BBI Contributing Editor

CHICAGO, Illinois Like many other aspects of modern life, the practice of radiology is steadily converting to the digital world. Initially, there was the drawback that the digitization was too coarse, and digitized images were judged inferior to their analog counterparts. But once a threshold of fine detail has been passed, digital images are in every way superior they are not degraded upon copying; they do not decay with age while in storage; they can be duplicated with ease and displayed in multiple locations simultaneously; they can be processed and reprocessed to enhance desired features without losing the original data; and they may well be less expensive.

This conversion was fully in evidence at the annual meeting of the Radiological Society of North America (RSNA; Oak Brook, Illinois) in Chicago's sprawling McCormick Place convention center last December. All of the modalities for making medical images except X-ray imaging already are fully digital, and X-ray imaging is gradually yielding to advances in digital technology wrought by computed radiography (CR) and digital radiography (DR). Film still is used as the capture medium for the majority of X-ray images, but it is clear that CR, which captures X-ray images on a reusable phosphor plate, and DR, which captures images on a large-scale transistor array that immediately digitizes the data, eventually will supplant film.

Improved images, lower costs

CR now has been available for more than 20 years, and each year brings new developments that provide better images and reduced costs. FujiFilm Medical Systems (Stamford, Connecticut) was the inventor of CR, but both Eastman Kodak (Rochester, New York) and Agfa (Ridgefield Park, New Jersey) are competitors. A new low-cost CR system from Orex Computed Radiography (Yokneam, Israel) also was on display at RSNA, advertised as inexpensive enough so that CR facilities do not have to be centralized. On exposure to X-rays, CR captures a latent image in a manner analogous to film. In CR, however, the developing process is not chemical but electronic. In some of its models, Fuji integrates the processing system with the table or wall-mount apparatus so that images become available almost immediately after exposure, thus providing what had seemed to be a unique feature of DR systems. Nevertheless, CR systems were slower than DR systems because of the time it took to scan the latent image. This year Fuji introduced an integrated system that scans multiple lines at one time, thus providing an image almost immediately after exposure and further erasing operational differences between the two kinds of receptor systems.

DR systems are generally a good deal more expensive than CR, a fact that has impeded their adoption, though it has not impeded companies from developing DR systems, which now are available from Canon (Irvine California), Cares Built (Keyport, New Jersey), CMT Medical Technologies (Haifa, Israel), Dalsa Life Sciences (Tucson, Arizona), Edge Medical Devices (Hackensack, New Jersey), Hologic (Bedford, Massachusetts), Imaging Dynamics (Calgary, Alberta), Nucletron (Columbia, Maryland), Sterne Medical Equipment (Brampton, Ontario), Swissray (Elmsford, New York), Trixell (Totowa, New Jersey), Varian (Palo Alto, California) and Wuestec (Mobile, Alabama).

There are various technical approaches to DR. The least-expensive systems, such as those from Cares Built and Swissray, convert the X-ray image to light using a scintillating screen and then focus the light image on one or more CCDs to create the digital image. Such systems have the drawback that the optical path creates the need for a relatively thick receptor, which cannot be retrofitted into existing machines. Other DR systems, such as those from Hologic, use selenium on top of a transistor array, and still others, such as those from Trixell and Varian, feature a scintillating screen on top of silicon on top of a transistor array. These systems tend to be more expensive because of the difficulty of manufacturing large blemish-free arrays. Edge Medical uses selenium but sidesteps the difficulty of making large arrays by mechanically scanning the charged selenium layer, an approach that may lead to an effective DR technology at a fully competitive price.

Controversy over digital mammography

Still controversial is digital mammography, which for the most part is done using film/screen detectors. Where most medical images are adequately resolved using 100-micron to 200-micron pixels, mammography is thought to require 50-micron pixels. Smaller pixels are more subject to noise, and all digital systems use digital processing to reduce the noise and to enhance features indicative of disease.

Digital mammography systems are available from several manufacturers, including GE Medical Systems (Waukesha, Wisconsin), Fischer Imaging (Denver, Colorado) and Hologic. A comparison of commercial systems conducted by Dr. Etta Pisano on behalf of the American College of Radiology (Reston, Virginia) did not produce definitive results. Pisano complained that she was hampered by manufacturers' unwillingness to disclose details on their processing algorithms. She said she felt that manufacturers were guided more by aesthetic considerations than by diagnostic clarity. In any case, the results were hard to interpret, being dependent on the machine, the readers, the disease and the algorithms.

