| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiodiagnostic, Baskent University Faculty of Medicine, Fevzi
Çakmak caddesi 10.sokak, No: 45, 06490 Bahçelievler, Ankara,
Turkey.
2 Department of General Surgery, Baskent University Faculty of Medicine, Ankara,
Turkey.
3 Department of Public Health, Baskent University Faculty of Medicine, Ankara,
Turkey.
Received August 9, 2007;
accepted after revision December 12, 2007.
Address correspondenceto C. Ba
aran
(ceylab{at}baskent-ank.edu.tr).
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Forty potential LDLT donors who ranged in age from 19 to 54 years were prospectively evaluated with preoperative MRCP. MRCP was performed with a 1.5-T magnetic field using T2-weighted PACE TSE sequence. MRCP source data sets were processed with maximum-intensity-projection (MIP) and shaded surface display (SSD) algorithms. Findings were compared with intraoperative cholangiography. Biliary anatomy was classified according to the classification proposed by Huang and colleagues. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of MRCP for the detection of aberrant biliary anatomy were calculated.
RESULTS. Intraoperative cholangiography and biliary exploration revealed that 27 donor candidates (67.5%) had conventional and 13 (32.5%) had aberrant biliary anatomy. Two donors (5%) had type B biliary anatomy; eight donors (20%), type C; two donors (5%), type D; and one donor (2.5%), unclassified. The sensitivity of MRCP source data sets in differentiating aberrant biliary anatomies from nonaberrant ones was 100%, the specificity was 88.9%, and the accuracy was 92.5%. PPV and NPV were 81.3% and 100%, respectively. The sensitivity of MIP images in differentiating aberrant biliary anatomies was 100%, the specificity was 88.9%, and the accuracy was 92.5%. PPV and NPV were 81.3% and 100%, respectively. The sensitivity, specificity, accuracy, PPV, and NPV of the SSD images in detecting aberrant biliary anatomies were 100%, 77.8%, 85%, 68.4%, and 100%, respectively.
CONCLUSION. Preoperative MRCP using a respiratory navigator-triggered T2-weighted TSE sequence with a PACE technique accurately depicts the biliary anatomy in LDLT donors and may guide intraoperative management of the biliary tract.
Keywords: anatomy • biliary system • liver disease • liver transplantation • living donor liver transplantation • MR cholangiopancreatography
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
At several liver transplantation centers, physicians also preoperatively evaluate the biliary tract anatomy because variant biliary anatomy is seen in up to 45% of the population. Imaging of the biliary anatomy also helps minimize postoperative biliary complications [4, 5]. The biliary anatomy is mainly evaluated by ERCP, MR cholangiopancreatography (MRCP), or intraoperative cholangiography.
MRCP, a noninvasive imaging technique that is useful in evaluating the biliary system, is being performed with increasing frequency [6]. MRCP using thick- and thin-slab heavily T2-weighted sequences has been used widely to depict the biliary tree. Although good results in evaluating relevant biliary anatomy before laparoscopic surgery and in evaluating living liver donors have been reported with this technique, detecting and defining intrahepatic anatomic anomalies, particularly in nondilated systems, are often inadequate [7-9].
In the current study, we used a T2-weighted prospective acquisition correction (PACE) technique turbo spin-echo (TSE) sequence. This pulse sequence has all the advantages of a regular 3D pulse sequence. Moreover, this pulse sequence allows the patient to breathe freely during the PACE technique. This technique allows acquisition of images free of motion artifacts in the abdomen without the need for apnea, which is useful in uncooperative patients. Whereas previous studies have compared the accuracy of standard MRCP using thick- and thin-slab heavily T2-weighted sequences and of mangafodipir trisodium-enhanced MRCP and gadobenate dimeglumine-enhanced MRCP using a maximum intensity-projection (MIP) algorithm to depict the biliary anatomy in LDLT donors, knowledge regarding the accuracy of a respiratory-triggered T2-weighted TSE sequence with PACE using MIP and shaded surface display (SSD) algorithms is nonexistent.
