BBI Contributing Editor
CHICAGO, Illinois — As much as any medical specialty, radiology has been the beneficiary of technological developments born of worldwide industrial competition, government and private research, and collaboration with practicing radiologists, mainly in academic centers. This symbiosis has produced remarkable advances in medical imaging, and a great deal of attendees' time during the annual gathering of the Radiological Society of North America (RSNA; Oak Brook, Illinois) is usually devoted to learning about the newest offerings from vendors. More than 600 vendors — only slightly less than the previous year — had their wares on display during the late-November RSNA meeting, held as usual in the mammoth McCormick Place convention center.
Though medical imaging is still vibrant with technical advances, it must be said that this meeting — the society's 87th such event — showed little that was genuinely novel. There are no new methods for imaging with the potential that computed tomography (CT) had in the 1970s or magnetic resonance (MR) in the 1980s. Instead, the gathering revealed steady improvements in all the modalities. Attendance at this year's meeting, pegged at about 55,000, was down by an estimated 12%, attributed partly to post-Sept. 11 travel fears, but also to limitations in hospital budgets, especially at teaching hospitals. The meeting included 2,100 research presentations and posters covering 15 subspecialties, as well as 283 refresher courses and 1,100 educational exhibits.
In recent years there have been important rearrangements of the corporate landscape of medical imaging. The leading manufacturers were GE Medical (Waukesha, Wisconsin), Siemens Medical (Erlangen, Germany) and Toshiba (Tokyo), offering a full line of imaging equipment for X-ray, CT, MR, ultrasound and nuclear medicine. The other important companies had nearly full lines, but lacked one or two of the modalities. Philips (Eindhoven, the Netherlands) lacked ultrasound and nuclear medicine; Hitachi (Tokyo) offered only MR and ultrasound in the U.S.; Marconi (Cleveland, Ohio) lacked ultrasound. Despite the power of these companies, the ultrasound market was dominated by a different set that included Agilent (formerly Hewlett-Packard; Andover, Massachusetts), Acuson (Mountain View, California) and ATL Ultrasound (Bothell, Washington). Now all that is changed.
In 1998, Philips acquired ATL, thus establishing itself as an important player in ultrasound, and in 2001 Philips also acquired Agilent, extending its ultrasound line and adding patient monitoring equipment. Philips went on to buy ADAC Laboratories (Milpitas, California), a maker of nuclear medicine equipment. Late in the year, Philips acquired Marconi (formerly Picker, a subsidiary of General Electric Company of England, no relation to GE in America). That deal expanded its X-ray, CT, and MR product lines and, with ADAC, added nuclear medicine, a field from which Philips had been an absentee. Siemens acquired Acuson. And so it has come about that Philips, whose medical division had once been referred to by Royal Philips' chairman as "disposable," is now a first-rank player alongside GE Medical and Siemens, while the formerly more-independent market for ultrasound is now the province of the major imaging companies.
This is a pattern seen over and over again in this field. At one time, in the years after CT was introduced, there were 25 companies — many of them upstarts — that offered CT systems, but ultimately the only important CT companies were the former X-ray companies. An almost identical pattern was observed 10 years later with MR. And now it is true of ultrasound as well. And it may come to be the case with picture archiving and communication systems (PACS), which now can be considered a sixth part (in addition to the five imaging modalities) of the medical imaging field. This matter will be treated in the second installment of our RSNA report next month.
Judah Folkman's lecture
The truly novel revelations at the RSNA meeting lay in speculations about developments for the future, the most exciting of which was an invited lecture by Judah Folkman, MD, a researcher at Children's Hospital (Boston, Massachusetts). Folkman conceived, and eventually demonstrated, the pivotal importance of angiogenesis in the growth of cancerous tumors, thus opening up a totally new avenue for treatment. Angiogenesis describes the ability of cancerous cells to stimulate formation of new blood vessels necessary for a tumor to grow. Unlike normal blood vessels, these vessels are poorly formed and leaky, a feature that is routinely exploited in X-ray, CT and MR, where contrast agents leak more quickly into the extracellular space in tumors than in normal tissues, making tumors easier to spot in an image. Other clues include the necrotic (dead) centers of tumors no longer fed by blood vessels. This effect was thought to be just a characteristic of tumors, but Folkman asserted that it is attributable to increased pressure from leaky vessels on the periphery of the tumor, which compress blood vessels near the center, thus starving the center of nutrients.
Folkman pointed out that very early tumors 1 mm to 2 mm in diameter are invisible with current imaging techniques, not only because they are small but also because they have not yet produced an angiogenesis-stimulating product that allows them to grow. If it were possible to detect angiogenic foci before they have begun to grow, to distinguish newly created vessels from established ones, and to recognize early regression of vessels under treatment, cancer detection and therapy would be greatly enhanced.
In practice, Folkman's elegant approach has produced mixed results. Some experiments with angiogenesis-inhibiting products, called angiostatins, have produced spectacular regressions of cancers that had resisted other forms of treatment, but other experiments have been disappointing. It turns out that for some reason tumors produce both angiogenesis-stimulating factors, which appear to be necessary to induce metastasis, and angiogenesis-inhibiting factors, thus confounding the effects of exogenous angiostatins. These factors have different half-lives in vivo, further confusing the picture.
