British Journal of Radiology (2005) 78, 1078-1085
© 2005 British Institute of Radiology
doi: 10.1259/bjr/14176682
Volume rendering of three-dimensional drip infusion CT cholangiography in patients with suspected obstructive biliary disease: a retrospective study
A Persson, MD1,
N Dahlström, MD2,
Ö Smedby, MD, PhD1 and
T B Brismar, MD, PhD3
1 Center for Medical Image Science and Visualization (CMIV), Linköping University Hospital, Sweden, 2 Department of Radiology, Hudiksvall Hospital, Sweden and 3 Department of Radiology, Karolinska University Hospital, Huddinge, Stockholm, Sweden
 |
Abstract
|
|---|
The purpose of this study was to evaluate the diagnostic potential of prolonged drip infusion CT cholangiography (DIC-CT) using meglumine iotroxate (Biliscopin®) and 3D volume rendering in patients with suspected obstructive biliary disease. From a material of 142 patients who had undergone a drip infusion CT, all cases with a verified surgical or endoscopic retrograde cholangiography (ERC) diagnosis (n=33) were selected. Age-matched controls were selected from the remaining examinations. Three radiologists reviewed all 66 examinations in retrospect, independently as well as in consensus. The image quality and the estimated diagnostic quality were rated as good or moderate in 91% of the 198 reviews. The consensus sensitivity and specificity for diagnosing biliary stones was 88% and 94%, respectively (with sensitivities ranging from 88% to 94% for individual observers, and specificities from 86% to 96%). Two out of three strictures were observed. No false positive strictures were described. The use of volume rendering technique (VRT) improved diagnostic certainty in 28/198 (14%) of the evaluations. The visualization of ductal stones was improved in 18/48 (38%). No differences in diagnostic quality between single and multislice CT were observed. We conclude that a detailed image of the biliary tree with good sensitivity and specificity can be obtained by means of bilirubin-governed infusion time DIC-CT with volume rendering reconstruction.
|
Introduction
|
|---|
Suspected biliary tract disease is a common clinical problem. A full radiological evaluation of the biliary tract should not only provide information regarding the presence of an obstruction, but also define the level and the cause of the obstruction. Advances in ultrasound, CT and MR technology over the past decade have greatly improved our ability to make this evaluation. In spite of the indisputable success of ultrasound and MRI, an alternative non-invasive diagnostic method is still required in those cases with inconclusive ultrasound findings or contraindications to MRI. In patients with unclear abdominal pain, CT is often used. However, CT without contrast administration often fails to resolve the non-dilated bile duct from the surrounding tissue. Also, stones in the common bile duct (CBD) are isodense with bile in up to 80% of cases [1]. When an intravenous biliary contrast agent is used, however, gallstones are more easily detected.
The aim of this study was to retrospectively evaluate the diagnostic value of CT cholangiography with volume rendering following the administration of biliary contrast material, meglumine iotroxate (Biliscopin®; Schering AG, Berlin, Germany), with a new scheme of prolonged drip infusion CT cholangiography (DIC-CT) in patients with suspected obstructive biliary disease.
|
Materials and methods
|
|---|
From a consecutive retrospective material of 153 DIC-CT examinations (142 patients, 68 men and 74 women, mean age 69 years, range 24–95 years) performed at Hudiksvall Hospital, Sweden, during the period from January 1996 to January 2003 all examinations with a final diagnosis (n=33) made through surgery (n=15), endoscopic retrograde cholangiography (ERC) (n=17) and percutaneous transhepatic cholangiography (PTC) (n=1) were selected. Age-matched controls were selected from the remaining 120 examinations. The controls were also matched regarding protocol parameters (i.e. CT scanner, slice thickness, pitch, etc.). The medical records of all controls, i.e. the files from the departments of radiology, surgery and internal medicine were reviewed for repeated hospital care to confirm true negative DIC-CT (Table 1
). The study was approved by the local ethics committee.
