| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Gastrointestinal Imaging |
1 From the Departments of Radiology (J.A.S.) and Gastroenterology (O.A.), Universidad de Antioquia, Hospital Universitario San Vicente de Paúl, Calle 64 x Kra. 51D, Medellín, Colombia; Department of Radiology, Boston Medical Center, Mass (M.A.B.); and Department of Radiology, Children's Hospital Medical Center, Cincinnati, Ohio (S.M.). From the 1998 RSNA scientific assembly. Received February 1, 1999; revision requested April 2; final revision received August 27; accepted August 30. Address correspondance to J.A.S. (e-mail: JorgeASoto@aol.com).
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: MR cholangiography was performed in 57 patients who were suspected of having choledocholithiasis and referred for endoscopic retrograde cholangiography. Non–breath-hold three-dimensional fast spin-echo, breath-hold single-section half-Fourier rapid acquisition with relaxation enhancement (RARE), and breath-hold multisection half-Fourier RARE sequences were compared. Two radiologists independently interpreted the MR cholangiograms. Evaluated diagnostic performance parameters included sensitivity, specificity, receiver operating characteristic (ROC) curves, and interobserver agreement (
statistics). Endoscopic retrograde cholangiography was the standard of reference.
RESULTS: Eight patients were excluded because of incomplete MR examinations (n = 4) or failure in the cannulation of the bile duct at retrograde cholangiography (n = 4). In 49 patients, the three MR cholangiographic sequences were completed successfully. In 24 (49%) of these patients, retrograde cholangiography demonstrated stones. Sensitivity and specificity of MR cholangiography exceeded 90%, and the area under the ROC curve was greater than 0.95 for both radiologists and for the three sequences. Interobserver agreement for presence of bile duct stones was at least 0.80 (very good) for the three sequences.
CONCLUSION: The three MR cholangiographic sequences had similarly high sensitivities and specificities for the detection of choledocholithiasis.
Index terms: Bile ducts, calculi, 766.289 • Bile ducts, MR, 766.121411, 766.121415, 766.121416, 766.289 • Bile ducts, stenosis or obstruction, 766.289 • Endoscopic retrograde cholangiopancreatography (ERCP), 766.1222 • Magnetic resonance (MR), comparative studies, 766.121411, 766.121415, 766.121416 • Magnetic resonance (MR), half-Fourier imaging, 766.121411, 766.121415, 766.121416
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
By using these pulse sequences, high accuracy rates in the detection of choledocholithiasis have been reported consistently (1–21). Irie et al (24) compared four MR cholangiopancreatographic sequences and concluded that the half-Fourier RARE sequence was superior for evaluation of the pancreatic duct. However, to our knowledge, there is no consensus about which sequence is preferable for the detection of bile duct stones. The purpose of this investigation was to compare the diagnostic performance of 3D fast SE, single-section half-Fourier RARE, and multisection half-Fourier RARE sequences in a group of patients suspected of having choledocholithiasis.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
In the remaining 49 patients, the three MR cholangiographic sequences were completed, and retrograde cholangiography was successful. Fourteen male and 35 female patients (mean age, 52 years; age range, 17–89 years) composed our study population. The study was approved by the investigation review board of Hospital Universitario San Vicente de Paúl, Medellín, Colombia, and informed consent was obtained from all patients.
Imaging
MR cholangiography preceded endoscopic retrograde cholangiography in all patients, and both procedures were completed within a 72-hour period. All MR cholangiographic examinations were performed after a fasting period of at least 6 hours. MR cholangiographic examinations were performed with a clinical 1.5-T MR imaging system (Philips ACS NT; Philips Medical Systems, Best, the Netherlands) with a standard body coil, since a torso phased-array coil was not available when the study was conducted. At our institution, we do not orally administer negative contrast agents or antiperistaltic medications at MR cholangiography.
We started the MR cholangiographic examinations with a respiratory-triggered, fat-saturated, two-dimensional, T2-weighted, fast SE sequence (3,000/100 [repetition time msec/echo time msec]; section thickness, 6 mm; gap, 0.6 mm; number of sections, 24; echo train length, 21; echo spacing, 12.5 msec; matrix, 230 x 256; number of signals acquired, six; and imaging time, 4 minutes 50 seconds). This sequence was performed in the transverse plane and was used to localize the imaging volume for subsequent MR cholangiographic sequences.
