If the acceptance of picture archiving and communications systems (PACS) as the modus operandi of the next century is inevitable, conventional wisdom suggests that hospitals not already initiating a planning task force ought to do so now. A broad array of hardware, software, and computer options already exists. Numerous network solutions are available as well. Experts suggest that the key question is not if a PACS network of some type should be implemented, but what kind, when, and how.

Should PACS implementation occur in carefully planned and executed phases? Should hospitals begin with a single-modality mini-PACS or should an entire multifaceted network be planned, acquired, and installed in one stroke?

The availability of funds and in-house technical resources and support will define the available options for many hospitals. Few institutions can afford a one-shot, multimillion-dollar investment. Even those that can, however, urge caution because PACS technologies are still evolving and many equipment interface and networking issues have yet to be resolved. Furthermore, no one manufacturer or systems integrator has all the answers, experts say. Many advocate an open-systems architecture approach that enables facilities to select off-the-shelf PACS system components for their unique situations and needs.


Even defining the nature and role of PACS varies among hospital constituencies, as Jim Grosskopf discovered in 1992 when he became administrative director of medical imaging and cardiovascular services at John Muir Medical Center (JMMC-Walnut Creek, Calif). As Grosskopf inventoried his aging equipment and diagnostic and interventional facilities that year, he realized that he needed something far more comprehensive and complex than a simple 5-year equipment replacement plan.

“Instead of merely budgeting for equipment replacement, we asked ourselves where we wanted to be in 5 to 10 years in terms of imaging and how we wanted to get there,” Grosskopf explains. Deciding that various internal and external resources would be required to produce a plan containing viable working conclusions, he assembled a multidisciplinary team. Members included the chairman of the medical imaging department, the administrative director of information technology systems (ITS), the radiology information systems coordinator, the administrative director of the angiography/cardiac catheterization laboratory, and the vice president of clinical and administrative services.

“The subject was complex and we recognized that even as a group, we needed to know more in order to proceed. We agreed to hire a consultant with the requisite background and experience in PACS technology and planning to bolster the team’s experience and to help us draft a plan,” Grosskopf explains.

Six months later, the first 5-year technology replacement plan had evolved into the Electronic Image and Information Management Strategic Plan, with the ultimate goal of taking JMMC’s clinical imaging services into a world without film.

The strategic plan covered 1993-1997. It included an overview of existing operations; projected data volumes and network traffic; diagrams to illustrate imaging equipment, PACS components, and related peripherals; a series of annotated block diagrams demonstrating the team’s multiphase PACS strategy; budget projections; and a cost/benefit budget justification model. Grosskopf updated the 5-year plan in 1995-1996 and is currently writing a third edition in light of JMMC’s recent merger with nearby Mt Diablo Hospital.

“Although JMMC and Mt Diablo are essentially writing separate plans,” he says, “we are developing a network plan to tie our imaging departments together because of our need to share images. Later, we plan to share a common hospital information and radiology information system.”

To achieve a fully filmless operation as described in the strategic plan, a basic network was needed to move images and data throughout the facility. Because of concerns with soft-copy reading in 1993, the group decided that the first phase of PACS implementation would be the creation of a print network to support digital and video image acquisition while continuing the use of film until everyone was satisfied with monitor and workstation image resolution.

“Our ITS department favored an open architecture solution for the print network that would also enable us to comply with DICOM [digital imaging and communications in medicine] 3.0 and TCP/IP [Transmission Control Protocol/Internet Protocol] standards,” Grosskopf says. “In addition to open architecture, our vendor provided the equipment interface technology and expertise needed to link our various scanning and imaging equipment to our four laser printers. The objective was to help us reduce our reliance on film and chemistry, transitioning us into a filmless environment,” he explains.

The heart of the JMMC print network is a hospital-wide 100-base T fiber backbone ATM network. The print network accommodated both digital and video print interfaces, enabling the connected imaging modalities to print to any of the four laser printers, providing a built-in backup system.

“By 1994, the print network was in place and we proceeded with the second phase of our 5-year PACS implementation plan — the connection of imaging modalities,” Grosskopf says. “Although the plan called for the connection of only one ultrasound scanner initially, three went online by the end of the first week. We added a fourth ultrasound machine in 1998. The next step entailed the connection of three gamma cameras from nuclear medicine and some of our digital radiography/fluoroscopy rooms,” he notes, “followed by angiography and the cardiac catheterization laboratory. Our two CT scanners and MRI system were the last to go online.”

The next phase in the JMMC multiyear implementation plan calls for the acquisition of a computed radiography (CR) system and the beginning of an archive.

“We moved successfully through our first 5-year PACS implementation plan, and as of now, we are 2.5 to 3 years into the updated plan that we launched in 1996,” Grosskopf notes. “Last year we sought funding for the first phase of acquiring a CR system and basic elements of an archive. We were funded for the parts requested and are returning to the committee again this year as we pitch our year 2000 capital budget request.”

