Computed tomographic angiography (CTA) is a valuable imaging modality for the evaluation of the vascular system. During the past decade, several studies have demonstrated that CTA has important advantages over the gold standard of conventional angiography, including increased safety, greater patient acceptance, lower cost, and reduced radiation exposure. ¹ Consequently, CTA has gained increasing attention as a potential minimally invasive diagnostic alternative to conventional angiography.

The development of multislice spiral CT technology has represented a significant advancement, particularly in the case of CTA studies of mesenteric vasculature, where distal branches may also be easily evaluated. First, multislice spiral CT allows for acquisition times up to eight-fold faster than conventional single-slice spiral CT scanners; this feature enables larger anatomic coverage, and diminishes misregistration and respiratory motion artifacts. Second, thin-slice collimation protocols can be implemented routinely with greatly enhanced spatial resolution along the longitudinal axis, which provides virtually isotropic 3D voxels; consequently, image quality is improved with better diagnostic capabilities. Third, optimization of contrast enhancement is now possible due to more accurate timing and better separation between the arterial and venous phases. However, in order to entirely exploit the benefits of the use of multislice CT, it is of paramount importance not only to optimize new study protocols but also to improve image display. Three-dimensional datasets are reconstructed and evaluated interactively on dedicated workstations using axial and two-dimensional multiplanar images, as well as more complex rendering techniques, such as maximum intensity projections (MIP), surface-shaded displays (SSD), and volume rendering. ²

Multislice CTA technique

In our experience, images were obtained using a multislice spiral CT scanner (Somatom Plus 4 Volume Zoom; Siemens Medical Solutions, Erlangen, Germany) equipped with flying spot, adaptive array matrix, and a gantry rotation time of 0.5 sec. ³ Before the study, patients received 800 mL of water as an oral contrast agent in order to produce negative contrast in the stomach and small bowel. A total of 140 mL of nonionic iodinated contrast medium (Xenetix 350; Guerbet, Aulnay-Sous-Bois, France) were infused intravenously at a rate of 4 to 5 mL/sec into a peripheral, typically antecubital, vein. Following a bolus injection of 20 mL of contrast medium, sequential dynamic slices were acquired at the level of the superior mesenteric artery origin in order to evaluate the optimal delay scanning time. Two spiral CT scans of the abdomen and pelvis were obtained at the arterial phase (delay calculated on bolus test injection) and portal venous phase (delay of 60 to 70 sec). Images were acquired using 1-mm slice collimation, 1-mm slice thickness, 6 mm/sec table feed, 1.0-mm reconstruction interval, 120 mAs, and 120 kVp. Breath-hold time ranged between 30 and 35 sec, depending on the size of the patient. Patients were instructed to hyperventilate before scanning and to exhale slowly if they could not suspend respiration for the entire examination time.

Since the use of a 1-mm slice collimation protocol produced a considerable number of images (300 to 400 images per scan per patient), we limited the image reconstruction interval to 1 mm, with no image overlapping. In addition, since a lower flow rate could have impaired optimal opacification of smaller arterial vessels due to inadequate intravessel concentration of contrast material, contrast medium was injected at a flow rate of at least 4 mL/sec.

Once acquired, images were downloaded to an off-line dedicated workstation (Kayak PC workstation; 2 parallel 700 MHz kernel processors, 1024 Mb RAM memory; Hewlett Packard, Palo Alto, CA). Image analysis was performed using Vitrea 2.2 (Vital Images, Minneapolis, MN), a software package with volume-rendering capabilities.

Selective vessel representation was obtained using different rendering curves. A panoramic overview of the entire main abdominal branches could be observed using a preset opacity curve showing only the vascular surface. The evaluation of minor vessels (i.e., second, third, and more distal orders of collateral branches) required the analysis of 3D data sets using interactive multiplanar cut planes ("oblique trim") and by modulating the opacity of the anatomical structures under evaluation and window/level parameters in order to see vessels "through" abdominal organs. Image analysis required direct operator interpretation at the workstation, where 2D axial and multiplanar reconstructions, and 3D images were available simultaneously on a 21-inch monitor. A complete analysis of arterial and venous vessels required a mean interpretation time of 20 minutes.