MR cholangiograms were obtained by using the following three pulse sequences: multislab 3D fast SE with respiratory triggering, single-section half-Fourier RARE, and multisection half-Fourier RARE. Imaging parameters used for the three sequences are listed in Table 1. A chemical-selective fat-saturation prepulse was applied with all sequences.
|
For the half-Fourier RARE sequences, a 65% partial k-space filling was applied. The inherent symmetry of k-space along the phase axis allowed for filling of the remainder of the phase-encoding steps (20,21). The two half-Fourier RARE sequences were implemented during single breath holds. Total imaging times were 10 seconds for the single-section half-Fourier RARE sequence (2.5 seconds per projection, four projections) and 20 seconds for the multisection half-Fourier RARE sequence (2 seconds per section, 10 sections).
The multislab 3D fast SE and multisection half-Fourier RARE sequences were performed in a right anterior oblique-coronal plane. With the single-section half-Fourier RARE sequence, four oblique-coronal images were obtained. The planes were defined by using the transverse T2-weighted fast SE image that best demonstrated the distal common bile duct and the head of the pancreas. The first plane was defined as the one (usually the right anterior coronal plane) that included the distal common bile duct, the head of the gland, and as much of the body and tail as possible. The other three planes were rotated in counterclockwise fashion from the initial plane at 30° intervals around an imaginary line running through the long axis of the common bile duct; therefore, the last projection was oriented at 90° with respect to the first projection. With the single-section half-Fourier RARE sequence, a coronal saturation slab was placed over the spine and kidneys to prevent inclusion of cerebrospinal fluid and urine-filled renal collecting systems in the projection images.
Postprocessing and Image Analysis
MR images were interpreted independently by two radiologists (J.A.S., M.A.B.) who completed fellowships in body imaging and who had extensive experience in biliary tract MR imaging. The radiologists were unaware of the clinical and laboratory data and the results of other imaging tests. Postprocessing and interpretation of MR cholangiograms was performed at an independent workstation (EasyVision, Philips Medical Systems). Postprocessing of the source images obtained with the two multisection sequences was performed by using the maximum intensity projection and multiplanar reformation algorithms. The radiologists were given the option to obtain as many two-dimensional and 3D renderings as were required to provide an interpretation. Since the single-section half-Fourier RARE sequence produced a thick-slab projection image of the ductal structures, no postprocessing of the images was required or was possible.
With each sequence, images for all patients were interpreted as a group by each radiologist, with periods of at least 4 weeks between the interpretation sessions for each group. Images were presented in the following order: single-section half-Fourier RARE, 3D fast SE, and multisection half-Fourier RARE. To decrease bias introduced by learning associated with the evaluation of each sequence, images were presented to the observers in random order, with no dates or patient names.
The radiologists were asked to evaluate the images for overall quality, especially for artifacts resulting from patient motion. If image degradation was severe enough to compromise assessment of the ducts, the image was considered nondiagnostic, and no further analysis was performed. Images considered adequate for diagnosis were evaluated for common bile duct dilatation (upper limits of normal were set at 6 mm for patients with a gallbladder and at 9 mm for patients with a history of cholecystectomy) and for the presence and location of stones (intrahepatic, common bile duct, or both). Stones were seen as intraductal foci of absent or reduced signal intensity that were partially or totally surrounded by bile. The radiologists were then asked to classify their level of confidence regarding the presence or absence of stones by using the following five-point confidence rating scale: 1, definitely absent; 2, probably absent; 3, indeterminate; 4, probably present; and 5, definitely present.
Endoscopic Retrograde Cholangiography
One gastroenterologist (O.A.) performed all endoscopic retrograde cholangiographic procedures with a standard technique. The gastroenterologist was unaware of the MR cholangiographic results and interpreted the retrograde cholangiograms immediately after performing the procedure. This interpretation was used in the clinical care of the patients.
For the purpose of this study, the standard of reference was a retrospective consensus interpretation of the retrograde cholangiograms by the gastroenterologist who performed the procedures and a radiologist (J.A.S.) who was also involved in the evaluation of MR cholangiograms. This retrospective interpretation was performed 6 weeks after the evaluation of the last group of MR images. This delay decreased the potential for bias introduced by recalling MR imaging findings when the retrograde cholangiograms were interpreted. In no case was the initial interpretation by the gastroenterologist changed at the retrospective reading.
Stone size was measured by the gastroenterologist who used the known outer caliber of the endoscope as a correction factor for magnification. Stone size was classified as larger than 5 mm or 5 mm or smaller.
