A 1979 report, A network of medical workstations for integrated work and picture communication for clinical medicine, is recognized today as one of the first publications to describe the concept of PACS. Given the significance of PACS in modern radiological workflow, the authors of this seminal work might be expected to take pride in their prescience. But Prof. Heinz Lemke, lead author and founder of the influential Computer Assisted Radiology and Surgery congress, claims to be embarrassed about the report.
"When I look back at it now, I get quite a red face," said Lemke, a pro-fessor of computer sciences at the Technical University of Berlin, which issued the publication. "It totally underestimated the complications in-volved. I was expecting to build a PACS within five years, using funds that turned out to be a fraction of what was required."
Despite Lemke's optimism in 1979, the first issue of Diagnostic Imaging International was launched six years later to a readership unfamiliar with the intricacies of PACS. The term "picture archiving and communications system" may have been spoken about in conferences, computer laboratories, and hospital IT departments, but
PACS technology was far from mainstream in the mid-1980s, and its
future in Europe appeared uncertain. Installation of a PACS at St Mary's Hospital in London was planned for 1985, but the funding was withdrawn when government backers realized that essential hardware and software systems were still in development. At about that time, groups in France, Germany, Austria, and Italy began to work with commercial partners on independent PACS projects.
In 1986, the Dutch PACS Project was initiated to evaluate a prototype at Utrecht University Hospital. In the same year, the University Hospital of Brussels inaugurated the world's first multivendor PACS project. No one was sure that PACS would gain sufficient support from governments, healthcare managers, radiologists, and manufacturers for the technology to be truly viable.
"This whole scene was restless, unstable. Nobody really knew whether we were at the crossroads of something significant or simply wandering in the wilderness," Lemke said.
He attributes that uncertainty to rapid develop-ments in information technology during this period. Once desktop PCs appeared, fears of the unknown diminished. Networking projects linking multiple computers began to demonstrate the power of digital communication. One significant hurdle remained, however: the high price of the technology.
"I remember that the PACS being installed in Berlin back in the late 1980s to communicate between two hospitals cost something in the region of DM 9 million. Buying a CT scanner at that time would cost about DM 2 million. So a PACS was four or five times more expensive than a CT scanner," Lemke said.
The cost of PACS has now dropped dramatically. Advances in microprocessor technology have boosted system capability and pushed the functionality threshold higher. Computational power essentially doubles every 18 months for the same price and is likely to do for at least the next 10 years, he said. Archive and storage capacity doubles every 12 months for a fixed price, and broadband capability doubles every nine months.
By the early 1990s, many more radiologists believed that PACS could become a practical reality. Confidence grew with favorable reports from a number of groundbreaking trial sites. The Danube Hospital in Vienna became the first European facility to operate with a filmless, fully digital radiology department when it opened in 1991. Plans to place a PACS in a London hospital finally received full government backing in 1990, and by spring 1996, the entire Hammersmith Hospital had gone filmless.
Pioneering radiologists who pushed through these projects at a time when the technology was not entirely proven made a major contribution to the acceptance of PACS. These early advocates faced considerable criticism from some quarters, and, to some extent, they put their careers and reputations on the line.
"With some technologies, you don't need people to stick their neck out. Take the Internet, for example. But other technologies, like PACS, take their time. These are technologies that change the pattern of people's behavior, their workflow, quite dramatically," Lemke said.
Introduction of digital imaging hardware, for example, has caused far less upheaval than establishing a digital image management solution. CT scanners may have revolutionized radiologists' view of anatomy, but the technology itself has not disrupted essential workflow processes. PACS, on the other hand, changes the way diagnoses are made and communicated and how radiologists interact with one another and with referring physicians.
"Once you run a PACS, you can't say one morning, 'Okay, today I won't use it.' PACS changes the infrastructure of radiology and hence the workflow of all the people who are part of that infrastructure," he said.