Contrast administration
The serum bilirubin level was measured before CT examination by means of standard laboratory methods. 100 ml meglumine iotroxate (Biliscopin®), 50 mg I ml–1, was administered by intravenous drip infusion through a plastic cannula, 18 gauge, in the right arm vein. To allow longer infusion times, the solution volume was increased by dilution with 500 ml isotonic sodium chloride. The infusion time was determined by the pre-examination bilirubin level according to a pre-defined scheme (Table 2) [unpublished results]. Following the guidelines from the manufacturer, the drip infusion was started at a low infusion rate (0.5 ml min–1) and increased to the desired infusion rate during the following 3–5 min. The patient was monitored during the infusion on the ward and during the transportation to the radiology department.
View this table:
[in this window]
[in a new window]
|
Table 2. Infusion time in relation to serum bilirubin level. The same total amount of iodine (5 g) was given to all patients
|
|
The CT scanning started immediately after the contrast administration. For distension of the distal duodenum, the patients ingested two glasses of drinking water immediately before the CT examination.
Scanning parameters
To optimize the location of the allocated water and to improve the visualization of the distal bile duct, the patient was examined in a position rotated 45° to the right from a supine position by means of thin-section single-breath-hold helical CT. Specific scan protocols varied depending on the CT scanner available at the time of examination. Between December 1995 and November 1999, 33 patients were scanned by means of a single slice CT scanner (Somatom Plus 4 A; Siemens Medical Systems, Forchheim, Germany) with 1 mm x 5.0 mm collimation. From December 1999 to November 2002, a multidetector four-slice CT scanner (Somatom Volume Zoom; Siemens Medical Systems, Forchheim, Germany) with 4 mm x 2.5 mm collimation was used in 33 patients. On both scanners a pitch of 1.5 with an increment of 1.0 mm, 200 mAs and 120 kV was used.
Image display
The resulting 66 exams were retrospectively re-evaluated in a random order regarding the presence of bile duct stones, tumour or duct stricture by three radiologists who were blinded to patient data. The readers (AP, ND and TB) were from three different hospitals. ND and TB had not earlier encountered any of the examinations and AP had only reviewed single examinations (8 of the cases, all reviewed more than 3 years before the study). The readers had different levels of experience of VRT software, from moderate to expert. The same software version and computer screen were used for all measurements. Time to complete each study ranged from approximately 20–30 min depending on the complexity of the case and the experience of the reader.
Ductal stones were defined as an intraluminal contrast defect. Stricture was defined as a >50% diameter reduction of the bile duct. Tumour was defined as a soft tissue mass external to the bile duct with or without stricture. For image display and image post-processing, a 3D workstation (Leonardo with 3D software and Inspace interactive 3D Viewer; Siemens Medical Solutions Computed Tomography, Forchheim, Germany) equipped with a real-time volume rendering graphic accelerator card was used. The retrospective evaluation of all examinations was first carried out individually. Disagreements were then solved in a consensus session by all three reviewers. The 2D axial slices and the 3D VRT display were assessed separately. The reviewer was allowed to alter the windowing on the 2D images and the transfer function in the VRT application during the review session.
The number of visible stones in the CBD was noted, but gall bladder stones, although frequently identified, were excluded from the analysis. Image quality in the 3D VRT display was evaluated by means of a four-point scale where 0 denoted a non-visualized bile duct, 1 poor, 2 moderate, and 3 good visualization of the CBD. It was recorded whether volume rendering display (VRT) transfer function alterations were necessary to gain diagnostic certainty. It was also noted whether 3D VRT provided any additional information of relevance to the diagnosis. Finally, the overall diagnostic quality was determined on a four-point scale, where 0=inconclusive, 1=poor, 2=moderate and 3=good quality. The attenuation of the main bile duct and liver was obtained in HU by each viewer by retrospective measurement on the restored digital images in all individuals with bilirubin >19 µmol l–1 (the upper limit of the clinically used reference interval of the test), i.e. in 90 patients out of a consecutive 142 patients [unpublished results]. The attenuation value in the main bile duct was measured perpendicular to the lumen by a region of interest (ROI) covering about 50% of the lumen area. Single measurements at a representative level of the main bile duct were obtained. The liver measurements were conducted in the posterior segment of the liver by a single 1–2 cm2 ROI that only covered parenchyma, avoiding bile ducts and vessels.