Anatomic representation

Arterial vascular system

The superior mesenteric artery (SMA) is identified in its entirety, including its origin and its major branches (Figure 1). SMA usually arises <1.5 cm below the origin of the celiac trunk. Major collateral branches are represented by 1) the inferior pancreaticoduodenal artery, which has an oblique course, anastomosing superiorly with the superior pancreaticoduodenal artery, a branch of the gastroduodenal artery; 2) the middle and right colic arteries, with the latter absent in up to 80% of normal individuals, supplying blood to ascending colon; 3) the jejunal branches, arising from the left side of the SMA; and 4) the ileocolic artery and ileal branches arising from the right side of the SMA, supplying blood to the ileum and cecum.

Anatomic variants such as an anomalous origin of the right hepatic artery from the SMA can be depicted easily (Figure 2).

The inferior mesenteric artery (IMA), which arises from the aorta approximately 7 cm below the origin of the SMA, can be visualized together with the major branches (the left colic artery, which has a straight superior course, supplying the transverse and descending colon; the sigmoid arteries, two to four vessels supplying the sigmoid colon and the superior hemorrhoidal artery, terminal branch for the upper rectum) (Figure 3).

Venous vascular system

All major venous vessels (portal vein, splenic vein, superior mesenteric vein, and inferior mesenteric vein) can be depicted accurately, as are the collateral branches (Figure 4). The SMV is usually a single trunk, receiving blood from the middle and right colic veins, the ileocolic vein, the gastrocolic vein, and from jejunal and ileal branches. The IMV receives blood from the left colic vein, the sigmoid veins and the superior hemorrhoidal vein.The major problem when evaluating the venous system was discerning small arteries from veins, both of which were opacified on delayed images. Correct image analysis required additional interpretation time on the workstation.

Pathologic conditions

Pathologic conditions where involvement of mesenteric vessels has to be evaluated are represented mainly by pancreatic cancer and mesenteric ischemia. ⁴ Infiltration of SMA is a known contraindication to pancreatic surgery. Simultaneous evaluation of axial and multiplanar reformatted planes has been demonstrated to be useful for assessing vascular invasion and in particular sagittal planes (Figure 5). It can also provide excellent evaluation of the posterior fat plane between the SMA and the abdominal aorta, which is especially useful to rule out vascular encasement. A more precise evaluation of the SMV and the portal vein is also extremely important, since a vessel infiltration <2 cm is considered a criterion for performing pancreatic surgery with a vascular graft. Again, multiplanar reformatted images (along coronal or coronal-oblique planes), provide excellent evaluation of possible vascular infiltration.

In the case of mesenteric ischemia, multislice spiral CTA may provide all necessary diagnostic information before planning adequate treatment. Multislice spiral CTA can demonstrate the underlying cause if the ischemia, which might be represented by narrowing or occlusion of the SMA (due to either atherosclerotic plaque or thrombus or neoplastic vascular infiltration), or by thrombosis or neoplastic encasement of SMV.

Finally, multislice spiral CTA might have a role in the assessment of vascular changes in cases of inflammatory bowel disease, small bowel tumors (i.e., carcinoid, lymphoma), and, less frequently, pathologic conditions (i.e., metastases, sclerosing mesenteritis, etc.).

Conclusion

With its greatly enhanced image resolution, multislice CTA seems able to overcome the technical challenge of depicting small mesenteric vessels. It provides a detailed anatomical view similar to that of conventional angiography but with the possibility of examination from innumerable viewing angles. Technical requirements include interactive evaluation on a workstation using dedicated reconstruction algorithms. Volume-rendering is the most suitable reconstruction technique, due to its ability to display the 3D spatial relationships among vessels and surrounding organs. Widespread applications for these techniques can be expected in the near future with the greater availability of multislice CT equipment, improvement in workstation design and performance, and specific education of radiologists in the use of 3D software. A cost-benefit analysis must be performed in order to better evaluate the impact of this new imaging modality on patient outcomes.

Although further studies are necessary to rigorously compare the results of multislice CTA with conventional angiography in terms of diagnostic accuracy, an increased use of this imaging modality in the study of mesenteric vessels can be anticipated, with multislice CTA progressively obviating the need for diagnostic conventional angiography.

References

1. Rubin GD, Shiau MA, Schmidt AJ, et al. Computed tomographic angiography: Historical perspective and new state-of-the-art using multi detector-row helical computed tomography. J Comput Assist Tomogr . 1999;23(Suppl1):S83-S90.

