There is nothing magical about what makes up a picture archiving and communications (PAC) system. A PAC system is nothing more than a collection of peripheral devices connected by applications software. The uniqueness of any particular system is determined by the way the software interacts with the peripherals chosen in the vendor’s system design. Most systems use identical or similar components often running the same operating system, database management software, and, in some instances, even applications software.
Consequently, peripherals play a fairly significant role in the design of a PACS, as much if not more so than the applications software itself. These peripherals include workstations, archives, file and Web servers, film digitizers, computed or digital radiography systems, monitors, display adapters, modality and software interface devices, networks, and virtually every other piece of the PACS puzzle.
Historically, the independent PACS vendors tend to integrate the latest technology first, while the big six, who control between 60% and 70% of the market, wait to assess either acceptance or missteps before adopting a change in hardware platforms. This has been especially evident with archive technology, a key to the widespread acceptance and implementation of PACS. In the past 2 years alone, archiving has quadrupled in capacity while prices of the storage devices, at least to the vendors, have been halved. The double-edged sword facing buyers is making sure they have middle-of-the-road cost-effective technology–leading edge, not bleeding edge or dull edge.
With very few exceptions, virtually every workstation in the PACS marketplace runs on a Windows NT platform. The hardware provider and configuration used vary from PACS vendor to PACS vendor and are typically transparent to the end user. With random access memory (RAM) prices at record low costs, many vendors have been loading the system with up to 1 GB of RAM to improve overall workstation performance. Average hard disk space, which used to be in the 2 GB range, now begins at 10 GB and often will go up to 20 GB or more, providing radiologists’ reading stations with several days’ worth of images locally. Of course, this capacity is also needed to accommodate the six- to eight-fold increase in slices coming from multislice computed tomography (CT). It is interesting to note that any incremental increases in processor speeds once? 600 MHz is exceeded-and all do now-provide little if any advantage to the workstation operation.
Workstation monitors have shown dramatic changes, with flat panel technology providing one of the greatest technological breakthroughs seen in years. Flat panel’s ability to allow for significantly reduced space and weight requirements, combined with reduced power consumption (and heat generation), longer display life, the elimination of special furniture requirements, and a host of other factors, has made it the display of choice for many PACS users. While prices remain about 50% higher than traditional CRT (cathode ray tube) prices in the medical arena, the advantages that flat panel offers provide a near break-even comparison when considering the total cost of ownership over a 3-year return on investment period. Additionally, prices on LCD displays have dropped nearly 50% in the commercial sector over the past 6 months alone, and presumably this price change will impact medical as well, although somewhat less dramatically due to medical’s limited volume use.
Despite significant changes and advances in both technology and price with display card adapters in the commercial sector, attempts to transfer this technology over to the medical marketplace have all but fallen flat. The few PACS vendors who have tried to use off-the-shelf cards have all come back to using specialized high-quality display adapters provided by a few select vendors. This no doubt is due to the specialized nature of medical display and image processing requirements and the high degree of reliability that this application dictates.
One of the most interesting changes in the workstation sector is a movement away from 2K resolution displays to 1K resolution displays. The reasons for this change can be attributed not just to price, although with savings of up to 35% or more over 2K display workstations, cost definitely is a factor. Many radiologists have become comfortable with reading images at 1K resolution provided the full 2K data set is available in memory for display.
A Role for CAD
In the next year or two, the reasons for using 1K display will be even more compelling, once the inclusion of computer aided detection (CAD) devices similar to those used for mammography today are included in PACS design. The inclusion of CAD in a PACS design will allow the CAD system to review the 2K data set for the radiologist once an image is received by the file server. With a simple click on the CAD button, the system will allow the radiologist to either magnify the suspect area at 2K resolution or simply advance to the next image if no suspect area is identified by the CAD device, significantly streamlining radiologist interpretation of general radiographic images. The first of these systems is due out at the end of this year and will address chest and lung first, with other anatomical regions following suit in subsequent years.
Along the same lines, use of dedicated workstations in CT and MRI is beginning to disappear, instead replaced by either application-specific software that can reside on the PACS workstation or independent third-party software solutions that run on personal computers. The advent of the software-only solutions for 3-D imaging heralds this new era, replacing six-figure dedicated stand-alone 3-D solutions and independent workstations with values of $70,000-80,000 each (although often bundled as freebies with the purchase of a new scanner), with full solutions (hardware and software) that cost less than $20,000 to provide many, if not all, of the same applications.
