While the technology side of PACS is strong, it’s the way you implement your PACS--and especially the way you integrate your PACS with your hospital and radiology information system--that makes all the difference.
You name the technology--either available now or in the near future--and you can use it in your PACS. For example, the latest hard drives--faster, roomier, smaller and cheaper--make a lot of sense as arrays for storing recent multi-megabyte medical image files locally. Inexpensive and plentiful RAM--standard on many platforms--simplifies manipulating those images on each workstation. Even off-the-shelf monitors are getting bigger, brighter and sharper, giving physicians a better look at the pictures.
Long-term storage of images--even on film--has always been a problem of not enough space and not fast enough access. That is changing with networked optical disk jukeboxes that can hold many terabytes (thousands of gigabytes or millions of megabytes). You’ll probably be archiving older images on digital linear-tape jukeboxes that offer even greater capacity. Some facilities opt for pressing their own CDs: small, cheap, non-degrading and at least 600 MB each. Throw in automatic lossless image compression (which guarantees that you get your original image back again) and you can easily double all these capacities.
Multiprocessor servers are just right for handling big image throughput fast. There are so many good ones available, you can even be choosy about platforms that match your other hardware.
Moving those images around from servers to workstations requires a heavy-duty network. Luckily, that’s exactly the standard these days. Run-of-the-mill 100 Mbps fast Ethernet may suffice for your backbone (imaging server, plus storage, plus archives), but you really should consider something on the order of switched high-speed 622 Mbps ATM, Fiber Channel, or a comparable alternative. (Most observers regard gigabit Ethernet as not quite ready for prime time--yet. ATM has the advantage of allowing quality of service features that can prioritize certain traffic--like images bound for the emergency room.) On the image acquisition end, a stressed PACS might be pushing about 2 GB per hour--per imager--and you want to be able to handle all that traffic: something on the order of switched 10 Mbps Ethernet or faster makes sense. On the diagnostic viewing end--where the radiologist, cardiologist or other specialist actually makes the diagnosis--you want at least 100 Mbps Ethernet. These are time-intensive and critical applications, and need the bandwidth. Naturally, you’ll want to make sure your hubs and routers are up to speed, too.
Standards are playing a key role in simplifying the PACS picture. For example, DICOM (Digital Imaging and Communications in Medicine) is the standard for connecting medical imaging equipment with networks. Version 3.0 of the DICOM standard is especially flexible. It covers existing imaging modes and allows for the definition of new modes as technology advances. Middleware standards like CORBA are already proving useful in linking legacy imaging systems. William Langdon, executive director with Superior Consultant Company, Southfield, Mich., notes that open systems and standards keep vendors competitive.
The case for integration
While the technology side of the picture is rosy, it is the way you implement your PACS--and especially the way you integrate your PACS with your hospital or radiology information system--that is making all the difference these days. Part of the reason has to do with the hefty price tag for your PACS: from the hundreds of thousands to millions of dollars. That can be tough to justify if all your PACS does is help a handful of radiologists. Another part of the reason has to do with that ubiquitous partner of current health services, managed care. You have to do the right things at the right time in the right way--or you don’t get paid. Your PACS--by itself--can’t handle these conditions.
PACS is not just for radiology anymore. Sure, it can handle X-rays and even computed radiography for special effects, but a well-integrated PACS can also deal with ultrasound, fluoroscopy, computed tomography, magnetic resonance, nuclear medicine and other image modalities. In fact, one shrewd move is to have an image server physically close to each of these imaging areas: that’s probably where most of the diagnostic viewing is going to happen also, and you’ll save the rest of the network from unnecessary traffic.
Even if you are only considering radiology, for a geographically dispersed facility-- or set of facilities--PACS implementation is of special importance. You probably don’t want to maintain a complete radiology department in every building: it’s not cost effective. It makes more sense to bring the images from wherever they originate to wherever the radiologist, or other diagnosing physician, may be. (In fact, some enterprises are renting their radiology departments out to handle images from virtually any spot on earth.) Such a distributed system of diagnosis lets the diagnosers see more images, faster and easier, than any other system. Clearly, a PACS network that large must coincide with the HIS network also. Doesn’t it make sense to combine them?
As Jim Mulvaney, director of marketing for Dynamic Healthcare Technologies, Lake Mary, Fla., (a provider of RIS and PACS), points out, there are even more reasons to combine your PACS with your HIS. From a diagnostic point of view, it is now possible to get a complete package in front of the radiologist: not just an image and a name or number, but patient history, clinical information, symptoms, reason for presentation and previous examination history. Such collateral information can greatly aid in the interpretation of the images. Since the HIS probably has most of this information already, why not include it with the image? This can’t help but result in a better diagnosis, more appropriate treatment, better patient care and more effective use of services.
The radiologist can speak the diagnosis and use voice-to-text conversion to turn the diagnosis into words. With an integrated RIS, the radiologist can then proof-read the text on the spot, make any changes and incorporate the diagnostic text into the total package.
Now what? By itself, a PACS would simply store the image, because the PACS doesn’t know where the image should go next. But a RIS or HIS does, and zaps the image, diagnosis and other patient information off to the emergency room or wherever it needs to be. The treating physician can have the complete results in minutes, making a big difference in treatment decisions, patient length of stay, and other aspects both medical and financial. Anything that saves clinician time is a plus. The key here is that the RIS or HIS is handling the workflow management for the PACS images, improving speed and efficiency.
An integrated system also keeps the records that let you ask analytical questions about the treatment process. Was this MRI the best way to diagnose this condition? You can now find out by comparing similar situations, a process called outcomes analysis. The result is often an overall reduction in the number of services done.