2. Johnson PT, Heath DG, Kuszyk BS, Fishman EK. CT angiography with volume rendering: Advantages and applications in splanchnic vascular imaging. Radiology. 1996;200:564-568.

3. Laghi A, Iannaccone R, Catalano C, Passariello R. Multislice spiral computed tomography angiography of mesenteric arteries. Lancet . 2001;358:638-639.

4. Horton KM, Fishman EK. Volume-rendered 3D CT of the mesenteric vasculature: Normal anatomy, anatomic variants and pathologic conditions. RadioGraphics . 2002;22:161-172.

Optimizing Multislice CT for Imaging with the Goal of 3D Postprocessing

William J. Davros, PhD, ABMP(D)

The advent of multislice CT (MSCT) has introduced in a new era in medical imaging. High-quality cross-sectional images can be gathered faster than ever before without sacrificing quality. The new MSCT units are more complex than previously offered units but also hold out the promise to be more flexible. It has also opened the possibility of more accurate diagnosis by radiologists, CT-based screening programs, integrated CT radiotherapy treatment planning, and three-dimensional (3D) volume-rendered processing for the purpose of pre-operative planning. This article will focus on optimizing MSCT image data collection and reconstruction for 3D volume-rendered processing.

Optimizing MSCT can be divided into two categories: optimization of scanning parameters and optimization of image reconstruction parameters. Scan parameters are set prior to the initiation of raw data collection and can be preprogrammed by the vendor or stored as site-specific protocols. Regardless of the selection process, they are a major determining factor in the quality of a postprocessed product. Reconstruction parameters can be set before or after scanning is completed. These parameters control the look of the final image and, therefore, also play an integral role in the quality of the final product.

Parameters available for optimization are X-ray tube potential (kV), the X-ray tube current (mA), the scan time per gantry revolution (s), and the X-ray detector configuration. The tube potential is the most important parameter in controlling the amount of image contrast. Lower X-ray tube potential will augment image contrast, especially when iodinated contrast medium is used. This is due to the fact that the absorption of X-rays is greater at lower photon incident energies. In practical terms, 80 kV will produce superior image contrast than 120 kV, and 120kV will produce greater image contrast than 140 kV. Most facilities use 120 kV because it is a good compromise between image contrast and penetrability of the photons. The X-ray tube current-time product (mAs) is proportional to the number of photons. When more photons are used, lower image noise will result. Image noise hampers postprocessing by making the job of discriminating between two structures whose CT numbers are close together more difficult. Detector configuration is the scan parameter that, in part, determines the final image thickness. When scanning with the thought of postprocessing, it is recommended to choose a detector configuration that will permit very thin image reconstruction. These scanners will not permit image thickness to be less than the smallest z-axis dimension of the detector configuration chosen. For example, if a 1-mm image thickness is desired, the user might select the 0.5 mm * 2 or 4 * 1 mm configurations, but not the 4 * 2.5 mm configuration.

Once raw data has been gathered by scanning, images need to be reconstructed. The parameters available for image reconstruction are image thickness, image-to-image spacing, reconstructed field-of-view (FOV), and reconstruction kernel.

* Image thickness is a measure of the z-axis distance over which scan projections are combined to produce a two-dimensional image. Thinner image thickness usually produces more pleasing postprocessed work than do thicker images.

* Image-to-image spacing is a measure of the z-axis distance between image centers. Overlapped images exist if the image-to-image spacing is less than the image thickness. Contiguous images exist if the image-to-image spacing is equal to the image thickness. If the image-to-image spacing is greater than the image thickness gaps between images exist. Optimal postprocessing is achieved when the image-to-image spacing is about one-half to one-third of the image thickness. For example, if the image thickness is 7 mm, the image-to-image spacing should be 2.5 to 3.5 mm. For the sake of simplicity many sites use integer values for image-to-image spacing by rounding, but never using a value larger than one-half of the image thickness.

* The reconstructed FOV is the parameter that determines the pixel dimension in the axial imaging plane. The pixel dimension is equal to the FOV in millimeters divided by 512. For example, an abdominal image may have been reconstructed with a 330-mm FOV, in which case the pixel size is 0.625 * 0.625 mm. In general, 3D postprocessed images have a greater potential to show fine structures when the pixel size is small. This leads to a recommendation that the FOV be as small as possible consistent with viewing all necessary anatomy.