Nowhere has change been more apparent than with archive technology. The alphabet soup of acronyms continues to confuse and confound both users and vendors alike, with new capacities, technology, and pricing appearing daily. In medical archiving, magneto-optical disk (MOD) and digital linear tape (DLT) remain the preferred technologies, with variations of these being adopted in high-capacity archive applications. Both advanced intelligent tape (AIT) and digital versatile disks (DVD) are also being explored by a few vendors as cost-effective options for archiving, while write-once, read-many (WORM) and digital audio tape (DAT) have fallen from favor in all but very limited applications. MOD capacity has also increased fourfold in the past 2 years alone (from 1.2 GB per platter to 5.4 GB), while initial DVDs are said to offer as much as 9 GB each, although delivery of these is slowly rolling out due to concerns about format issues that mimic the VHS versus beta videotape scenario in years past. DLT can also offer up to 30 GB per tape, making it ideal in multi-terabyte archive applications, especially where bundled with a high speed tape reader that can provide end-to-end searches in less than 2 minutes per tape.
Networks have evolved to where gigabit backbones have almost become the norm, supplanting fiber distributed data interface (FDDI) and fiber channel networks that were promoted just a few years back. While other 100 MB networks like fast Ethernet and 100-Base-T are still being used extensively, it is not uncommon to see 100 MB to the desktop in limited applications any more, especially while riding on ultra-high-speed network backbones including both asynchronous transfer method (ATM) and gigabit. Facilities also are planning their network utilization to include not just radiology PACS applications, but also cardiology PACS, electronic medical records (EMR), distance learning (videoconferencing), and much more.
Servers have become bigger, faster, and cheaper, with the exception of the use of high-capacity RAID (redundant array of inexpensive disks) drives that can store a month’s worth of images at the server level; however, the changes that can be considered dramatic with server technology have been limited. Web server use has begun to take off as a means of distributing images to select primary care physicians (PCPs), although facilities are beginning to look closer at the impact the Healthcare Insurance Portability and Accountability Act (HIPAA) will have on image distribution via the World Wide Web. The requirement to adhere to HIPAA security and confidentiality requirements is at least 18 months out and possibly longer, depending on current changes being considered by Health Care Financing Administration (HCFA). Most of the major PACS vendors have already embraced solutions supplied by independent third parties that utilize progressive image streaming to speed up image transfer, while the independents themselves have secured more than? 50 clients to distribute images to PCPs.
Modality interfaces really have not shown significant changes in either price or performance. Most newer imaging modalities provide for native DICOM (Digital Imaging and Communications in Medicine) output, either directly from the modality itself or via a dedicated workstation. Most clinical information systems (hospital and radiology information systems) also support at least one version of the Health Level Seven (HL-7) standard, with software upgrades typically bringing the system up to the current 2.3 revision level. Very few vendors offer DICOM as anything other than a standard part of their offering, although there are a few cases-notably in ultrasound-where the modality is promoted as DICOM-compliant yet native DICOM output is a costly option.
Digitizer Choices, CR and DR
Film digitizer use has evolved from a single laser scanner provider scenario to one where charge-coupled device (CCD)-based units are now being implemented on a widespread basis. There are several reasons for this change. Service and maintenance costs for CCDs are considerably less than for lasers, which require quarterly calibration. Improvements in CCD technology during the past few years have allowed CCDs to be favorably compared to lasers with regards to image quality. And? CCDs have been proven to be considerably more reliable than lasers, with high mean time between failures (MTBF). Lastly, entry costs with CCDs are about 25-30% lower than lasers, allowing vendors to either make higher profit margins or pass on the savings to the end user.
Computed radiography (CR) has changed significantly in price, performance, and application. The development of tabletop CR has allowed for image processing to take place where the images are taken (in ICU and CCU applications, for example, as well as low-volume imaging centers) instead of having to transport plates to the department for centralized processing. Costs for CR devices have dropped as well, with moderate throughput CR systems available in the low $100s and tabletop CR in the mid $60s. While postprocessing algorithms and added features can drive up the price significantly, offering these improvements as options instead of bundled has helped keep CR pricing in check.
Digital radiography (DR) has grown from a single source to a market with nearly a dozen players, driving down costs by nearly 50% to under $200K in some applications. While vendors may argue the merits of one technology over the other, the primary driver in vendor selection continues to be price, as neither throughput nor reimbursement changes depend on the DR technology selected. There are other factors that can impact the price of DR systems, namely the ability of a vendor to retrofit rooms using existing generators, and this needs to be taken into consideration when evaluating DR vendors.
Many PACS providers offer multiple options for peripherals, typically offering two or more for each peripheral under consideration, enabling the buyer the latitude to select a best-of-breed solution from a single provider. By looking closely at the offerings the vendors present and evaluating the interplay between the applications software and components making up the PACS, health care organizations should be able to implement a system that will serve for several years to come.
Michael J. Cannavo is president, Image Management Consultants, Winter Springs, Fla, [email protected].