Despite some prior funding roadblocks and added glitches, Grosskopf and his co-workers predict that a CR system and the foundation of a PACS archive will be in place late this year or early next year. Information from nine PACS archive vendors is being analyzed and requests for proposal (RFPs) will be sent to three finalists. Grosskopf expects a CR system vendor to be selected by the end of June, with purchase and installation in the neonatal intensive care unit and emergency department shortly thereafter.

“We are beginning there because patients typically do not have relevant prior studies, so we can immediately begin to put those studies into digital format,” Grosskopf explains. “The review stations will be ready as part of the installation.”

Doing the homework and developing a detailed implementation plan with the help of expert in-house and external resource teams is clearly an ideal way to phase in a PACS network. Even the most sophisticated plans, however, can experience setbacks due to unforeseen events.


As often happens, when renovation or new facility construction is contemplated, thoughts of increasing revenues and enhanced delivery of care lead to strategic planning to accommodate the new technologies that make them possible. The launch of strategic planning for a PACS program in 1994 preceded the construction of a new outpatient center in 1996, according to Jennifer Michael, assistant director of image information management services for the University of Chicago Hospitals, and Fred M. Behlen, PhD, assistant professor of radiology at the University of Chicago.

“Working with a team of diverse in-house experts, our methodology was based on what we called an infrastructure approach, in which we focused more of the initial resources on building the information infrastructure — the ability to capture, store, and move data — with deployment of workstations occurring later,” Behlen explains. “It quickly became clear to us, based on the responses to our RFPs, that we were not going to construct a filmless outpatient clinic in such a short time frame unless we were willing to spend $12-$13 million.”

Rather than commit to such a large one-time investment with its concomitant uncertainties, Michael and Behlen opted to implement their PACS network in carefully planned phases.

“The word phased has two meanings to us,” Michael explains. “One definition relates to the phase-in of a filmless environment and the kind of team planning that is required. The second definition relates to the selection and installation of different PACS systems for various modalities or different types of radiology. Thus, we formed two teams: a PACS committee comprised of radiology physicians and technical staff, the hospital’s information systems people, and referring physicians; and a similarly constituted group that also included obstetrics-gynecology staff,” she adds. “Our rationale for the second team was based on the decision to implement an ultrasound mini-PACS system that would serve the Department of Obstetrics-Gynecology as well as Radiology. The resulting ultrasound mini-PACS provided hardware redundancy and backup while simultaneously meeting each group’s very different imaging needs.”

The PACS team issued an RFP in 1994 for an archiving system to support digital modalities including CT, MRI, and chest x-ray. Companies responding to the RFP fell in two categories: equipment manufacturers with limited systems integration experience and systems integration firms with limited medical imaging experience. In the belief that the University of Chicago Hospitals already possessed the necessary medical imaging expertise, a contract was signed with Lockheed-Martin in August 1995 for a system using a DICOM archive approach that could integrate with the existing radiology information system.

No one could have foreseen what happened next. “We awarded the deal to a Martin-Marietta unit,” Behlen explains. “Martin-Marietta then merged with Lockheed, which then acquired [another PACS vendor], and then closed the operation from which we had purchased our system in order to support [the other PACS] rather than the Martin-Marietta system.”

Despite the setback, Michael and Behlen proceeded with the respective PACS implementation plans. An ultrasound mini-PACS went online in 1998 and is currently 85% functional in its ability to acquire and display images. Although some studies are reviewed on workstations, for internal reasons film is still printed, Michael says.

The archive component of the Lockheed-Martin system remains operational, capturing and storing CT, MRI, and digital chest images. As the PACS implementation progresses, the two anticipate archiving lossy wavelet-compressed data on RAIDs (redundant arrays of inexpensive disks). Original lossless image data will be stored on high-density tape in a robotic library. Given current pricing, Behlen says, the hospital can afford to keep several years of wavelet-compressed data online in RAIDs. By purchasing additional disk space each year to expand the archive while maintaining several years of data online, the expected number of retrievals from the longer-term tape-based archive will be small.

“We have a little more than 2 years of data in that archive,” Michael says. “Because we have only one diagnostic review station, we are doing minimal soft-copy reading of chest images. Our radiologists typically use it when they cannot locate a film that must be reported. Although its use has yet to become a routine part of their daily operation, it does assist them in maintaining quality assurance in terms of unreported studies.”

Behlen points out that a CR unit and image distribution system have been installed to serve the institution’s seven intensive care units (ICUs). “We print hard copy in the Department of Radiology and have equipped each of the ICUs with a dual-screen workstation for soft-copy viewing,” he explains.

Michael and Behlen believe that their phased approach to PACS implementation is gradually transitioning the institution toward the goal of filmless departments.

The PACS implementation at the University of Chicago Hospitals has also progressed in terms of image distribution with a Web-based image server, Michael says. Workstations equipped with a plug-in are currently in use in the radiology department and a few outlying locations. “I can go to a workstation that has this plug-in and ask for an image,” she notes. “If we have captured it anywhere in our system, it will display it for me.”

Michael and Behlen anticipate that PACS display will be largely Web-based within 3 to 5 years. “It’s kind of a foregone conclusion,” Behlen states. “The question now is how we get from here to there. Right now, we are committed to pushing Web technology as far as it will go, perhaps even including certain aspects of mammography.”