* The reconstruction kernel is a mathematical device that is used to control how sharp edges will be displayed in the final image set, as it controls how much image noise will be displayed. Smoother kernels used in soft-tissue imaging display less noise and thus the images have a soft look to them. The main advantage to this type of image is the ability to see low-contrast lesions, such as a liver tumor surrounded by normal liver tissue, in a non-contrast scan. Detail or bone kernels display sharp edges with high fidelity but display a great deal of noise. In this type of imaging, the noise is less of a problem in that the scan was presumably done to image fine detail such as bone fractures or high-resolution lung imaging. Postprocessed images tend to look best when the kernel is at the smoother end of the available choices. Surface-shaded models and volume-rendered models have a more pleasing surface appearance because the inherent low noise in the 2D images translates to less pitted surfaces on 3D models.

Postprocessed imagery is a powerful tool in the arsenal of medical imaging. It permits surgeons to see anatomy in a familiar frame of reference. Optimizing multislice CT for postprocessing is the key to unlocking the potential of postprocessed imagery.

Indications and Utilization of Abdominal CT in the Emergency Department

Max P. Rosen, MD, MPH

The advent of multislice CT (MSCT) scanners has lead to an expansion of the application of CT for suspected abdominal pathology. One area in which it has had its greatest impact is in the Emergency Department (ED). Our experience at a level 1, adult, academic trauma center shows the number of abdominal CT scans performed has increased from 24.56 per 1000 ED visits in 1999 to 28.34 per 1000 ED visits in 2000. The "top 5" indications for abdominal CT in our ED are: 1) suspected appendicitis (17%), 2) renal colic (15%), 3) suspected abscess (12%), 4) suspected diverticulitis (10%), and 5) suspected small bowel obstruction (7%). Between 1997 and 2000, the proportion of abdominal CT scans performed in our ED for suspected bowel pathology has increased from 37% to 50%.

The proliferation of CT for suspected bowel pathology is likely due to a combination of the publication of several studies demonstrating the sensitivity and specificity of CT for diagnosing the above abdominal conditions. A Medline search of the terms "appendicitis" and "CT scan" for 1990 yielded only 8 articles. This increased to 25 in 1995, and 79 in 2000. We have found that advances in imaging technology published in the literature translate into an increase in clinical requests for an imaging study within 1 year of the publication.

In addition, the maturation of CT technology has paralleled the penetration of managed care. In the past, many patients with abdominal pain would have been admitted to the hospital for observation. However, managed care has placed a premium on avoiding hospital admission. As one of the major functions of the ED is to triage patients between hospital admission and outpatient management, any technology that enhances the triage ability of ED physicians is especially valuable in the managed care environment.

We recently studied our experience with 536 consecutive patients with non-traumatic abdominal pain who underwent a CT scan in the ED. Prior to ordering the CT, each physician was asked to identify his/her planned management. We then followed each patient and studied his/her actual management. Prior to performing the CT, 402 of 536 (75%) would have been admitted to the hospital. However, following CT, only 312 of 536 (58%) were actually admitted. The net effect of performing CT was to obviate the need for hospital admission in 90 of 536 (17%) of patients. Prior to CT, 67 of 536 (13%) of patients were thought to need immediate surgery. Only 25 of 536 (4.6%) actually underwent surgery. Of the 67 patients thought to need surgery, only 25 (37%) actually required immediate surgery. Among patients with the four most common pre-CT diagnoses (appendicitis, renal colic, abscess, and diverticulitis), CT had the greatest impact on hospital admission rates (28% reduction) and surgical management (40% altered management) for patients with suspected appendicitis.

The apparent reliance on CT to evaluate patients with suspected abdominal pain in the ED likely reflects the high sensitivity and specificity of CT and the elusive nature of clinical diagnostic certainty in assessing patients with abdominal pain (Figure 1). Among a group of 57 patients who presented to our ED with abdominal pain, we found that CT had a sensitivity and specificity of 96% and 86.5%, respectively. ¹ However, our approach to evaluating patients with abdominal pain includes an initial assessment by an intern or resident, who are often uncertain of their clinical diagnosis. For example, when asked to rate their level of certainty in their clinical (pre-CT diagnosis), the mean level of certainty expressed by our interns and residents was only 2.4 on a 5-point scale (1 = least certain, 5 = most certain).

