PACS through the years definitely has been a "lessons learned" experience. This article describes some of the history, and evolution of these systems as well as early experiences. The first large-scale PACS installation was in 1982 at the University of Kansas, Kansas City. As is typical with early experiments, people learned more about how NOT to do it than they did how to do it. The experiences from this installation taught that PACS requires abundant data exchange, and that an Ethernet connection is not necessarily the best method to do this (remember the fastest connection at that time was 10 MBit/sec, and routers and bridges were not really commonplace).
That brings us to one of the big "drivers" in this technology; network standardization. In the early '80s, there was no one single standard. TCP/IP was just one of the several options available. The government was pushing for the OSI standard, and major manufacturers, notably GM, were trying to enforce broadband instead of Ethernet standards. As a matter of fact, the first PACS by Philips used broadband technology. There were also "ad-hoc" developments using inventive solutions in the '80s, such as the one at Michigan State University, whereby images from three CT scanners were sent to the University diagnostic center for reading via the commercial CATV cable system; a predecessor to cable modems! The early versions of DICOM, called ACR-NEMA, only specified a dedicated point-to-point connection, leaving it up to the manufacturer to exchange the data via their network of choice. It took about 10 years for the network standard to emerge, something we take for granted now. Today, everyone uses the TCP/IP as the basis for network communication, making it easy for new developments at the physical level such as gigabit/sec Ethernet to be deployed.
PACS & PIXELS
Display technology has also come a long way. Initially, there were only 1k by 1k monitors, and there was a big question as to whether the resolution was going to be sufficient. It is interesting to note that this discussion is still very valid, not because higher resolution monitors are unavailable, but because there is a definite price trade-off. For example, should one use a 3 Mpixel monitor for reading of chest images, or is a 5 MPixel more efficient? Is it OK for referring physicians to use their high-quality PC monitor, or should they use a medical grade monitor with a special video board? It is not only a question of using the appropriate hardware, but also using the correct display look-up tables to provide a consistent presentation. Another issue is still the appropriate number of viewing monitors. Early PACS viewing stations mimicked the electronic film alternators they were supposed to replace, and most of them therefore had two rows of four monitors stacked above each other. Most institutions now seem to be happy with two monitors, especially since more people like to use "stack view" with digital modalities such as CT and MR. As for chest, there is still the obvious cost factor and limitation of space. Interestingly enough, some institutions use four monitors for all their diagnostic viewing, while others use two. Talking with radiologists from either institution, they both seem to make a convincing story justifying their configuration.
Economically speaking, it is interesting to see that over the years, the question as to whether PACS is cost-justifiable has not gotten any easier. The early work at the Hospital of the University of Pennsylvania, as well as at Washington University in Seattle, provided some early numbers and a framework to use, however, a clear "savings-model" is still difficult to formulate. The challenge is that one cannot just look at how much is saved by eliminating film, but that the true savings lie more in the increases in efficiency. Productivity studies by the VA in Baltimore in the early '90s have helped in this regard. However, one has to realize that, as Dr. Eliot Siegel from the VA in Baltimore strongly advocates, one has to re-engineer a department and its workflow to make use of the advantages of this new technology to really realize the benefits. As you can imagine, the early PAC systems only replaced their film-based operation with a softcopy environment without the emphasizing re-engineering.
One of the true drivers that made PACS technology affordable was the implementation of standards, especially HL7 in the IT domain, and