the process of creating a cancer treatment center that may or may not be outfitted with a CR of its own, depending on budgetary concerns, Andriole reveals.

At Moffitt-Long, the deployment includes one CR system in the radiology department for imaging of inpatients able to be moved from their beds, one in the emergency department, and one in the 52-bed newborn pediatric intensive care unit.

"We actually have more ICUs than thatthere's one on every floor of the hospital from the sixth to the 15th," says Andriole. "We'd like to one day perhaps have a CR in each of those."

Because CR is a digital modality, it is possible to immediately transmit images to multiple, remote locations. That has been a boon for UCSF, considering how images previously traveled to where they needed to go in the days when everything was on film. Back then, the radiology department located on Moffett-Long's 3rd floorwould send a technologist with a portable radiography machine to whichever ICU the patient was in, take the images, come back downstairs, and then hand the exposed film to a full-time person dedicated to making film copies. The film-copier technician would then run the duplicate back up to that ICU so that clinicians there could view it while a radiologist downstairs read the original. This was a costly process because of the equipment, materials, and labor involved. It also yielded a copy that was not quite as good as the original, although it was still suitable for review purposes.

"Radiologists were often unable to give timely input on x-ray images," Andriole says. "But now, thanks to CR, the images are in digital form and are immediately routed to radiology. That's enabling our radiologists to offer contemporaneous input and thus make more of an impact on the delivery of care. It's also enabling our radiologists to have a stronger relationship with our referring colleagues."

LOOSELY STRUCTURED STRATEGY

In addition to looking at CR images at view stations deployed in the ED and ICUs, about 300 UCSF referring clinicians today are able to access CR images on their office and home desktop PCs via a web server. The number of these physicians is growing at a pace of about 25 new users each month, says Andriole.

"When we got started with all this," she recounts, "we had just one CR system and that was the extent of it for the next 5 years, although in that time we did have a second system for our research area and made it available as needed as a backup. The primary system was deployed in the radiology department. Seemingly, that was a woefully insufficient deployment for a large health care enterprise, but it did make a very large impactwe were doing between 100 and 150 imaging examinations a day.

"Then, in 1998, we swapped out our film darkroom in the ED for space dedicated to CR. Our ED is not the trauma center for the city, so we found that one CR there was sufficient."

Andriole reveals that UCSF had only a loosely structured strategy for deploying CR.

"We had in the beginning a vision of what we felt could be accomplished with CR, but not a specific plan that set target deployment dates and numbers of systems," she says. "The timing of our acquisitions was entirely dependent on the availability of capital in the budgetCR back in 1992 wasn't cheap; a system then cost about $500,000, compared to today's price of around $100,000.

"To justify that kind of capital expenditure, we calculated that we would need to perform about 100 CR examinations each day. This would cover the purchase of the equipment and would reflect the savings in film stock, chemicals for processing, and technologist and radiologist FTEs. What it did not includebecause it would have been impossible to attach a number to itwas the value of improved quality of care resulting from the ability to have immediate access to current and prior images at multiple locations at the same time."

In UCSF's case, infrastructure costs for CR were easier to bear because the hospital was concurrently installing a PACS.

SURPRISED BY ACCEPTANCE

Based on the experiences of UCSF, Andriole says an institution can find success with CR as a distributed radiology solution no matter which way it approaches implementationeither in phases or in one fell swoop, enterprise-widealthough she believes that most facilities perhaps will find it best to start with just one small piece before deploying further.

"In a phased approach, you're able to gain experience with the technology and more easily prove its value to your institution," she says. "That's a good strategy and one that makes the proposal to acquire CR seem less financially burdensome or risky, since significant benefits can accrue from the installation of even a single CR device.

"However, other facilities might find it better to install everything all at oncesort of the Big Bang approach. This is most feasible if a large, enterprise-wide infrastructure already is in place or is being contemplated. The reason is that, once the infrastructure is present, it's not a difficult task to put in CR one at a time, everywhere you believe it should go."

A surprise for UCSF during the initial round of CR deployment had to do with the not-insignificant matter of obtaining buy-in from end users of the modality.

"We were convinced during the planning for CR that our referring clinicians would be the most reluctant to embrace the technology," Andriole says. "But it turned out that a few of our radiologists were the most reluctant. The problem was that they had their particular preferences for how they wanted to do their work, and CR did not initially fit in with those preferences. But eventually, after they received training and began to gain an appreciation for the way CR could make their work routines better, they came around."

In some instances, buy-in was fairly easy to obtain. For example, at least one group of referring clinicianshaving seen the CR review station in another section's ICUapproached the radiology department team responsible for implementing CR and demanded to know how long they would have to wait until receiving a review station of their own.

"They were very eager to start utilizing CR," Andriole says. "The same was true of our radiology technologists and film librariansthey very quickly recognized the value of this technology and how it could save them time and effort."

In recent years, the effort to obtain buy-in from reluctant new users of CR has become less challenging thanks to the numerous peer-reviewed studies that have been published with regard to CR's virtues and applications.

"Having data available to show to potential end users of CR can help turn apprehension into enthusiasm," Andriole says. "But then, too, CR is itself a very exciting technology with a great future ahead of itCR continues to become faster, smaller, and less expensive, all of which serves to make CR more indispensable for hospitals and clinics of all sizes that are trying to transition from film to filmless and to distribute radiology across the enterprise."

Rich Smith is a contributing writer for Decisions in Imaging Economics