Statistical Analysis
Endoscopic retrograde cholangiography was the standard of reference for the determination of sensitivity and specificity of MR cholangiography. The initial assessment of the presence or absence of stones was compared with the final diagnosis to determine the sensitivity and specificity. We derived 95% CIs (25). Using statistics, we also used the results of the independent readings to measure the degree of interobserver agreement (26,27). Receiver operating characteristic (ROC) curve analysis of the five-point confidence scale was performed by using the ROC Analyzer Program for Windows (Centor RM, University of Alabama, Birmingham). The area under the curve was calculated by using the nonparametric trapezoidal rule (28). The areas under the ROC curves of the MR cholangiographic sequences were compared to determine differences in diagnostic performance (29). In addition, the areas under the curve for the two radiologists were compared to determine interobserver variability on the basis of the five-point confidence rating scale.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Statistics
Interobserver agreement ( statistics) for the presence of stones in the bile duct was 0.92 for the 3D fast SE sequence, 0.84 for the multisection half-Fourier RARE sequence, and 0.80 for the single-section half-Fourier RARE sequence. Interobserver agreement for the three MR cholangiographic sequences was considered to be very good (28).
ROC curves for both radiologists are shown in Figure 5. By using the nonparametric method, the area under the curve for radiologist 1 was 0.99 with the 3D fast SE sequence, 0.99 with the single-section half-Fourier RARE sequence, and 0.97 with the multisection half-Fourier RARE sequence. For radiologist 2, the area under the curve was 0.99 with the 3D fast SE sequence, 0.96 with the single-section half-Fourier RARE sequence, and 0.96 with the multisection half-Fourier RARE sequence. Comparison of the areas under the ROC curves for the two radiologists and for the three sequences demonstrated no statistically significant difference (P > .05). The diagnostic performance of the three sequences for both radiologists was excellent.
|
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
More recently, the use of breath-hold techniques has become possible with the introduction of single-shot RARE (16–18) and half-Fourier RARE (19–21) techniques. With single-shot RARE sequences, imaging time can be reduced to 4–5 seconds (15–18). The addition of half-Fourier acquisition allows even faster imaging. Because of the inherent symmetry of k space along the phase axis, as few as one-half of the full number of phase-encoding steps are acquired (31). In practice, usually slightly more than half of the k space is actually acquired to compensate for phase errors in the data. With the half-Fourier techniques, snapshot MR images of the pancreaticobiliary tree are generated with an imaging time of 2 seconds or less (19–21,32). The main drawbacks of single-shot techniques are a decreased signal-to-noise ratio and an increased risk of blurring in the phase-encoding direction (21,24,33). Half-Fourier techniques can be implemented with single thick sections (30–70 mm) or with multiple sequential sections (typically 5 mm thick).
Fast SE and multisection half-Fourier RARE sequences produce multiple, tomographic, thin sections that display intraductal filling defects as areas of signal void surrounded by bright bile. An area of complete signal void is produced when the diameter of the stone in the section-selection direction equals or exceeds the section thickness. For the detection of intraductal stones, this complete void leads to high reader confidence, although interpretation time may be longer, since all of the source images must be reviewed carefully.
A limitation of the multisection half-Fourier RARE sequence is its relatively low signal-to-noise ratio, which demands use of sections (usually 5-mm sections) that are thicker than those typically used with two-dimensional and 3D fast SE sequences (usually 2–3-mm sections); the thicker sections may compromise detection of small stones because of partial-volume averaging. The single-section half-Fourier RARE method displays stones as regions of lower signal intensity, not necessarily as areas of signal void, since this projection image represents an average of the signal intensities of the structures included within the entire thick slab. Therefore, the single thick-section method may be limited in its ability to depict stones in markedly dilated bile ducts, since the high signal intensity of bile may obliterate the low signal intensity of the stones (23,33).
Several groups of investigators have compared pulse sequences for MR cholangiography. Reinhold et al (30) compared a two-dimensional fast SE sequence with a steady-state free precession sequence and found that the fast SE sequence enabled better visualization of the ducts. Ichikawa et al (34) compared MR cholangiopancreatographic images obtained with half-Fourier RARE, two-dimensional fast SE, and steady-state free precession sequences and reported that the best image quality was obtained with the half-Fourier RARE sequence.
Irie et al (24) compared steady-state free precession, two-dimensional fast SE, 3D fast SE, and multisection half-Fourier RARE sequences in a phantom, in healthy volunteers, and in a group of patients suspected of having pancreatic abnormalities. They found that the half-Fourier RARE sequence had the highest contrast-to-noise ratio and spatial resolution and that this technique provided the best image quality. They also compared various section thicknesses (2–7 mm) and concluded that 5-mm-thick sections with no intersection gap were most appropriate for imaging the pancreatic duct. However, in the clinical patients in this same study, the detection rate of filling defects within the pancreatic duct with the half-Fourier method was poor (mean sensitivity, 21%).
