| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
EDUCATION EXHIBIT |
Department of Radiology, University of California, San Francisco
CT colonography (also known as virtual colonoscopy) has undergone significant development over the past 10 years. Technologic advances with multidetector CT have allowed the use of thinner collimation. Image displays have improved in terms of functionality and image quality. The ability to navigate automatically through the colon with the 3D endoluminal view is standard on most current CT colonography software platforms, and some even allow automatic navigation on the 2D axial view. New displays are undergoing continuing improvements in an attempt to allow a larger viewing area at one time. One such example is known as the virtual dissection view, in which the colon is split along its long axis and flattened for evaluation. Computer-aided detection of lesions will be an option that will be added to computer workstations in the near future.
Radiologists’ experience in CT colonography continues to grow as the total number of procedures performed in this country increases. The results of single-center studies evaluating polyp detection with CT colonography have been variable. However, numerous large trials document that CT colonography has excellent sensitivity (>90%) for polyps 10 mm and larger (1–4). Multicenter studies evaluating use of CT colonography in asymptomatic patients will continue to help define the role of CT colonography in colorectal screening.
In the preceding article, Dr Pickhardt presents an excellent review of polypoid lesions seen with CT colonography (5). He is to be congratulated for accomplishing his stated goal of demonstrating the spectrum of lesions included in the differential diagnosis of colorectal polypoid lesions seen at screening CT colonography. In addition to excellent CT colonographic and colonoscopic images, the text provides the reader with a useful summary of lesion histologic characteristics and clinical significance. Neoplastic mucosal lesions including tubular adenomas, tubulovillous adenomas, and villous adenomas are reviewed, as well as the indistinguishable nonneoplastic lesions such as hyperplastic polyps and normal epithelium. Both submucosal lesions (such as lipomas and pneumatosis cystoides coli) and extrinsic lesions are shown to be particularly well demonstrated on 2D views processed with soft-tissue window settings and levels. As noted in other reviews, Dr Pickhardt indicates that the most common causes of false-positive lesions are residual stool and thickened folds.
Dr Pickhardt emphasizes his conclusion that primary 3D evaluation for polyps with findings confirmed with 2D images yields the best results for screening detection. The reader should note (and Dr Pickhardt points out) that this recommendation is based on his experience with a specific software platform (Viatronix V3D). This software platform for 3D endoluminal fly-through examination was used along with 2D images in a study of asymptomatic, largely average-risk patients conducted by Dr Pickhardt and his colleagues (6). This study, which is the basis of the review of CT colonography published in this issue of RadioGraphics, demonstrated a sensitivity of 93.8% for the detection of adenomatous polyps that were at least 10 mm in size.
We urge readers not to be discouraged by not having available to them the commercial software used in the Pickhardt et al trial (6). Dr Pickhardt states clearly in this review the importance he gives to this software. He further suggests a reimbursement benefit with its use, according to his experience with third-party payment, which has been "closely linked" to their results with the Viatronix V3D system. It is not clear whether it was the use of this vendor-specific 3D endoluminal fly-through for primary interpretation or a combination of other variables that led to the excellent results of this trial. It should be noted that this study (6) differs from others in several respects, including the following: a combination of barium/iodine fecal and fluid tagging and a rigorous colonic cleansing protocol consisting of double-dose phospho-soda, a study group of relatively young and healthy patients, and, of course, the use of 3D endoluminal fly-through for primary interpretation. The translucency rendering option, which was used to distinguish tagged stool from soft-tissue polyp, is available solely on the workstation used for this study. The Pickhardt et al study (6) is the only published trial to date that has employed this particular method, and it was carried out on a large asymptomatic, predominantly average-risk patient cohort. Furthermore, the study was not performed using a direct comparison of primary 2D versus primary 3D interpretation for determination of lesion detection accuracy. Consequently, the trial in itself does not directly prove that the use of primary 3D interpretation will result in improved lesion detection as compared with primary 2D interpretation.
Dr Pickhardt does acknowledge that many radiologists rely on 2D views for primary interpretation and use 3D views only for problem-solving. We would emphasize that others who have used such a protocol have reported excellent detection rates for large polyps (1,2,4). We anticipate that radiologists will use an interpretation strategy with which they are comfortable on the specific workstation available to them. We also predict that they can achieve accurate results with CT colonography irrespective of whether a primary 2D or primary 3D technique is used. The ability to distinguish true lesions from false-positive lesions on both 2D and 3D views is critical, and it is possible with numerous other software platforms. As with every procedure that we perform in radiology, practice, experience, and clinical correlation are key.
Financial Interest: Both authors have no financial relationships to disclose.
 |
References |
|---|
 Top References References |
|---|
Department of Radiology, University of Wisconsin Medical School, Madison, Wisconsin
I thank and commend Drs Yee and Wall for their thoughtful comments on my overview of evaluating polypoid lesions at CT colonography. With regard to CT colonography interpretation, I would like to further clarify my approach to polyp detection in relation to the 2D multiplanar reformatted and 3D endoluminal displays, since this is a critical performance issue. Use of the term primary 3D for polyp detection has perhaps led to the misconception that the 2D images are not used for detection. In fact, my approach (and that used in the New England Journal of Medicine screening trial) employs both the 2D and 3D displays for polyp detection, effectively combining a primary 2D and primary 3D interpretation. Because these displays are complementary in nature, the overall polyp detection rate is optimized. This approach is somewhat analogous to the barium upper gastrointestinal examination, whereby both single- and double-contrast techniques are combined to provide complementary evaluation that exceeds the results provided by either approach in isolation. Although essentially all available CT colonography software systems allow for an effective primary 2D evaluation, it is important to note that most systems do not yet allow for an effective, time-efficient primary 3D evaluation.
Although the primary 2D approach fared well in the "polyp-rich" patient populations that composed the earlier feasibility trials, it must be understood that these performance results cannot be simply extrapolated to a screening population with a low prevalence of polyps. To wit, the three large CT colonography trials (all >600 patients) that subsequently used a primary 2D approach in a screening population setting have all yielded disappointing results (1–3). In my opinion, we need to accept the fact that primary 2D evaluation alone is inadequate for CT colonography screening and move on to more pressing matters (such as expanding insurance coverage for CT colonography screening beyond the University of Wisconsin). For the best chance of success, future CT colonography screening trials should not only combine primary 2D and 3D interpretation, but also employ fluid and stool tagging, adequate colonic cleansing, and multidetector CT, as we have done. By definition, any true screening trial will consist of "relatively young and healthy patients," since this population is the intended target group for large-scale asymptomatic screening with CT colonography.
I believe that a caveat is needed with regard to the 3D "virtual dissection" view, which effectively fillets and flattens the endoluminal perspective. Although the concept is very attractive because it allows for even more rapid 3D evaluation, the software versions currently available create too much distortion for reliable polyp detection, in my opinion. Virtual dissection, however, along with computer-aided detection, remain important areas for further investigation.
As we await the results of future validation trials, I strongly believe that clinical implementation of primary CT colonography screening should proceed without delay by those who are currently able to demonstrate an acceptable accuracy. The need for immediate action is self evident, as far too many people continue to be diagnosed with an often lethal disease that could have been easily prevented by effective routine screening.
|
References |
|---|
|
TopReferencesReferences |
|---|
Related Article
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOGRAPHICS | RADIOLOGY | RSNA JOURNALS ONLINE |