A large part of the impetus toward digital X-ray systems comes from the requirements for PACS (picture archiving and communication systems, as medical image management systems have come to be known). There is a bewildering variety of PACS available from literally 100 suppliers. The situation is reminiscent of introduction of hospital information systems. That field is now dominated by a handful of large companies Cerner (Kansas City, Missouri), IDX (Burlington, Vermont), McKesson (Alpharetta, Georgia), Meditech (Westwood, Massachusetts) and Siemens Medical Systems (Erlangen, Germany), which bought Shared Medical Systems (Malvern, Pennsylvania). All of them were present at the RSNA meeting. But in addition to the large companies, there are 100 or so smaller companies with more specialized products that gained a foothold in the early days of information systems and have managed to hang on to their original customers because users find that changing systems is so disruptive and costly.

There are similarities with PACS. The large equipment suppliers including GE Medical, Philips Medical Systems (Bothell, Washington), Siemens Medical Systems and Toshiba America Medical Systems (Tustin, California) all offer full-scale PACS as do the large film companies Agfa, Eastman Kodak and Fuji. But in addition to these, there are dozens of other smaller suppliers whose businesses are surviving and in some cases flourishing.

Smaller companies find this an easy field to enter because most of the initial costs lie in software development, which is inexpensive compared with hardware development costs. The companies include Algotec (Raanana, Israel), Amicas (Brighton, Massachusetts), BRIT Systems (Dallas, Texas), DeJarnette (Towson, Maryland), DR Systems (San Diego, California), eMed (Lexington, Masachusetts), IMCO (Pewaukee, Wisconsin), Merge eFilm (Milwaukee, Wisconsin), Stentor (Brisbane, California), StorComm (Jacksonville, Florida) and many others. Despite fierce competition, which makes it exceedingly difficult for each of these small companies to make itself known and to distinguish itself in the midst of the cacophony of competing claims, many of them are likely to survive in niches that they find for themselves. The bulk of the market will go to the larger companies, as it has in the case of hospital information systems, but there still will be a place for the smaller companies.

Related to digital X-ray is computer-assisted diagnosis, pioneered by R2 Technologies (Sunnyvale, California), based on work at the University of Chicago (Chicago, Illinois). CAD has been a long time in gestation as researchers hunted for reliable clues that could be recognized by computer analysis of images to enhance diagnostic accuracy. Now there is more and more evidence that CAD systems make a measurable improvement in the accuracy of interpretations for breast cancer, notably in radiographically dense breasts, typical of younger women, as reported by Dr. Rachel Brem. Beside R2 Technologies, CADx (Laval, Quebec) and iCAD (Hudson, New Hampshire) offer systems, all of which work by having a computer indicate suspicious areas on a digitized version of a mammogram. To be useful, a system must not have too many false positives (finding evidence of disease when disease is absent) nor too many false negatives (failing to find disease when it is present). A paper given by Dr. Michael Nelson evaluated systems from the three manufacturers for their ability to detect breast cancer and concluded that there were no significant differences among them. R2 is suing the other two companies, claiming infringement of its patents.

Technical developments continue in MR

In digital modalities, magnetic resonance (MR) continues to unroll more and more technical developments. The three major equipment suppliers GE, Philips and Siemens all offer 3-tesla systems, which yield better images and permit more sophisticated examinations like functional MRI. While these systems offer some dramatic images, they are expensive, and the market is mainly concerned with systems at 1.5 tesla and below. All the major manufacturers the three just mentioned plus Hitachi (Twinsburg, Ohio) and Toshiba offer 1.5 tesla systems as well as so-called open systems, which use permanent magnets to obtain the magnetic field. Now that magnetic resonance can be said to be a maturing technology, there is a convergence in the nature of the products offered by each manufacturer.

The dream of every manufacturer is that its technology be adopted and reimbursed for screening applications. Now MR is being touted as a tool for breast cancer screening in women at high risk. Well, maybe. The studies came from Germany and Italy, and they indicated that contrast-enhanced MR or a special technique called MR elastography achieved better accuracy than CT or mammography. While the findings may be valid, the economics of MR clearly mitigate against the conclusion that MR is suitable for screening.