Statistical methods
Sensitivity and specificity were calculated with 95% confidence intervals (computed exactly from the binomial distribution). The correlation between measured and rated data was computed as Pearson's correlation coefficient. Image and diagnostic quality of examinations (sums of the scores of all three observers) carried out with single and multidetector equipment were compared using the Mann-Whitney U-test.
|
Results
|
|---|
Out of the patients with a proven final diagnosis (n=33), ERC and surgery were normal in 14 cases. A total number of 27 biliary stones were observed in 16 patients during ERC (n=14) or during surgery (n=2). In the retrospective consensus evaluation, CBD stones were observed in 14 of these patients, resulting in a sensitivity of 88% (Table 3). Both of the false negative examinations showed a poor excretion of contrast to the bile ducts. In the first of these examinations, one of the reviewers proposed a bile duct stone, but the consensus was that the examination was inconclusive (Figure 1). In the second false negative examination, two of the reviewers proposed a bile duct stone, but the consensus was that the examination was negative. The sensitivity for individual observers ranged from 88% to 94%. False positive CBD stones (Figures 2 and 3) were described in three cases in the consensus session, corresponding to a specificity of 94%. The specificity for individual observers ranged from 86% to 96%. Two out of three surgically verified strictures were observed during DIC-CT. The false negative examination was graded as poor quality by two reviewers and as inconclusive by one (Figure 4). No false positive strictures were described during DIC-CT. The image quality in the 198 assessments (3 observers x 66 examinations) was rated as good or moderate in 91% of the assessments (87% for single slice and 96% for multislice CT) (Figure 5), and the estimated diagnostic quality was rated as good or moderate in 91% of all assessments and 92% in consensus (Figure 6). There was no difference between single and multislice CT regarding image quality and the estimated diagnostic quality (p>0.1). The observed attenuation in the main bile duct and liver at different bilirubin values is shown in Figure 7. The use of VRT improved diagnostic certainty in 28/198 (14%) of the evaluations and the visualization of ductal stones was improved in 18/48 (38%) of the positive cases (Figure 8). The VRT parameter settings were changed interactively in 128 of the 198 evaluations (Figure 9). There was a weak but significant correlation (r=0.26, p<0.001) between the rated diagnostic quality and the measured attenuation in the CBD.
View this table:
[in this window]
[in a new window]
|
Table 3. The accuracy of drip infusion CT cholangiography (DIC-CT) to diagnose biliary stones. Viewer 1=expert level in volume rendering technique (VRT), viewer 2=advanced level and viewer 3=moderate level
|
|
View larger version (159K):
[in this window]
[in a new window]
|
Figure 1. No contrast excretion is observed in this case. At the consensus reading, the examination was regarded as not conclusive. One reader described a calcified bile duct stone (arrow), which was verified with endoscopic retrograde cholangiography.
|
|
View larger version (53K):
[in this window]
[in a new window]
|
Figure 2. (a, b) Axial 2D images, (c) volume rendering technique (VRT). Three small bile duct stones (arrows) were observed in common bile duct (consensus), no stone was observed at endoscopic retrograde cholangiography.
|
|
View larger version (94K):
[in this window]
[in a new window]
|
Figure 3. Axial 2D images. (a) One small stone (arrow) was observed in the common bile duct (consensus), (b) no stone was observed at endoscopic retrograde cholangiography.
|
|
View larger version (87K):
[in this window]
[in a new window]
|
Figure 4. This case shows poor excretion of contrast. There is an area with low attenuation (arrow) close to the confluence of the main bile duct, which was proposed by consensus to represent a stricture, but the examination was ranked as inconclusive, requiring further investigation. Final diagnosis at endoscopic retrograde cholangiography confirmed the existence of a total post-operative occlusion of the common bile duct.
|
|
View larger version (17K):
[in this window]
[in a new window]
|
Figure 5. Image quality in the 3D volume rendering technique (VRT) display for the 198 assessments (3 observersx66 examinations). The number at the top of each bar denotes the absolute number of assessments, and the bar height represents the percentage.