This lack of certainty in their clinical diagnosis is supported by the findings of another study conducted at our hospital in which we reviewed 100 consecutive CT scans performed for suspected appendicitis. ² Among these 100 patients presenting with right lower quadrant pain, CT correctly identified 23 of 24 (95.8%) patients with surgically proven appendicitis. However, CT also identified a varied constellation of other pelvic conditions that cause right lower quadrant pain including: uterine/adnexal pathology (N = 16), diverticular disease (N = 9), inflammatory bowel disease (N =3 ), urinary tract disease (N = 2), and mesenteric adenitis (N = 2). In addition, CT identified pathology above the level of the iliac crest in an additional 7 patients. In the days prior to cross-sectional imaging, it is safe to say that the vast majority of these patients would have required hospital admission and either would been observed as their physician waited for the etiology of the abdominal pain to "declare" itself, or would have undergone to exploratory surgery.

Conclusion

We have found that the use of CT to evaluate patients with suspected abdominal pain on our ED continues to increase. This is likely due to the rapid evolution of MSCT technology, the high sensitivity and specificity of CT for accurately identifying the etiology of abdominal pathology, and the need for rapid triage in the ED under managed care.

References

1. Rosen MP, Sands DZ, Longmaid HE, et al. Impact of abdominal CT on the management of patients presenting to the emergency department with acute abdominal pain. AJR Am J Roentgenol. 2000;174:1391-1396.

2. Kamel IR, Goldberg SN, Keogan MT, et al. Right lower quadrant pain and suspected appendicitis: Non-focused appendiceal CT­­Review of 100 cases. Radiology. 2000;217:159-163.

New Screening Technologies

Judith V. Douglas, MS, MPH

The scientific community is now questioning the assumption that early detection and treatment hold value in all instances. Researchers are reviewing evidence from studies of screening for different diseases, asking whether tests, even when relatively inexpensive and non-invasive, save lives and in fact "do no harm." In addition to the furor surrounding mammography and the PSA test, questions are arising regarding other procedures used to diagnose cancer and other diseases.

Some procedures are relatively untested, although market forces are clearly involved. In 2001, for example, one manufacturer of electronic beam computerized tomography (EBCT) used to diagnose heart disease reported sales of $19.3 million for one quarter alone. ¹ The same year, a newly opened "virtual physical" center in Baltimore reported that its first 57 scans generated 27 recommendations for patients to see a doctor for follow-up, including 12 cardiology referrals. ² To date, however, there is slim support among the medical establishment for full-body scans. Offered by freestanding entrepreneurial centers to those willing and able to pay out of their own pockets, virtual physicals remain outside routine care.

Since colon cancer is the second leading cancer killer in the United States and is 90% preventable, virtual colonoscopy may become part of routine care. Although traditional colonoscopy has widespread support among medical professionals as screening for patients aged >=50, many patients avoid it as unpleasant. Less invasive, virtual colonoscopy requires less rigorous preparation, a shorter examination, and no anesthesia. Costs are debatable, some studies cite the virtual procedure as less expensive ($500 to $1000 compared with $1000 to $2000), ³ while others suggest that the additional time needed for analysis makes it more expensive. ⁴ Neither covered by many insurers nor generally available, virtual colonoscopy has other drawbacks. In a study of 300 patients examined with both methods, the virtual procedure missed 30% of the polyps (more smaller ones) found using the traditional video. ⁵ Approximately 10% of those screened virtually will need to undergo the conventional procedure to remove the polyps. ⁶ Nonetheless, until totally noninvasive molecular tests are available, virtual colonoscopy may offer reluctant patients an acceptable option.

Another test yet to find its way into standard care screens smokers and ex-smokers for lung cancer. ⁷ Lung cancer is the leading cause of cancer deaths in the United States, with the same 5-year survival rate it had over 30 years ago. The screening uses powerful spiral CT scanners to detect tumors that are much smaller than those found with conventional chest X-rays. According to one study, 23% of those tested using spiral CT had suspicious nodules, but only 2.7% were in fact cancerous ⁸ ; almost all of these were in Stage 1 and had not spread. ⁸ Chest x-rays found suspicious nodules in only 7%, 0.7% of them cancerous; half of these were Stage 1 and the other half were more advanced. Such findings raise more questions than answers. Dr. Barnett Kramer, ⁹ director of the Office of Disease Prevention at the National Institutes of Health, cites a study in Japan in which spiral CT found similar number of cancers in nonsmokers as in smokers, yet the latter are ten times more likely to die of lung cancer. In his opinion, the new test, like the old one, is "finding cancers that are not dangerous."