Yamashita et al (33) conducted a comparison study in a phantom and in 108 patients. They compared two half-Fourier RARE sequences—a multisection technique with 5-mm section thickness and a single-shot, single thick section (projection) technique. They found that the projection technique with a section thickness of 30 or 50 mm best demonstrated the pancreaticobiliary tree and periampullary region but that the source images acquired with the multisection sequence best depicted stones in the bile duct. However, their study included only 10 patients with stones in the common bile duct. Therefore, there is no consensus about which sequence is most appropriate for imaging the biliary tree and for the demonstration of stones in the bile duct.
In our study, we compared the diagnostic performance of three MR cholangiographic sequences—non–breath-hold 3D fast SE, breath-hold single-section half-Fourier RARE, and breath-hold multisection half-Fourier RARE—for the detection of bile duct stones. With the single-section half-Fourier RARE sequence, we chose a section thickness of 30 mm, which Yamashita et al (33) found to be optimal. For the multisection half-Fourier RARE sequence, we used a section thickness of 5 mm with no intersection gap, as Irie et al (24) suggested. The 3D fast SE sequence we used has been described and tested previously (13,14).
Our results showed that the three sequences had similarly high sensitivity and specificity for the detection of bile duct stones. The detection rate of choledocholithiasis varied slightly but not significantly between the three sequences. In addition, the very good interobserver agreement and diagnostic performance of the three sequences, as shown by statistics and ROC curve analysis, respectively, indicated that interpretation of MR cholangiograms is objective and reliable.
The false-negative and false-positive interpretations that occurred with the 3D fast SE sequence were likely caused by degradation of image quality due to artifacts that resulted from irregular breathing patterns. It is well known that non–breath-hold techniques such as 3D fast SE imaging are sensitive to degradation caused by respiratory motion artifacts, even when respiratory triggering is used (22,23). The false-negative interpretations of single-section projection half-Fourier images probably occurred because subtle areas of decreased signal intensity in markedly dilated bile ducts were overlooked.
With the multisection half-Fourier method, partial-volume averaging resulting from the use of a 5-mm section thickness was likely the cause of the false-negative interpretations made by both radiologists in a patient with a small stone. Flow-related artifacts can cause intraductal foci of signal dropout on images obtained by using single-shot RARE techniques (21). These artifacts can mimic stones and may have been the cause of the false-positive interpretations of breath-hold half-Fourier images in our study. Other potential causes of false-positive interpretations include the presence of pneumobilia and ampullary tumors. However, neither of these conditions was demonstrated with endoscopic retrograde cholangiography in our study population.
Limitations of our study include the relatively small population of patients with stones (n = 24). Although the 95% CIs for the sensitivity and specificity were high and had narrow ranges, a larger series would be useful in the reproduction of the results obtained in our study. The high diagnostic performance of the three MR cholangiographic sequences compared in this study applied to two radiologists who were experienced in the interpretation of MR cholangiograms. A similar diagnostic performance may not be obtained by less-experienced radiologists.
The exclusion of patients in whom all three sequences were not completed may have biased the data in favor of the 3D fast SE sequence, since this sequence could not be performed in three patients with claustrophobia. Another potential for bias in our study results from the presentation of all images from each sequence as a group to the radiologists for interpretation. Since there is a learning curve associated with interpretation of MR cholangiograms, this method of presentation could have favored the performance of the sequence presented last (multisection half-Fourier RARE). However, as mentioned previously, the radiologists who provided the interpretations already had extensive experience with MR cholangiography; in our opinion, this prior experience decreased the effect of this potential source of bias. Finally, although bias could have been introduced by the fact that retrograde cholangiograms were interpreted by one radiologist (among others) who had previously interpreted the MR images, in no case was the initial interpretation made by the gastroenterologist at retrograde cholangiography changed during the retrospective reading. Therefore, it is unlikely that recall of findings on MR images influenced the final results of our study.
In summary, the results of this study show that the three MR cholangiographic sequences we compared had similarly high sensitivities and specificities for the detection of bile duct stones, although the multislab 3D fast SE method had a slightly higher (though not significantly different) confidence rating in the interpretations by the two radiologists. In clinical practice, any one of these MR cholangiographic sequences is likely to result in adequate image quality, but attention to technical details is important to reduce the number of misinterpreted studies.