|
|
View larger version (13K):
[in this window]
[in a new window]
|
Figure 6. Estimated diagnostic quality of the 3D drip infusion CT cholangiography (DIC-CT) data sets. The number at the top of each bar denotes the absolute number of assessments, bar height represents the percentage.
|
|
View larger version (28K):
[in this window]
[in a new window]
|
Figure 7. Attenuation in the common bile duct and liver parenchyma at drip infusion CT cholangiography (DIC-CT) as a function of serum bilirubin in all individuals with >19 µmol l–1, i.e. in 90 patients out of a consecutive 142 patients.
|
|
View larger version (84K):
[in this window]
[in a new window]
|
Figure 8. Volume rendering 3D images with two different transfer functions. (a) A linear ramp transfer function giving high opacity (up to 50%) to all voxels inside the bile duct. (b) Rendered with lower opacity (12%) and a transfer function emphasising attenuation values in a narrow intermediate interval, corresponding to the border between the bile duct lumen and surrounding tissue. A surgically verified stone is observed in the distal bile duct.
|
|
View larger version (42K):
[in this window]
[in a new window]
|
Figure 9. Volume rendering 3D images with (a) high (50%) and (b) low opacity (12%). The same transfer functions as in Figure 2 a, b have been used. (c) The same data set as in (a) but with an applied multiplanar cut plane in order to omit obscuring parts of the volume. Three stones are visualized in the common bile duct. Two of these were verified at endoscopic retrograde cholangiography.
|
|
|
Discussion
|
|---|
Currently, most examinations of the biliary system are made with ultrasound or MRI. Ultrasound is widely available and comparatively inexpensive, but when used for imaging the lower part of the choledochus and CBD, it has poor sensitivity for detecting stones (range 22–75%) [2–4]. The poor sensitivity of ultrasound to detect CBD stones is mainly due to incomplete bile duct visualization and overlying bowel gas [3, 4]. Ultrasound must therefore be considered primarily a screening method for assessing the presence and extent of dilated intrahepatic and extra-hepatic bile ducts.
In patients with an inconclusive ultrasound examination, MR cholangiography (MRC) can be performed prior to ERC and surgery [5–8]. MRC is non-invasive, does not require any contrast agent nor expose patients to radiation. Bile ducts proximal to stenoses are easily visualized. However, the spatial resolution is limited, and small bile duct stones are not detectable. The sensitivity and the specificity of MRC are approximately equal to those of ERC: 88% to 95% and 89% to 96%, respectively [9]. MRI is also considered to constitute the state of the art when diagnosing choledocholithiasis, cystic disorders, benign stenoses and cholangiocarcinoma.
Endoscopic retrograde cholangiography (ERC), which remains the reference method for biliary stone imaging, allows diagnosis, treatment, and the evaluation of the therapeutic effect in a single examination. It provides a high spatial resolution image and is not dependent on biliary excretion rates, since the contrast medium is administered directly. On the other hand, it cannot be used for the functional analysis of excretion mechanisms, and the procedure is also invasive, time-consuming and requires well-trained clinicians. The evaluation of bile duct carcinomas is difficult or impossible when these are proximal to total or subtotal stenoses.
In patients with inconclusive ultrasound examination and contraindications to MRI, DIC-CT with 2D images and 3D volume rendering reconstruction may be a diagnostic alternative to ERC. In these patients, DIC-CT may also provide sufficient information concerning resectability and the effect of palliative treatment in malignant disorders of the bile ducts. This examination may provide similar pre-therapeutic information as MRI in obstruction with significant dilatation. By using volume rendering, images similar to MRC can also be obtained in cases where MRC is contraindicated or not available. It is safe compared with direct cholangiography (percutaneous transhepatic cholangiography or ERC), but does not offer any immediate therapeutic option.
The protein-binding characteristics essential for biliary contrast media may, however, increase the risk of adverse reactions [3, 10]. Our experience from 153 consecutive DIC-CT exams [unpublished results] is that adverse reactions are rare when a slow infusion is used (<1%).