Toward that end, the University of Chicago Hospitals operates an ATM network backbone with 10 and 100 megabit Ethernet? to serve individual workstations and desktop PCs.

Despite its continuing reliance on film, however, the hospital has not bought any dry lasers. While various units have been demonstrated, Michael and Behlen say that acquiring the technology is more of an endgame that could become useful once in-house users and referring physician groups agree that dry production should replace wet film chemistry.

Although the University of Chicago Hospitals is in many ways typical of the phased-in approach to PACS implementation, Michael does see potential advantages to an all-at-once approach for a facility whose key imaging modalities already capture images digitally.

“The kind of equipment you have is key,” she says. “If your equipment can digitally transmit images to an archive, I might recommend trying to do more of the PACS implementation at one time. One caveat: I do not think that a lot of hospitals have all the digital equipment in place that they are going to need.”

Nor are gradual implementations always the easiest, according to Behlen. “Consider an amputation performed in stages,” he suggests. “I believe that a fair amount of shaking out of the technology still needs to be done, but I think we are getting to the point that we can discuss the acquisition of a large system in one chunk.”


If the University of Chicago Hospitals and John Muir Medical Center exemplify teaching and nonteaching proponents of the phased-in approach to PACS implementation, Next Generation Radiology (New York City) represents a unique new model: a single group practice approach.

“We are a phased implementation,” acknowledges David Katz, MD, one of the group’s four partners. “We went into PACS because I was very interested in getting our office linked to several university centers of excellence. To do so, we had to implement a PACS system.”

Next Generation Radiology primarily utilizes experts located at Harvard University/Brigham and Women’s Hospital in Boston, the Hospital of the University of Pennsylvania in Philadelphia, and Wake Forest University in Winston-Salem, NC. The University of California San Francisco/Stanford and Emory University in Atlanta are also part of the novel network.

Because of Next Generation Radiology’s size, Katz and his partners opted for a pragmatic, phased approach to PACS rather than an enormous one-time capital investment. They devised a system that was part teleradiology, part PACS, and part wide area network. The system utilizes lossy wavelet compression and includes DICOM-compatible CT and MRI scanners. It went live in late 1998.

“We routinely send CT and MR images to one of five university partners,” Katz says. “We also installed one dual-monitor workstation in our two office sites. With the system, we can read soft-copy CT, MR, ultrasound, and digital fluoroscopy images. We have one dry laser printer if film is needed for referring physicians. Mammography is still filmed with conventional wet processing.”

Katz points out that the initial phase of PACS implementation did not include a jukebox, so no long-term archive is yet in place. Images are stored on optical disk and literally shelved until needed for comparison with more current examinations.

The result, says Katz, is exceptionally cost-effective. Instead of hiring expensive specialists to work for Next Generation Radiology, the group obtains expert opinions from its university-based consultants. Images are transmitted rapidly and reports are sent equally efficiently, he notes.

“Managed care and third-party payors have no problem with this arrangement because we are still the physicians of record. I am absorbing the cost of the PACS network and the consultation fees from my university colleagues,” he notes. “In essence, my colleagues and I are taking a lower reimbursement in order to offer this service. Payors know that their insureds are receiving the highest quality care.”

Katz indicates that many of his referring physicians also are pleased with the novel consulting arrangement because they know that their patients’ images will be read by leading experts within 24 hours.

“They are buying into the whole concept for good reason,” Katz says. “There is a level of expertise at these universities that simply is not available within the average radiology group. That’s what PACS technology is enabling us to offer.”

At present, the Next Generation Radiology network can only traffic in images. University reports are returned by fax.

“That is going to change soon,” Katz says, “as we enter the next phase. We plan to begin using email and Internet technology to replace fax transmissions between our offices and our university colleagues.”

Within the next few years, Katz and his associates plan to install an archiving system to replace their optical disks. They also intend to connect their referral base to the PACS network both for the upgrade in convenience and speed and as a means of gradually reducing the reliance on film.

“Many of the radiologists [outside the practice] are still reading from film,” Katz explains, “which means that we have to pay someone to deliver laser-printed images. Once we go electronic, the deliveries and occasionally lost prints will all disappear.”

As with Grosskopf, Michael, and Behlen, Katz is by no means certain that his PACS system is producing bottom-line savings.

“If I hadn’t gone into PACS,” he suspects, “I think I would have saved a lot of money. In the long run, I’m sure it will prove more cost-effective. For now, I’m creating a completely new way of doing radiology, creating a unique practice model. I just wish the technology would catch up with what we’re trying to achieve. Three years from now, this will be a lot easier for someone to do than it is for us today.” n


Suggested Reading

Behlen FM, Michael J, Ranahan LA. An infrastructure approach to PACS: designing for the “vendor interface.” In: Lemke HU, Vannier MW, Inamura K, eds. Computer Assisted Radiology and Surgery. Elsevier Science BV; 1997:458-464.

Grosskopf EJ. On the road to PACS — confronting the issues. Radiology Management. September/October 1998:26-33.

Sheldon M. Stern is a freelance medical writer in Irvine, Calif, and a contributing writer for Decisions in Axis Imaging News.