Looking back with the benefit of hindsight, Lemke suggests that the early PACS pioneers, himself included, should have tried to reengineer radiological workflow before introducing digital management systems. Instead, they focused first on developing the technology and then faced problems trying to fit it into conventionally run departments.
"I blame myself for not pushing the workflow issue in the beginning," he said. "That paper in 1979 included just one or two sentences where I said that the work profile of physicians will change dramatically. It was a feeling in my stomach, but I never realized how important it was."
He recommends that today's PACS purchasers consider exactly what they want the technology to deliver before making a selection. This is best achieved through intensive roundtable discussions with representatives from all groups whose work will be affected by the PACS. Thorough analysis may cost the equivalent of one person-year in time, or about 3% of the cost of PACS, but the investment is worth it.
"At least you get some assurance that PACS will do the work you expect it to do. It you try to build or buy a PACS as if it were an off-the-shelf product, you will finish up with dissatisfied customers," he said.
Though not all hospitals in Europe have PACS yet, the complete digitization of radiology seems inevitable. Acceptance of a filmless department has been encouraged by similar trends in other fields. When the technology was in its infancy, radiologists claimed they would never read images on anything other than light boxes. But just as many other professionals have discarded paper-driven procedures in favor of soft-copy workflow, radiologists now routinely view imaging results on computer screens.
State-of-the-art diagnostic monitors provide improved resolution and functionality. Radiologists can magnify regions of interest and adjust contrast to visualize suspect areas more clearly. But their initial reluctance to abandon hard copy was not due solely to the lack of high-quality monitors, Lemke said. Familiarity with existing ways of working played a large part.
"If, back in 1895, Wilhelm Conrad Roentgen had used a display unit with a resolution of 500 x 500 pixels to demonstrate his radiographs, hard copy would never have been developed. It is only because for close to 100 years we were using film, and the whole accustomed mindset was so strong in this direction, that so many things that could have been done on a monitor were not done," he said.
The battle for complete acceptance of PACS may be won in radiology, but the technology still will not sell itself. Radiologists need management systems that are capable of moving, manipulating, storing, and retrieving increasingly complex, information-rich images.
"It is an open-ended spiral," Lemke said. "Computers and communications in the scanners generate vast amounts of data, so we need more computers and communication systems to process the data and display the images to radiologists and other clinicians. This is a never-ending process."
To address this issue, computer-aided detection (or even diagnosis) packages will almost certainly become an important part of PACS in the future. Fiber optics and frequency multiplexing will make it possible to move huge data sets. Smart algorithms will then use what is known as "progressive compression," which allows doctors to work on low-resolution images while the full high-resolution version is loading. Once imaging informatics experts understand the perceptional processes radiologists use, they can tailor the algorithms to aid diagnosis, Lemke said.
He shares the growing view that PACS will eventually become hospital-wide rather than remaining a tool for radiologists. The technology is already making inroads into other areas where images are essential, such as pathology, endoscopy, and cardiology. The greatest interest, however, appears to be in surgery. Many companies are already developing dedicated surgical PACS, or S-PACS. The market for S-PACS could potentially surpass the radiology PACS market, given the proportion of hospital space devoted to surgery and the rising importance of imaging in therapy.
One key difference between S-PACS and its radiological predecessor is that the A for archiving has been replaced with an A for acquisition. This reflects surgeons' greater need to capture views in real-time rather than to store pictures. The ideal system should be capable of merging functional and morphologic data to display the effects of anatomic intervention on physiology.
Learning from history, Lemke is assigning top priority to workflow reengineering in his own S-PACS work. The question of where images fit into surgical workflow will be addressed before the technology reaches operating rooms. A new group of pioneers willing to trial the S-PACS prototypes will also be needed, but this should not be a problem, he said. Although radiologists' acceptance of PACS took many years, surgeons seem less skeptical of S-PACS, possibly because PACS is now regarded as a mature technology.
"It is a sign of the times. The idea of PACS is not so weird and frightening any more," Lemke said.