It's simple, says Michael Gray of Novato, Calif.-based Gray Consulting : "The less you spend, the less you get."

Gray understands, of course, that most freestanding imaging centers and group practices, who represent a minority of PACS purchasers, are on very tight budgets. (Only 25 percent of respondents to a survey published in the June 2007 issue of Medical Imaging Magazine were not hospitals or medical centers. This isn't a direct indicator of who's buying, but it's a good clue.) But he cautions that cutting corners on a PACS may be very costly over the typical five-year lifespan of a system. "You end up end up with a PACS that doesn't do some very important things," he says. "You're not getting the work done nearly as efficiently as you could."

The more efficient your PACS, the more studies you can read. And efficiency is directly related to the sophistication of the PACS. "The ability to create hanging protocols with great complexity, and to create an unlimited number of them is hugely important," says Gray. "One guy in a practice looks at a study one way, another looks at a study another way." And delivering what's wanted is the key to doing good business -- and more business. Unfortunately, radiologists pay little attention to the needs of the referring physicians when they consider what they want in a PACS. Delivering the images the way they're wanted is key. And making access to the images easy -- which often means delivery on a clearly marked CD -- is another piece of the puzzle.

Deciding which PACS to buy can be a dizzying and confusing experience. There are about 80 PACS manufacturers, according to Medical Imaging Magazine, and hundreds more vendors who repackage those systems to sell under their own brand name. While price is a crucial consideration, of course, functionality should be the trump card, argues Gray. He speaks from experience. "I've done a lot of RFPs," Gray says. "I've worked on more than 50 projects. And by the time we get down to the vendors we're serious about, that are matched up on features, functionality, architecture, the systems will be within five percent of each other in terms of cost. You would think there's collusion, but there's not."

It's important not to look at the prices that PACS vendors start with, says Gray, because those vary widely. But key capabilities that should be examined, in rough order, are:

• The ability to create an unlimited number of complex hanging protocols;
• RIS integration, including how easily and accurately paper-based information can be imported into the PACS;
• The ability to create flexible worklists;
• Pure DICOM storage. Many PACS take DICOM data and immediately transform it into proprietary formats or, in Fuji's case, as Microsoft objects. When it comes time to get a new PACS, this can be a huge and costly problem, because it means if your practice switches vendors, it has to break everything back into DICOM format to import into the new system. Both the cost and difficulty of this process should be assessed if the vendor isn't sticking with DICOM throughout.
• How much can the technologists do with a study before it gets to the radiologist? Can they merge, split, and delete studies? Can they check the accuracy of information and the hanging protocols? All of these capabilities make for greater efficiency.
• Is the hardware solid and stable? Michael Cannavo , the founder and President of IMC (Image Management Consultants), which is based in Winter Springs, Fla., points out that most systems use similar hardware, provided by either Dell or Hewlett-Packard. But some low-end systems are built on PCs running Windows and a SQL directory, a combination much less reliable than more expensive hardware supporting Linux and Oracle software.
• The GUI (graphical user interface). Cannavo argues that it may be the key differentiator between PACS that are otherwise nearly identical, and that smaller groups need to pick a GUI they're comfortable with. Gray argues that while you have to like the GUI, it's a relatively minor consideration.
• Vendor service, which includes software upgrades, user support, and training. "PACS from larger vendors -- GE, Philips, Siemens, and AGFA, for example, are typically 20 to 25 percent more expensive" (than smaller vendors), says Cannavo. "That's because they offer better support, and it's worth the difference."

In the end, say both Gray and Cannavo, there will always be tradeoffs. For example, Gray tells of a small client that may purchase a PACS that stores data in a proprietary format -- a no-no, in Gray's book. The practice does about 10,000 procedures a year. "I told them that's going to cost you when you move to a new system. They ask, 'How much?' I run the data and tell them $25,000. They say, 'Well, the systems going to cost me a half a million and five years from now it's only going to cost $25,000 more. I can live with that."

Small groups shouldn't be too concerned about such tradeoffs, says Gray, as long as they're getting the crucial features they need to work efficiently and satisfy their clients -- the physicians who are ordering the studies. "You need to look at the upper end, the full-featured PACS," he says. Find a vendor you like that can scale its technology down to your size (this capability isn't a given). In the end, Gray emphasizes, "You don't need Big Bertha, but you need all the features that any other radiology group would have."