In this retrospective material, the diagnostic value of DIC-CT was rated as moderate or good in more than 90% of the examinations, and the sensitivity as well as the specificity for diagnosing stones was high. In the two false negative exams, the contrast excretion was too low to visualize the bile duct. In three cases, stones were observed during DIC-CT but not during ERC (Figures 2, 3, 9). Since ERC and surgery have been considered the gold standard in this study, these three cases are regarded as false positive, but a conceivable explanation for this is that they were too small to be visualized at ERC, or that they had been excreted from the bile duct before the ERC examination.
There was no obvious correlation between accuracy and reader experience in VRT (Table 3). Although the number of final positive stone findings was small (n=16), the lack of correlation suggests that VRT of DIC-CT is rather simple to interpret for an experienced radiologist even when the experience in VRT is moderate.
In our experience, the CT acquisition technique is crucial. In order to enable the visualization of small biliary stones, the slice thickness or reconstruction intervals, i.e. the increment, must be minimal (less than 1–2 mm). Since the examined volume is approximately 30 cm, the minimum number of images to be handled by the workstation is around 300.
While most of the diagnostic information is available in axial CT images [11, 12], the definition of anatomy and anatomical variants is often best appreciated in three-dimensional reformats, e.g. maximum intensity projection (MIP) [13] or VRT [14–16]. In contrast to MIP, the anatomical spatial relation between the organs is retained with VRT. In order to optimize the visualization of the pathology or anatomy of interest, the rendering parameters (opacity, brightness and colour) are altered interactively. The influence on the final image from every single voxel that the ray passes through is thereby changed. For instance, biliary walls can be shown with low opacity, while stones are more easily shown with high opacity (Figure 8). In order not to obscure the regions of clinical interest, overlying structures are either assigned a high transparency value or removed by means of cut planes. This is beneficial, since the biliary tree branches in three dimensions, which is why the traditional axial perspective is not always optimal for diagnostic interpretation (Figure 9). Most often, a pre-defined transfer function is used initially and subsequently, small adjustments may be made.
In this study, the use of VRT improved diagnostic certainty in 14% of the evaluations, and the visualization of ductal stones was improved in 38% of the positive cases. The main reasons for changing the VRT parameter settings were (a) to confirm the initial finding, (b) to optimize the image for the image display and (c) to improve the contrast difference between the biliary duct and the surroundings in cases with poor image quality. No differences between single and multislice CT regarding estimated diagnostic quality were observed. A possible explanation is that the reconstruction of the images at 1 mm intervals for both single and multislice CT compensated for the difference in slice thickness (5 mm for single slice and 2.5 mm for multislice). This indicates that even if a single slice CT is used, 3D DIC-CT can still be performed with acceptable diagnostic value. The weak but significant correlation between the rated diagnostic quality and the measured attenuation in the main biliary duct can be explained by the fact that the contrast between the HU values in the bile duct and liver parenchyma decreased from 261 vs 68 HU for bilirubin levels under 19 µmol l–1 to no difference at levels above 100 µmol l–1 (Figure 7).
A limitation of the study is its retrospective nature. It is, however, unlikely that observation of biliary stones or strictures at ERC or surgery was not noted in the medical records. The follow-up period in this study was also relatively long (57 months). All patients with stones or biliary strictures of clinical interest should therefore have returned to the hospital during this period. The hospital is the only one in the region. The probability that patients with undiagnosed stones or strictures (i.e. false negative cases) might have been diagnosed and treated at another hospital is therefore small.
This material included only three surgically verified strictures. The frequency of strictures was thereby too low to allow any estimation of sensitivity and specificity. However, the high rating of image and diagnostic quality indicates that there is a potential for DIC-CT to visualize partial strictures, although this must be verified in a larger material. No false positive strictures were observed.
|
Conclusion
|
|---|
DIC-CT with bilirubin-governed infusion time and volume rendering post-processing produces detailed images of the biliary tree resulting in good sensitivity and specificity.
Received for publication April 12, 2005.
Revision received May 11, 2005.
