| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Clinical Observations |
1 All authors: Radiology Department, Boston Medical Center, 88 East Newton Street, 2nd Floor, Boston, MA 02215.
Received March 15, 2005;
accepted after revision July 10, 2005.
Address correspondence to S. W. Anderson
(Stephan.Anderson{at}bmc.org).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CONCLUSION. Unenhanced and contrast-enhanced MDCT images, interpreted in PACS workstations with axial images, are moderately sensitive and specific for showing choledocholithiasis.
Keywords: biliary system • CT imaging • gastrointestinal radiology
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Although CT is not the imaging technique of choice for patients with clinical suspicion of choledocholithiasis, it is commonly used in patients with jaundice, abnormal liver function test results, or other symptoms possibly related to the biliary tract. Previous studies have reported the performance of CT using different techniques for showing common bile duct stones. These include incremental [9] and helical CT [10-13] acquired both with [11] and without the addition of an IV contrast agent [12, 13]. Reported sensitivities have ranged from 20% [14] to as high as 88% [12].
In the emergency setting, CT scans are being acquired with increasing frequency in patients presenting with pain or other nonspecific abdominal complaints. Thus, CT often provides the first opportunity to detect bile duct stones in patients undergoing imaging for various medical conditions. Not surprisingly, CT scans with different protocols are acquired in many patients with common bile duct stones. With an almost universal use of MDCT in emergency radiology departments, acquisition of images with high spatial resolution is now routine. Thus, it is important to recognize the variable appearance of bile duct stones on CT images acquired with MDCT technology. In this study, we assessed the ability of MDCT performed with and without an IV contrast agent to detect common bile duct stones.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
|
Image Analysis
The CT scan data sets were transferred to PACS workstations for analysis
(Aurora, software 6.5; Merge eFilm). CT images were interpreted independently
by two radiologists, observer 1 and observer 2, who had 9 and 3 years of
experience, respectively, in cross-sectional abdominal imaging. Observers were
free to use the window settings they preferred, which included narrow settings
if a common duct stone was not initially identified on soft-tissue window
settings. The radiologists had the option to generate multiplanar reformations
of the axial images using a postprocessing software package incorporated into
the PACS workstations (Voxar 3D; Barco). The radiologists received no clinical
information or results of earlier or subsequent imaging studies. However, they
were aware of the aim of the study.
Each radiologist was shown the group 1 (unenhanced) CT images separately in a randomized order with all patient details removed for analysis. After this, the observers were shown the group 2 (IV contrast-enhanced) CT images independently with the patient details removed for analysis. A diagnosis of choledocholithiasis was made when there was direct visualization of a stone as a high-attenuation focus or filling defect with surrounding bile (target sign, crescent sign) seen within the bile duct. Common bile duct caliber was not measured, and bile duct features such as dilatation, sharp amputation, or both were not used as ancillary signs of choledocholithiasis. When stones were identified, an analysis of the CT attenuation characteristics was made to classify them as diffusely calcified, partially calcified, or of soft-tissue attenuation.
All ERCP images were also interpreted on the PACS workstation by a gastrointestinal radiologist with more than 8 years of experience in cross-sectional abdominal imaging. Studies were classified as positive or negative for the presence of stones in the biliary tract. Stones in the gallbladder or cystic duct were not assessed in this study. If stones were found, the number of stones (up to a maximum of 10) was recorded, and the size of the largest stone was measured using electronic calipers on the workstation. The size of a stone was determined by its size relative to that of the endoscope taken to measure 11 mm. The results from the ERCP were used as the gold standard in determining the accuracy of CT findings.
Finally, after the results of the ERCP were disclosed to the radiologists, they were asked to perform a consensus review of the CT data sets to determine a possible explanation for any false-positive or false-negative interpretations made.
The originally dictated reports by the attending radiologists interpreting the CT scans at the time of completion of the studies were also evaluated. It was noted, based on the dictated reports, whether common bile duct stones were directly visualized or not at that time. A total of eight attending radiologists with variable experience (2-25 years of experience) were responsible for the original dictations.
Statistical Analysis
The independent interpretations by the two radiologists were used to
determine the following diagnostic performance parameters for MDCT images:
sensitivity, specificity, positive predictive value (PPV), negative predictive
value (NPV), and diagnostic accuracy. We calculated the 95% confidence
intervals (CIs) for each of these parameters. This analysis was done
separately for studies performed with IV contrast material and for the studies
performed without IV contrast material. Finally, we evaluated the agreement
between the two observers using the kappa statistics. Agreement was classified
as follows: 0.00-0.20, poor; 0.21-0.40, fair; 0.41-0.60, moderate; 0.61-0.80,
good; 0.81-1.00, very good.