Still, Folkman is optimistic that study of these factors will elucidate at least some of the mechanisms of metastasis and lead to effective avenues for treatment, particularly if there were techniques that enabled the detection of early effects. He said that there are presently some 20 angiogenesis inhibitors in clinical trials, seven of which have reached Phase III. Because these compounds are either naturally occurring or closely related to natural ones, they have none of the harsh side effects typically associated with chemical treatments for cancer.
Finally, Folkman observed that the concept of angiogenesis with its circulating factors that produce new blood vessels in cancerous tumors is applicable to other proliferative diseases, including macular degeneration in ophthalmology, endometriosis in gynecology, psoriasis in dermatology, and certain forms of vascular disease in cardiology. It is possible that when angiogenesis is well understood, it will point the way to therapies for these diseases as well.
Multidetector helical CT
Turning to technical developments actually on display on the sprawling McCormick Place exhibit floor, the most spectacular was the evident growth of multidetector helical CT. Pioneered by Elscint (Haifa, Israel) before the division was sold to GE, multidetector systems are now offered by all the major manufacturers, including GE, Siemens, Philips and Toshiba. Traditional CT systems have a single row of detectors and collect data for a single cross-sectional slice with each rotation of the X-ray source. Elscint put in a second row of detectors, thus collecting data for two slices with one rotation of the source, in effect cutting in half the time necessary to collect data for a full study. Once the idea had gained a foothold, it was a short step to increasing the number of detectors by larger factors — to 256 in a scheme shown by Toshiba as a works-in-progress.
In traditional CT systems that collect data slice by slice, the X-ray beam is collimated to a "fan beam," that is, a shape like a Japanese fan partially opened to an angle around 60 degrees, and just a few millimeters thick in the axial direction. When there are multiple rows of detectors, the collimation of the X-ray beam must be widened in the axial direction so that the beam more nearly resembles a pyramid. A little reflection reveals that the irradiation pattern when the X-ray source rotates around the patient is related to the slices that make up the study in a complicated way. The key to doing the mathematical reconstruction of images is fast computation, the essential ingredient previously missing. Now that very fast computers are available and inexpensive, this development became possible.
High-speed data collection using CT is of some value in terms of throughput, but its real value lies in making it possible to images organs that move, like the lungs, or to image organs of large extent, like the colon or peripheral blood vessels. Thus multidetector CT has opened CT to new examinations, and a number of papers presented at the meeting attested to its efficacy in these new roles.
One that is of special interest to radiologists is virtual colonoscopy, wherein the CT data are reconstructed into a 3-D image of the interior of the colon that appears as it would to an endoscopist. Colon cancer is one of the leading causes of death. The cancers typically grow from precancerous polyps, but the progression can be interrupted if the polyps can be detected and removed. For this reason, many doctors believe that colonoscopy should be a regular screening procedure for people over 50. But there are problems with this recommendation. Colonoscopy is expensive and time-consuming. It is not a pleasant examination, and it carries some risk of perforating the intestine. Additionally, there are not enough internists to perform the procedure on so many patients.
Under these circumstances, virtual colonoscopy using CT seems an attractive alternative. It requires the same purging regimen as colonoscopy, which most patients find very unpleasant, but the examination itself is simple and safe. The question has been whether CT finds polyps as reliably. Papers presented at this meeting suggest that it does. But there remain some difficult hurdles to making virtual colonoscopy a standard substitute for real colonoscopy. In real colonoscopy, a snare can be used to extract any polyps as soon as they are seen, whereas virtual colonoscopy must be followed by real colonoscopy to extract polyps unless none were found. In addition, because internists earn a good deal of money performing colonoscopies and control the patients, they may be reluctant to refer them to radiologists. The issue remains unsettled.
Fast MR imaging
One reason for the growing importance of magnetic resonance imaging is its technical richness — its ability to make images based on a variety of different tissue characteristics, thus making it possible to distinguish tissues that would otherwise be indistinguishable. Just by changing the sequence of excitation and readout, MR images may be made sensitive to the MR parameters T1 or T2, to concentration, flow, diffusion or perfusion. Exploring all the possibilities opened up by these capabilities will take years.
MR also has available techniques for collecting data rapidly, and, as in the case of CT, this speed can be used not only to increase throughput, but also to image moving organs or extensive organs. MR now has a role in examining the arteries, where it has the advantage of not needing a contrast agent as CT does. Fast MR means that it is possible to image extended organs like the arteries in the leg. Speed can be used as well to collect more data for images sensitive to several different characteristics of tissue, thus enhancing its diagnostic value without extending the time to perform the examination.
MR also has an important place in research, and one application is functional MR imaging (fMRI), a technique for discovering which areas of the brain are active during reception of a stimulus or performance of a physical or mental task. An intriguing finding reported at RSNA, based on research at the University of Rochester (Rochester, New York), involved using fMRI to see which areas of the brain were active during personal decision-making. Subjects were asked first to choose the better of two desirable events (such as taking a warm bath or eating a good meal), or to choose the worse of two undesirable events (such as being in a car accident or being robbed). During the first scan, they were instructed to answer the question based on how their choice would affect them personally, but in a second scan they were instructed to consider the question dispassionately, based solely on cost.
The scans showed that the prefrontal lobe, known to be the part of the brain typically involved in emotions, was active when making the personal decision, but it remained nearly equally as active when making the supposedly rational decision. Even rational decisions involve emotions. The finding may not be of much practical importance, but it is an indication of the increasing sophistication provided by these new windows into the mind.