Early detection of cancer is an uncertain enterprise. Dr. Stephen Swensen, chairman of radiology at the Mayo Clinic, screened 1520 smokers and former smokers with spiral CT. Of more than 2800 suspicious nodules, 37 were found to be malignant tumors. In some cases, the testing entailed chest surgery, with a 4% chance of death. Although patients undergoing spiral CT "assume that this could save their lives," said Dr. Swensen, "That is absolutely, unequivocally unproven." ¹⁰

Dr. Steven Goodman, associate professor of pediatrics and epidemiology at the Johns Hopkins University School of Medicine, notes that screening is different from any other medical program. "We bring it to healthy people, and those who test positive become sick people, the subjects of medical intervention. We need to be awfully careful." ⁹

References

1. Farella C. Is popular heart scan sound or scam? Nursing Spectrum. Published October 9, 2001. Available at www.virtualphysical.com/nursingspectrum1000901.html. Accessed April 10, 2002.

2. Salganik MW. Full-body scanning seeks a new image. Baltimore Sun. April 13, 2001. Available at www.virtualphysical.com/BaltSun010413.html. Accessed April 10, 2002.

3. Lahn J. Colonoscopy could go virtual. Fortune . January 2000. Available at www.business2.com/articles/mag/print/p,1643,36643,FF.html. Accessed February 11, 2002.

4. Boyd K. Virtual colonoscopy effective for colon cancer screening. UCSF Medical Center Health & Medical News press release dated May 30, 2002. Available at www.ucsfhealth.org/news.org/hm_news/30may01.htm. Accessed June 2002.

5. Yee Y, Akerkar GA, Yung RK. Colorectal neoplasia: Performance characteristics of CT colonography for detection in 300 patients. Radiology. 2001;218:685-692.

6. Cropper CM. Colon cancer: An easier diagnosis. Business Week Online . October 15, 2001. Available at www.virtualphysical.com/bwonline101501.html. Accessed April 10, 2002.

7. Kolata G. Lung cancer test is much in demand, but benefit is murky. New York Times . June 21, 2000. Available via the New York Times Premium Archive at www.nytimes.com. Accessed March 14, 2002.

8. Henschke CI, McCauley DI, Yankelevi DF, et al. Early Lung Cancer Action Project: Overall design and findings from baseline screening. Lancet. 1999;354:99-105.

9. Kolata G. Test proves fruitless, fueling new debate on cancer screening. New York Times. April 9, 2002:D1,D4.

10. Kolata G. Questions grow over usefulness of some routine cancer tests. New York Times. December 30, 2001. Available via The New York Times Premium Archive at www.nytimes.com. Accessed March 14, 2002.

Dr. Silverman is a Professor of Radiology, the Gerald D. Dodd, Jr. Distinguished Chair of the Diagnostic Imaging Section of Body Imaging, Section of Body Imaging University of Texas M.D. Anderson Cancer Center Houston, TX. Dr. Davros is Section Head of Medical Physics, Department of Diagnostic Radiology. Cleveland Clinic Foundation, Cleveland OH. Dr. Rosen is Director of Radiology, Beth Israel Deaconess Medical Center­Medical Care Center North, Chelsea, MA. Ms. Douglas is an adjunct faculty member at the Johns Hopkins University School of Nursing, Baltimore, MD.

In compliance with the Essentials and Standards of the ACCME, the authors of this CME tutorial are required to disclose any significant financial or other relationships they may have with the manufacturer(s) of any commercial product(s) or provider(s) of any commercial service(s) discussed in this program.

Dr. Laghi, Dr. Davros, Dr. Rosen, and Ms. Douglas have disclosed that they have no such relationships. Dr. Silverman discloses a relationship with Amersham Health through their sponsorship of this newsletter series, for which he serves as editor-in-chief.

CT Update is published by Anderson Publishing, Ltd., 1301 West Park Ave., Ocean, NJ 07712; (732) 695-0600. O. Oliver Anderson, Publisher, Elizabeth A. McDonald, Managing Editor; Felice Ponger, Art Director.

Sponsored by a grant from Amersham Health. The views and opinions expressed in this publication are those of the authors and do not necessarily reflect those of
the publisher or sponsor. Full and complete prescribing information should be reviewed regarding any product mentioned prior to use.

2002 Anderson Publishing, Ltd.