Accepted for publication May 13, 2005.
|
References
|
|---|
- Stockberger SM, Wass JL, Sherman S, Lehman GA, Kopecky KK. Intravenous cholangiography with helical CT: comparison with endoscopic retrograde cholangiography. Radiology 1994;192:675–80.[Abstract/Free Full Text]
- Dong B, Chen M. Improved sonographic visualization of choledocholithiasis. J Clin Ultrasound 1987;15:185–90.[Medline]
- Pasanen P, Partanen K, Pikkarainen P, Alhava E, Pirinen A, Janatuinen E. Ultrasonography, CT, and ERCP in the diagnosis of choledochal stones. Acta Radiol 1992;33:53–6.[Medline]
- Stott MA, Farrands PA, Guyer PB, Dewbury KC, Browning JJ, Sulton R. Ultrasound of the common bile duct in patients undergoing cholecystectomy. J Clin Ultrasound 1991;19:73–6.[Medline]
- Barish MA, Yucel EK, Soto JA, Chuttani R, Ferracci JT. MR cholangiopancreatography: efficacy of three-dimensional turbo spin-echo technique. AJR Am J Roentgenol 1995;165:295–300.[Abstract/Free Full Text]
- Bret PM, Reinhold C, Taourel P, Guibaud L, Atri M, Barkun AN. Pancreas divisum: evaluation with MR pancreatography. Radiology 1996;199:99–103.[Abstract/Free Full Text]
- Chan YL, Chan AC, Lam WW, Lee DW, Chung SS, Sung JJ, et al. Choledocholithiasis: comparison of MR cholangiography and endoscopic retrograde cholangiography. Radiology 1996;200:85–9.[Abstract/Free Full Text]
- Regan F, Fradin J, Khazan R, Bohlman M, Magnuson T. Choledocholithiasis: evaluation with MR cholangiography. AJR Am J Roentgenol 1996;167:1441–5.[Abstract/Free Full Text]
- Kaltenthaler E, Vergel YB, Chilcott J, Thomas S, Blakeborough T, Walters SJ, et al. A systematic review and economic evaluation of magnetic resonance cholangiopancreatography compared with diagnostic endoscopic retrograde cholangiopancreatography. Health Technol Assess 2004;8:1–89.[Medline]
- Nilsson U. Adverse reactions to iotroxate at intravenous cholangiography. A prospective clinical investigation and review of the literature. Acta Radiol 1987;28:571–5.[Medline]
- Klein HM, Wein B, Truong S, Pfingsten FP, Gunther RW. Computed tomographic cholangiography using spiral scanning and 3D image processing. Br J Radiol 1993;66:762–72.[Abstract]
- Van Beers BE, Lacrosse M, Trigaux JP, de Cannieri L, De Ronde T, Pringot J. Noninvasive imaging of the biliary tract before or after laparoscopic cholecystectomy: use of three dimensional spiral CT choloangiography. AJR Am J Roentgenol 1994;162:1331–5.[Abstract/Free Full Text]
- Fleischmann D, Ringl H, Schoefl R, Potzi R, Kontrus M, Henk C, et al. Three-dimensional spiral CT cholangiography in patients with suspected obstructive biliary disease: comparison with endoscopic retrograde cholangiography. Radiology 1996;198:861–7.[Abstract/Free Full Text]
- Mallouhi A, Chemelli A, Judmaier W, Giacomuzzi S, Jaschke WR, Waldenberger P. Investigation of cerebrovascular disease with MR angiography: comparison of volume rendering and maximum intensity projection algorithms-initial assessment. Neuroradiology 2002;44:961–7.[CrossRef][Medline]
- Mallouhi A, Felber S, Chemelli A, Dessl A, Auer A, Schocke M, et al. Detection and characterization of intracranial aneurysms with MR angiography: comparison of volume-rendering and maximum-intensity-projection algorithms. AJR Am J Roentgenol 2003;180:55–64.[Abstract/Free Full Text]
- Mallouhi A, Schocke M, Judmaier W, Wolf C, Dessl A, Czermak BV, et al. 3D MR angiography of renal arteries: comparison of volume rendering and maximum intensity projection algorithms. Radiology 2002;223:509–16.[Abstract/Free Full Text]
Top
Abstract
Introduction