Mulvaney suggests that a major impact of such integrated systems occurs at the very beginning of the process, when scheduling an imaging service. The healthcare enterprise, whether a hospital or clinic or medical office, usually has practice guidelines for treatment. For example, they may have decided to treat lower back pain with five days of rest before performing an MRI. You can incorporate these guidelines into the system, and the system can then reinforce the guidelines. When scheduling an MRI, the system may remind the physician about the five days of rest prerequisite. The physician may well overrule the system, but the system can track that also. The result of lots of overrulings may be a decision to modify the practice guidelines--or something on the order of, "Say, Dr. X, why do you use twice as many MRI scans as everyone else?"
Managed care rules
Not surprisingly, managed care can get into the act also. The managed care enterprise has its rules too, and these can be a part of the system. That sounds bad for the radiologist who has to work by those rules, but the result can be beneficial: the radiologist might actually get paid for the work. Since the system is reinforcing the guidelines of the third-party payor, it is guaranteed that the imaging will meet their payment standards. The ability to produce that grail of managed care--the clean bill--may change the behavior of the service provider some, but in a way that has financial merit at least.
Henri (Rik) Primo, director of information systems for the health services division of Siemens Medical Systems, Iselin, N.J., notes that the goal is really to make the person the focus of the solution. That actually means trying to avoid having the person become a patient, a goal that images presented in context can support. It is only now that we possess the technology that makes possible the integration of image and information.
While Primo sees the integration of PACS and RIS now as controlling what might otherwise be an overload of information, in the future the goal will become making medical information available to the appropriate people anywhere at any time. For example, wavelet or fractal compression of images effectively multiplies bandwidth to remote viewers--and makes sense for physicians who don’t necessarily need the detail that the radiologist does. Again, this represents the theme of technology enabling the intelligence of the system.
The marriage of PACS with RIS or HIS impacts the bottom line positively. PACS will never be a trivially inexpensive technology. But uniting PACS with RIS or HIS enables a more powerful--and reasonable--combination than either alone. PACS handles the images. RIS or HIS handles the data, including patient data, treatment guidelines and PACS images. And that is a beautiful picture.
CASE IN POINT
Baylor Medical Center, Irving, Texas
In many ways, the PACS at the Baylor Medical Center at Irving, Texas, has followed a nearly classical path in its development. Located just outside Dallas, the Irving facility includes a 300-bed hospital and several associated medical office buildings and medical centers. The Department of Radiology includes modalities such as diagnostic radiology, magnetic resonance imaging, computed tomography, ultrasound, angiography and nuclear medicine.
As Mona Martinez RTR, director of radiology, explains, the original motivation for investigating the possibility of a PACS was a common one in radiology departments: lack of storage space for traditional X-ray films. Faced with the prospect of the construction involved in expanding the file area within the hospital, it was actually the administrator who suggested that it would make more sense to put the money toward a full-fledged PACS. This turn of events helped avoid the usual radiology-trying-to-talk-administration-into-buying-PACS scenario. Instead, the immediate goal became the elimination of the file room--including the space, film expense and distance problems--and paper associated with the traditional system.
The first phase of PACS implementation was a connection with the hospital’s emergency room facilities. This permitted emergency room personnel to view diagnostic images rapidly, improving patient care and speeding up emergency room services. The Irving facility uses a PACS from Siemens Medical Systems, Iselin, N.J. The latest enhancements include a robotic optical disk jukebox. Intended for archival storage, the device holds 100 terabytes of images. RAID systems handle the most recent images before they move to the archives.
Next, the department of radiology had to handle a new hospital nearby, Baylor Medical Center at Irving/Coppell. Although not part of the original plan, the PACS integrated the Coppell’s imaging services. This had several positive results. Coppell was filmless from day one. Also, since Coppell had its own radiology department, the two departments could cooperate on diagnostic tasks: separation of the two facilities made no difference.
This collaboration was possible because the network carrying the PACS images is extremely fast. Robbie Perez, network services manager, explains that the internal network is a collapsed FDDI (Fiber Distributed Data Interface) ring running at 100 Mbps. (FDDI also has a reputation for robustness that appeals to managers of life-or-death information flow.) The connection to Coppell is a 150 Mbps extension of the FDDI ring, running over ATM. The PACS can move images to the radiologist as quickly as necessary. The flexibility of the original system enabled the accommodation of the new hospital.
Currently, Martinez is involved in moving the PACS into all the critical care areas. This will give physicians in, say, the intensive care unit the same rapid access to medical images that the emergency facilities already enjoy. Also, of possibly greater long-term importance, the facility is installing a new radiology information system (RIS). Martinez observes that it is a good idea to plan your PACS and RIS implementations together, to ensure that you have the most compatible overall system possible. (A recent Frost & Sullivan report comes to a similar conclusion. Market projections indicate the need for one-stop shopping for integrated HIS/RIS/PACS solutions.) The Irving group actually delayed completing this phase of the PACS implementation in order to install the RIS.
The goal is to integrate the existing PACS with the Cerner (Kansas City, Mo.) HNA Millennium RIS. The ultimate aim is for the RIS to control all the data, including the imaging data. The RIS will handle scheduling, moving images from the PACS to the radiologist and moving the diagnosis to its destination.
Teleradiology is an ongoing part of the PACS plan at Irving. The existing PACS system uses 128 Kbps ISDN lines to connect with remote sites in its affiliated medical office buildings and medical centers, as well as to the homes of on-call radiologists. This distributed system allows the images to move quickly where they are needed.
The next step in teleradiology at Irving is to allow Web access to the medical images. This will simplify and expand the diagnostic process further.
Edmund X. DeJesus is a freelance technical writer and former editor at BYTE magazine.