Diagnostic performance parameters were also calculated for the original interpretations of the CT scans at the time of their completion. Sensitivity, specificity, PPV and NPV, and diagnostic accuracy were calculated including 95% CIs. Again, this was done separately for those studies performed with IV contrast material and those performed without IV contrast material.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Unenhanced CT (Group 1)
Thirteen (33%) of 39 patients who had unenhanced CT had choledocholithiasis
at ERCP. Table 1 gives the
results for the diagnosis of choledocholithiasis per CT study for the two
observers. Between the two observers, a total of seven false-negative
interpretations were made in six patients. In one of these patients who had a
dual study, both observers missed a stone, a solitary 8-mm soft-tissue density
stone situated within the distal common bile duct. Even on retrospective
review, this stone was not apparent on the unenhanced CT, but both observers
detected it prospectively on the IV contrast-enhanced study (Figs.
4A and
4B). On consensus review, the
observers considered the stone impossible to detect because of its isodensity
with surrounding soft tissue and the lack of a target or crescent sign. The
remaining five studies were misinterpreted by only one of any of the two
observers. At CT, the stones were classified as soft-tissue attenuation in
two, as homogeneously calcified in two, and as having partial calcification
(calcified rim) in one. On retrospective review, the consensus of the two
observers was that all these stones were visible, but the major causes for not
detecting them were related to their small size, similar density to
surrounding tissue, confusion of dense lower bile duct stone for oral contrast
material in the duodenum, or all three
(Fig. 5).
|
|
|
|
Between the two observers, a total of four false-positive interpretations were made in three patients using unenhanced CT for detection of choledocholithiasis. Interpretations of both observers were falsely positive in one patient with a dual study. In this same patient, both observers also falsely identified a common bile duct stone on the corresponding IV contrast-enhanced study. In the remaining two patients, false-positive misinterpretations were made by only one of any of the two observers. On consensus review, all these misinterpretations were thought to result from volume averaging from surrounding soft tissue such as the pancreas or common bile duct wall.
Contrast-Enhanced CT (Group 2)
Fifteen (26.3%) of 57 patients in this group were found to have bile duct
stones at ERCP. Table 2 gives
the results for the diagnosis of stones per CT study for the two observers.
Between the two observers, a total of four false-negative interpretations were
made in three patients. Of these, both observers made a false-negative
interpretation in one patient. This was a patient with a 4.5-mm stone removed
at ERCP that was not apparent by CT even on retrospective review. The
remaining two false-negative interpretations occurred in two different
patients, and on retrospective review, both observers agreed the stones were
present: one as a minimally hyperattenuating stone and the other as a
soft-tissue attenuation focus. At ERCP, a single 4-mm common bile duct stone
was found in the first patient, and two stones measuring 3.5 mm and 6 mm were
found in the second patient, respectively. The reasons for these misses were
thought to be related to their small size and isodensity with surrounding
tissue.
|
The observers made a total of 12 false-positive interpretations in 10 patients using IV contrast-enhanced CT for detection of choledocholithiasis. Both observers made false-positive interpretations in two patients. Even at retrospective review, a calcified focus was thought to be clearly present within the common bile duct in both patients by both observers (Fig. 6). At ERCP performed between 1 and 6 days after the CT, no stones were found. On consensus review, the only explanation offered was that the stones likely passed while awaiting therapeutic ERCP. Of the remaining eight patients with false-positive CT studies, four were described as being of soft-tissue density; three were described as being partially calcified; and, in one patient, as being uniformly calcified. At consensus review, these false-positive findings were thought to have likely resulted from partial volume averaging from enhancing the wall of the common bile duct, interference from oral contrast material administered, and partial volume averaging from IV contrast-enhanced blood vessels coursing close to the bile duct (Fig. 7).
|
|
|
Table 4 gives the sensitivity, specificity, PPV, NPV, and diagnostic accuracy for the detection of choledocholithiasis using the original interpretations at the time of the CT scans. Overall, the sensitivity is much decreased for both IV contrast-enhanced and unenhanced studies compared with the results described earlier by the two observers during this study. Specificity, however, is somewhat increased, with 100% and 98% for the unenhanced and IV contrast-enhanced groups, respectively.
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The ability to detect bile duct stones at CT depends on a number of factors related to the stone (size, shape, position, density), bile duct (dilated vs nondilated), technology used (conventional vs helical CT), technique used (slice thickness, reconstruction interval, pitch, kVp, administration of contrast material), patient selection (screening population vs all comers), and interpreter variability (experience, anticipation of bile duct stones). The detection of nearly isoattenuating stones may be improved by narrow window settings, a technique that has been used in previous studies [12]. As noted, our radiologists used narrow settings if a common duct stone was not identified on standard soft-tissue window settings. It is well recognized that small stones situated within the intrahepatic ducts or impacted at the ampulla are difficult to identify, particularly in nondilated biliary ducts.