Breath-hold sequences have an obvious advantage in most clinical situations because of the short imaging time necessary. Because of the short times, it seems prudent to combine both single-section and multisection breath-hold sequences to capitalize on the relative advantages of each. The 3D fast SE sequence is advised in patients who are unable to hold their breath or who are unable to understand or follow instructions to do so, such as pediatric or elderly patients. Breath-hold sequences are definitely preferable in patients with claustrophobia or in patients in whom an extended examination is not possible. When findings obtained with a breath-hold sequence are not conclusive, the addition of a non–breath-hold sequence with thinner sections and a higher signal-to-noise ratio (such as a 3D fast SE technique) may be beneficial to increase the diagnostic confidence regarding the absence or presence of small calculi.
|
Footnotes |
|---|
Author contributions: Guarantor of integrity of entire study, J.A.S.; study concepts, J.A.S., M.A.B., S.M.; study design, J.A.S., S.M.; definition of intellectual content, J.A.S., M.A.B.; literature research, J.A.S.; clinical studies, J.A.S., M.A.B., O.A.; data acquisition, J.A.S., M.A.B., O.A.; data analysis, J.A.S., M.A.B., S.M.; statistical analysis, S.M.; manuscript preparation, J.A.S.; manuscript editing, M.A.B., O.A.; manuscript review, S.M.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  S Rickes and C Uhle Advances in the diagnosis of acute pancreatitis Postgrad. Med. J., April 1, 2009; 85(1002): 208 - 212. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. R. Nandalur, H. K. Hussain, W. J. Weadock, E. J. Wamsteker, T. D. Johnson, A. S. Khan, A. R. D'Amico, M. K. Ford, S. R. Nandalur, and T. L. Chenevert Possible Biliary Disease: Diagnostic Performance of High-Spatial-Resolution Isotropic 3D T2-weighted MRCP Radiology, December 1, 2008; 249(3): 883 - 890. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. S. Lim, M.-J. Kim, S. Myoung, M.-S. Park, J.-Y. Choi, J.-S. Choi, and S. I. Kim MR Cholangiography for Evaluation of Hilar Branching Anatomy in Transplantation of the Right Hepatic Lobe from a Living Donor Am. J. Roentgenol., August 1, 2008; 191(2): 537 - 545. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E J Williams, J Green, I Beckingham, R Parks, D Martin, and M Lombard Guidelines on the management of common bile duct stones (CBDS) Gut, July 1, 2008; 57(7): 1004 - 1021. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Zhang, G. M. Israel, E. M. Hecht, G. A. Krinsky, J. S. Babb, and V. S. Lee Isotropic 3D T2-Weighted MR Cholangiopancreatography with Parallel Imaging: Feasibility Study. Am. J. Roentgenol., December 1, 2006; 187(6): 1564 - 1570. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. M. Merkle, P. A. Haugan, J. Thomas, T. A. Jaffe, and C. Gullotto 3.0- Versus 1.5-T MR Cholangiography: A Pilot Study Am. J. Roentgenol., February 1, 2006; 186(2): 516 - 521. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Romagnuolo, M. Bardou, E. Rahme, L. Joseph, C. Reinhold, and A. N. Barkun Magnetic Resonance Cholangiopancreatography: A Meta-Analysis of Test Performance in Suspected Biliary Disease Ann Intern Med, October 7, 2003; 139(7): 547 - 557. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Sugita, E. Sugimura, M. Itoh, T. Ohisa, S. Takahashi, and N. Fujita Pseudolesion of the Bile Duct Caused by Flow Effect: A Diagnostic Pitfall of MR Cholangiopancreatography Am. J. Roentgenol., February 1, 2003; 180(2): 467 - 471. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. K. Kim, B. S. Kim, J. H. Kim, H. K. Ha, P. N. Kim, A. Y. Kim, and M.-G. Lee Diagnosis of Intrahepatic Stones: Superiority of MR Cholangiopancreatography over Endoscopic Retrograde Cholangiopancreatography Am. J. Roentgenol., August 1, 2002; 179(2): 429 - 434. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. H. Kim, M.-J. Kim, S. I. Park, J.-J. Chung, S. Y. Song, K. S. Kim, H. S. Yoo, J. T. Lee, and K. W. Kim MR Cholangiography in Symptomatic Gallstones: Diagnostic Accuracy according to Clinical Risk Group Radiology, August 1, 2002; 224(2): 410 - 416. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
/300), and (c) multisection half-Fourier RARE (