CT is another mature technology, but it continues to spawn innovations. CT is available from the same manufacturers as MR, although Hitachi does not market CTs in the U.S. The driving force behind CT technology has always been to reduce scan time, most recently by using spiral scanning and multiple rows of detectors. Reduced scan time may increase throughput, but more importantly it improves image quality by allowing collection of redundant data, thus reducing the effects of noise, and it permits thinner slices. It also brings into better focus objects that move, and in the body almost everything moves a bit as blood pulses through the arteries. This year GE introduced a 16-slice CT, while Philips and Siemens each had 10-slice CT systems.

An area of intense development is virtual colonoscopy using CT. The slice images from CT can be recombined to create images as they appear to a colonoscopist watching the image from a colonoscope as it is advanced through the colon. CT has the virtue of being less invasive but it generally has proved less sensitive than an actual colonoscopy and it has the drawback that therapy that is, removal of polyps must be done in a subsequent real colonoscopy. Researchers from the University of Chicago reported that the sensitivity of CT virtual colonoscopy could be improved using CAD on volume-rendered 3-D images, thus making virtual colonoscopy more competitive. Inasmuch as there are far too few gastroenterologists to do the all the screening colonoscopies now thought to be desirable and the fact that colonoscopy is an expensive procedure compared with CT, virtual colonoscopy may well earn a place in diagnostic medicine.

Because of the existence of fast CT, doctors are tempted to use it frequently, but two warning voices were heard at RSNA. One was a caution from Philip Goodman at Duke University (Durham, North Carolina), who pointed out that X-ray exposure from lung CT is many times greater than that from a chest X-ray, to which the procedure had been compared. Similar considerations apply to total-body CT scans, which are being promoted by some CT operators as a suitable screening procedure. According to a survey conducted in Boston, 80% of respondents, who knew next to nothing about the procedure, said they thought it sounded like a good idea.

Recently, the National Cancer Institute (Bethesda, Maryland) and the Society for Pediatric Radiology (Houston, Texas) mailed a warning to radiologists and pediatricians against using CT indiscriminately for children with head injuries because of the high dose and low yield of positive results. This is a concern likely to be heard with increasing frequency.

PET's growth story rolls on

A remarkable development during the last three or four years has been the growth of positron emission tomography (PET). Two major factors underlie the growth first, the willingness of the Centers for Medicare & Medicaid Services to pay for the procedures, and second, the proliferation of nuclear pharmacies that produce the fluorodeoxyglucose now used for most PET procedures.

In addition, the major CT manufacturers also provide PET machines combined with CT so that a clinician gains the functional information provided by PET combined with the anatomical detail provided by computed tomography. At the RSNA gathering, PET was reported to be useful for assessing stroke and coronary artery disease, for detection of various cancers including those of the brain, breast, lung, cervix, and testicles. Dr. Barry Siegel of the Mallinckrodt Institute of Radiology (St. Louis, Missouri) pointed out the value of PET in therapy planning and in obtaining early information about response to chemotherapy.

A move into mainstream for ultrasound

Until a year or two ago, ultrasound dwelt in a world largely apart from the mainstream of radiology. Today, thanks to acquisitions, the three main ultrasound suppliers are the same as the three main suppliers of CT and MR GE, Philips and Siemens. Toshiba, Hitachi and SonoSite (Bothell, Washington) also provide ultrasound equipment. Philips had long been almost entirely out of the ultrasound business, but with the acquisition of ATL and Agilent (formerly Hewlett-Packard Medical Systems), Philips now looks to be the largest ultrasound supplier.

Ultrasound technology is completely digital, and digital processing of both the outgoing ultrasound pulses and the received echoes continues to enhance the quality of images, though in subtle rather than dramatic ways.

With the inherent advantages of being inexpensive, portable, harmless and easily installed, ultrasound has a firm grip on its standard uses in cardiology, obstetrics, gynecology, urology, certain vascular studies and abdominal studies, ultrasound-guided biopsies and brachytherapy are commonplace.

Most ultrasound machines are the size and weight of a washing machine. They can be rolled from place to place but not carried. SonoSite offers a hand-held machine that received FDA clearance last May. The company asserts that it can be used to do ultrasound-guided procedures at the bedside. GE also introduced a notebook-sized ultrasound machine.

Virtually none of the products discussed above would exist if it were not for modern digital technology. More than any other medical specialty, radiology has been the beneficiary of this revolution. As a result, the profession is vastly understaffed in both technologists and radiologists. That is a boon for those already in the profession, but a hurdle for those who train and hire them and a potential problem for everyone who uses imaging services namely, all of us.