BBI Contributing Editor

CHICAGO, Illinois — The perennial question asked by all who attend the annual meeting of the Radiological Society of North America (RSNA; Oak Brook, Illinois) is, "What's new?" At the 2001 meeting, held in the sprawling McCormack Place convention center last December, the answer had to be in terms of evolution, not revolution. The evolution toward complete digitization of medical imaging continues, as digital systems demonstrate their ability to provide superior images, create novel presentations and make the operation of imaging facilities more efficient.

Picture archiving and communication systems (PACS) now can be considered a sixth component of medical imaging, ranking alongside the five major modalities: X-ray, computed tomography (CT), magnetic resonance (MR), ultrasound and nuclear medicine. According to a recently completed study from the Concord Consulting Group (Concord, Massachusetts), the U.S. market for PACS in 2001 stood at $800 million, making it comparable in size to the markets for each of the modalities. Dozens of companies offer PACS, including most of the major modality companies, and it is tempting to assert that a shakeout is imminent.

Three salient histories are those of CT, MR and digital angiography. At one time there were 25 or so companies offering CT products, but today only seven or eight companies — GE Medical (Waukesha, Wisconsin); Siemens (Erlangen, Germany), Philips (Eindhoven, the Netherlands), Toshiba (Tokyo), Hitachi (Tokyo) and a few others — offer these machines. Almost exactly the same tale can be told for MR — 25 companies ultimately whittled down to the same major players as for CT. Similarly for digital angiography. Now that the major modality companies are active players in the PACS arena, one wonders if they will soon come to dominate it as these histories suggest. If so, the dozens of small companies in the field have something to fear.

But the histories of hospital information systems (HIS) suggest that the evolution of PACS may be different. The market for HIS is dominated by a handful of large vendors — Cerner (Kansas City, Missouri), McKesson HBOC (Alpharetta, Georgia), Siemens SMS (Malvern, Pennsylvania), Meditech (Dedham, Massachusetts) and IDX (Burlington, Vermont) — but there are dozens of smaller vendors, many of whom have been in business for decades. Unlike the imaging modalities, where only large companies are able to sustain the required investments in technology and marketing, information systems do not necessarily require similar investments. The reason is that the computer companies make the big investments in technology, and the information system companies that supply the specialized software can ride on the same wave. Once a small company has a modest installed base, it can hang on to this base, taking advantage of new technology as it comes along, without investing a great deal in marketing because owners find it expensive to change suppliers. Thus, small companies in this field are not as susceptible to the forces of competition as were the small suppliers of CT, MR and similar devices.

But that does not mean the field is immune from competition. At RSNA 2000, dozens of companies were touting the so-called ASP (application service provider) model for PACS. The concept implied that most of the computing power and storage would reside off-site, while on-site there would be only "thin clients" — small computers to originate requests sent to remote computers and accept results from them, using the Internet for communication. Furthermore, the cost of these systems would be billed per-study stored rather than as an up-front capital cost.

The flaw in this concept was revealed in the vast reduction of companies following the ASP model. By our count, 16 of the 32 companies that offered off-site storage and image distribution systems at RSNA 2000 were not present at RSNA 2001. The flaw lies in the nature of PACS, which do not lend themselves to the "thin client" concept. Since all of the imaging devices and workstations, as well as the network that connects them, must be on-site, there is not much advantage, and indeed there are penalties in terms of response time and reliability, from moving the archive off-site. This conclusion is only reinforced by the plummeting cost of digital storage. The only remaining argument for off-site storage is security, but it is not difficult technically to arrange for redundant records at another local location. As for billing per study stored, hospital administrators recognize it as simply an alternative payment mechanism, in which the vendor assumes the cost and risk of the capital investment, but which the user must ultimately pay.

There remain companies committed to aspects of the ASP model. Stentor (South San Francisco, California) offers an image distribution system that uses wavelet compression in a manner that permits rapid transmission over existing networks not designed for the large files that represent images. Some 70% of Stentor's customers take advantage of the pay-per-study option. Emageon (Birmingham, Alabama) and Inphact (Nashville, Tennessee) both offer Internet-based PACS, using the pitch that such systems are more easily scaled to smaller hospitals. Medicalis (Chestnut Hill, Massachusetts) offers a remote, Internet-based order-entry system developed at Boston's Brigham and Women's Hospital. Likewise, Fuji (Stamford, Connecticut) offers an Internet-based system that has been well-received in the market, as do many of the larger PACS companies.

Digital radiography

Much in evidence at RSNA 2001 were digital radiography (DR) systems, which capture X-ray images and convert them immediately to digital form. These systems compete with computed radiography (CR), an earlier technology pioneered by Fuji, which substitutes a reusable phosphor plate for film. The latent image on the plate is read out and digitized in a processor analogous to a film processor except that its processing is electronic, not chemical. The main advantage of DR over CR is that there is no cassette handling and no delay between exposure and availability of the image for review, so that a technologist can know immediately if the image was properly exposed. While not decisive, these advantages would seem to recommend DR over CR.

But DR still suffers from an important drawback. It is much more expensive that CR, mainly because of the difficulty of making almost blemish-free, large-scale arrays of transistors necessary to realize direct radiography. GE Medical has offered a DR mammography machine for the last two years, but it costs about $450,000. At RSNA 2001, Fischer Imaging (Denver, Colorado) offered a comparable, FDA-approved machine for about the same price. But conventional mammography machines cost less than $100,000, and a dedicated CR processor adds no more than another $100,000. Thus, direct radiography cannot compete on an economic basis with computed radiography.