The attenuation of bile duct stones strongly influences the ability to detect them using CT. Heavily calcified stones are relatively easily identified, whereas soft-tissue density stones can be isoattenuating to surrounding tissue, making them difficult to identify. The attenuation of biliary stones varies with their composition. They can be made up of varying amounts of bile pigment, cholesterol, fatty acids, and calcium. Pure cholesterol stones are iso- or slightly hypoattenuating relative to bile, making them difficult, if not impossible, to detect. This imposes a theoretic upper limit for the CT detectability of choledocholithiasis of approximately 80% that cannot be improved on irrespective of any future advances in CT technology [17-19].
Even in light of the fact that Jeffrey et al. [16] achieved a sensitivity of 90% in diagnosing choledocholithiasis using conventional (incremental) CT, it is generally accepted that helical CT is superior in the diagnosis of choledocholithiasis. The potential for helical CT, especially using MDCT technology, to image the anatomy using thin slices in a single breath-hold and to reconstruct those slices retrospectively using variable overlap should reduce much of the image degradation previously experienced from motion artifacts and volume averaging. For our study, we used 4-MDCT and acquired images with 3.2-mm thickness through the entire abdomen and pelvis in a single breath-hold (averaging 12 seconds) with retrospective image reconstruction at 3-mm intervals. Our accuracy of 84-88% obtained using our technique may have suffered compared with that of other investigators who performed CT using thinner slices focused around the biliary tract only.
We also performed our CT examinations using various protocols for oral and IV contrast administration. Patients such as those with right flank pain and suspected renal stones had CT performed without any contrast material administration, patients such as those with suspected bowel disease had administration of both oral and IV contrast material, and patients being evaluated after major trauma had IV contrast material administration only. It is generally believed that administration of oral contrast material, IV contrast material, or both reduces the accuracy for diagnosis of choledocholithiasis. In our study, we encountered false-positive findings for stones resulting from reflux of oral contrast material into the lower bile duct and from partial volume averaging from IV contrast-enhanced bile duct mucosa. However, we also found at least one instance where administration of IV contrast material helped; this was in a patient with a soft-tissue density stone situated at the ampulla was only made obvious after IV contrast enhancement of the surrounding bile duct wall. In clinical practice, both oral and IV contrast material are frequently used because they make the biliary anatomy clearer and they help with diagnoses of mass lesions such as neoplasms. At least from our limited data, it appears that the administration of IV contrast material does not limit the accuracy of MDCT for diagnosis of choledocholithiasis in patients undergoing routine abdominal imaging.
In our study, we achieved a sensitivity of 69-87%, specificity of 83-92%, and accuracy of 84-88% in the CT diagnosis of choledocholithiasis. Twenty of the 21 patients with stones were identified by at least one of the two observers, and stones in only one patient were missed by both observers. The stones missed were mostly of soft-tissue or partially calcified density, measuring less than 6 mm, and situated in the lower common bile duct. The effect of bile duct size on sensitivity was not evaluated in this study. On retrospective review, the major reasons for missing stones were thought to be related to their small size, similar density to surrounding tissue, confusion of dense lower bile duct stones for oral contrast material in the duodenum, or all three. False-positive results occurred in our study related to partial volume averaging from surrounding tissues including opacified blood vessels coursing close to the bile duct, IV contrast enhancement of bile duct mucosa, and reflux of oral contrast material into the lower common bile duct. It is also highly likely that a few of our false-positive results were related to real bile duct stones that had passed by the time ERCP was performed. As CT technology continues to improve with the increasing use of 16-, 32-, and 64-MDCT, it is likely that the accuracy for detection of choledocholithiasis in patients undergoing routine abdominal CT will improve compared with our results. The increasing resolution that will be afforded and decreasing acquisition time limiting motion artifacts may eliminate some of the false-positive and negative interpretations described in our study. However, for the theoretic limitations just mentioned, it cannot be expected to rival the excellent results obtained with MR cholangiography or endoscopic sonography.
Because of its retrospective nature, our study has a number of limitations. First, the mean delay of 7 days for ERCP after CT probably led to passage of some stones, resulting in an increased false-positive rate at CT. Second, the fact that the observers in this study were primed for detection of choledocholithiasis likely led to an improved sensitivity for detection for bile duct stones compared with real-time interpretations at the time of the examination. This likely explains the large disparity between sensitivities achieved by the two observers in this study compared with that achieved at the original interpretations at the time of the CT scans. The fact that the radiologists involved in this study were primed for the detection of common bile duct stones likely also resulted in overdiagnosing choledocholithiasis, which explains the disparity between specificity achieved in this study and that achieved at the time of the original interpretation.
In conclusion, unenhanced and contrastenhanced CT performed with a 4-MDCT scanner is moderately accurate in the diagnosis of choledocholithiasis in patients undergoing imaging for various indications. The impact of further improvements in CT technology, such as acquisition of images with isotropic voxels, should be investigated.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  S. W. Anderson, E. Rho, and J. A. Soto Detection of Biliary Duct Narrowing and Choledocholithiasis: Accuracy of Portal Venous Phase Multidetector CT Radiology, May 1, 2008; 247(2): 418 - 427. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