Nevertheless, DR continues to develop, and there can be little doubt that the modality eventually will become competitive with CR. Hologic (Bedford, Massachusetts) showed direct radiography on the McCormack Place exhibit floor that grew out of developments originated at DuPont (Wilmington, Delaware) in the days when it was still in the medical imaging film and equipment business. Trixell, a consortium supported by Siemens, Philips and Thomson-CSF — now rechristened Thales Components Corp. (Totowa, New Jersey) — advertises that it is in full production mode of DR plates for incorporation into any X-ray system. Edge Medical (Hackensack, New Jersey) again showed its scanned flat-panel direct radiography detector, which avoids large transistor arrays and is therefore expected to be less expensive than existing DR systems.

Radiologists' fears

Despite the vast array of technological marvels now available to radiologists, a sense of beleaguerment pervaded some RSNA presentations. Times seem almost too good to last. During the last few years, radiologists have benefitted from the sharply rising use of imaging exams as primary care physicians and specialists increasingly seek the definitive results — positive or negative — that imaging provides. Usage of CT, MR and ultrasound has been increasing at more than 10% per year, and even traditional X-ray procedures, which often are replaced by digital procedures, show only slight declines. As a result, imaging facilities are busier than ever, and in spite of continued downward pressure on unit charges for examinations, radiologists' revenues are greater than ever.

But there are clouds on the distant horizon, as these large increases in usage are likely to draw regulatory and economic scrutiny. Managed care as a concept seems largely discredited at present, but with health insurance premiums rising 6% in 2000 and 9% in 2001, and even larger increases anticipated in years to come, demands for cost controls of some sort cannot be far behind. Since imaging accounts for a large part of the increase in costs, though not as dramatic as increases in costs for pharmaceuticals, it is likely to be subject to increasing scrutiny and control.

Over the past decade, the amount of imaging equipment in departments other than radiology and in physicians' offices has steadily increased, especially equipment for nuclear medicine and echocardiography in cardiologists' offices; ultrasound equipment in the offices of obstetricians, gynecologists, urologists, and vascular specialists; and X-ray equipment in the offices of orthopedic surgeons. David Levin, MD, reported that the proportion of noninvasive imaging done by nonradiologists increased from 27% to 32%. Levin attributes this increase to self-referral, that is, physicians referring patients for examinations using equipment that the physicians themselves own. As such equipment proliferates, this trend poses a threat to radiologists' business.

Computer-aided diagnosis (CAD) also is a threat. R2 Technologies (Los Altos, California) has had on the market for the last two years an FDA-approved device that digitizes mammographic films and indicates sites of suspicion on a display below the lightbox for the films. CADx (Laval, Quebec) and Intelligent Systems Software (Boca Raton, Florida) have similar systems, though not yet FDA-approved. These companies also have works in progress for CAD for chest X-rays and for CT of the lungs. All current systems are conceived as aids to diagnosis, diminishing the chances of false negatives, but the technology has the potential someday to eliminate the radiologist from the task of interpretation altogether.

Finally, radiologists worry about the possibility of being displaced by remote readers. With modern communication technology, as embodied in PACS, there is no need for the radiologist who interprets an image to be anywhere near the site where the image is created. In principle, remote readers can offer much better service than local radiologists because of economies of scale that would enable them to offer subspecialist expertise for every examination, off-hour coverage, high-speed turnaround and lower cost. American radiologists darkly imagine managed care companies contracting with radiologists in India and Pakistan who will provide readings for pennies.

All of these seem overblown fears. Medical imaging has climbed to its present, well-regarded position because it is safe, painless and usually definitive in diagnosing or ruling out disease. As RSNA 2001 demonstrated, the technologies continue to advance so that there is no reason to expect that the use of imaging in medical practice has anywhere to go but up. It is likely that imaging will come under increasing scrutiny as the amounts devoted to it increase, but imaging has a secure and growing role in cost-effective medical practice. While payments per examination may continue to decline, aggregate payments are likely to continue to rise.

While other physicians may take an increasing proportion of imaging examinations, radiology departments still perform the majority of examinations (two-thirds) and radiologists interpret them. Other physicians find it advantageous to own ultrasound and X-ray equipment, but CT and MR machines are generally too expensive for other physicians whose patients with legitimate need for such equipment are limited in number. Most physicians recognize the special skills that radiologists possess.

Remote reading of imaging examinations has a certain plausibility because it has already happened for off-hour readings, where local groups in a city cover for one another at night and on weekends. One large practice on the West Coast has rented a home in England where members reside for two-month periods reading images by day that were created at night at their home site. That practice is likely to grow more widespread, but daytime reading by remote radiologists is not so likely for several reasons. For one, local radiologists provide services beside reading, including supervision and management of the department. For another, local physicians generally prefer readings from someone whom they know and can trust. Finally, there is little economic incentive, even though remote readers might offer lower-cost services, because payers, who would have the most to gain, know they would encounter strong resistance from local physicians and patients if they imposed remote reading as a way of saving money. All things taken together, medical imaging — for patients, providers and suppliers — faces a bright future.