The purpose of this study was to evaluate the ability of a respiratory-triggered T2-weighted TSE sequence with PACE for MRCP to depict the biliary anatomy of LDLT donors, correlate these findings with those obtained with intraoperative cholangiography, and compare the accuracy of two rendering algorithms.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
MRCP Technique
Imaging was performed with a 1.5-T MR magnet (Magnetom Symphony, Siemens
Medical Solutions) using a 4-channel body phased-array surface coil as a
radiofrequency receiver. A respiratory-triggered T2-weighted TSE sequence with
the PACE technique (TR/TE, 1,600/678; flip angle, 170°; field of view, 400
mm; matrix size, 384 x 384; section thickness, 1.6 mm; gap, 0; distance
factor, 50) in axial and coronal planes was performed. The acquisition time
for the 3D MRCP sequence was 6 minutes 17 seconds in patients who breathed
slowly and rhythmically; however, it took more time in patients breathing
rapidly and irregularly. We processed MRCP data sets with MIP and SSD
algorithms. The source images were obtained in two planes, providing better
anatomic orientation. A standard defined protocol was used for 3D reformatted
images. For the image analysis, a series of 19 projections rotated by 10°
intervals from -90° to 90° was created for each rendering algorithm.
The reconstructions were obtained in the coronal plane.
MRCP source images and MIP and SSD images were closely monitored by the radiologist. In general, the radiologist evaluated visualization of the common duct, right and left intrahepatic ducts, and insertion of the right posterior lobe duct and left medial lobe duct.
Intraoperative Cholangiography
Intraoperative cholangiography was performed by surgeons before hepatectomy
in all 40 patients. After cholecystectomy, the cystic duct remnant was
cannulated, and 10-20 mL of iohexol (Omni paque, GE Healthcare) was hand
injected to opacify the intrahepatic biliary system under fluoroscopic
guidance. Anteroposterior and oblique views were obtained. Although the
preliminary surgical plan was based on MRCP, the final decision was based on
intraoperative cholangiography as the reference standard.
Image Analyses and Statistical Analyses
MIP and SSD reformations of MRCP were obtained using a dedicated
workstation (Leonardo, Siemens). Two radiologists experienced in MRCP
classified all source and reconstructed MR images at the workstation.
Discrepancies were resolved based on agreement of the radiologists.
Details about the biliary anatomy that were noted included configuration of the main division and drainage of second-order biliary radicals as classified by Huang and colleagues [10] (Figs. 1A, 1B, 1C, 1D and 1E). The most common configurations were, first, conventional anatomy of the right and left hepatic ducts forming a common hepatic duct (type A) (Figs. 2A and 2B); second, trifurcation formed by the right anterior sectoral branch, right posterior sectoral branch, and left hepatic duct (type B) (Figs. 3A, 3B and 3C); third, drainage of the right posterior sectoral branch into the left hepatic duct (type C) (Figs. 4A, 4B, 4C, 4D and 4E); fourth, drainage of the right posterior sectoral branch into the common hepatic duct (type D) (Figs. 5A and 5B); and, fifth, drainage of the right posterior sectoral branch into the cystic duct (Figs. 6A and 6B).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Image sets were presented in a different, randomized order to prevent consistent bias in the interpretation of one image set based on a prior viewing of a different image set of the same patient. Between analyses of the three different image sets, there was a minimal time interval of 1 week. Patient identifiers were removed for blind review.
Each reviewer classified 120 image sets (MRCP source data and MIP and SSD reconstructed images for each of 40 patients) for biliary anatomy according to Huang et al. [10] and compared the accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of each rendered algorithm and source data with the intraoperative biliary findings in all patients derived from intraoperative cholangiography. We used the results of intraoperative cholangiography as the reference standard for our study.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
MRCP Source Data Sets
MRCP source data sets indicated that 24 patients (60%) had conventional, 15
(37.5%) had aberrant (Figs.
4C,
4D and
4E), and one (2.5%) had
unclassified biliary anatomy. Among the donors with aberrant biliary anatomy,
MRCP source data sets were able to detect two of two patients with type B
anatomy, seven of eight with type C, and two of two with type D. Three
patients with conventional anatomy were indicated as having type B anatomy,
and one patient with type C was indicated as having type B. The overall
sensitivity of MRCP source data sets in differentiating variant donor biliary
anatomy from nonvariant anatomy was 100%, specificity was 88.9%, and accuracy
was 92.5%. PPV and NPV were 81.3% and 100%, respectively.
Maximum Intensity Projection
MIP images indicated that 24 patients (60%) had conventional, 15 (37.5%)
had aberrant (Figs. 3A,
4A, and
5A), and one (2.5%) had
unclassified biliary anatomy (Fig.
6A). MIP correctly predicted
conventional biliary anatomy (type A) in 24 of 27 patients (Fig.
2A). Three patients with type A
were indicated as having type B anatomy, and one patient with type C was
indicated as having type B anatomy. The overall sensitivity of MIP images in
differentiating variant donor biliary anatomy from nonvariant anatomy was
100%, specificity was 88.9%, and accuracy was 92.5%. PPV and NPV were 81.3%
and 100%, respectively.
Shaded Surface Display
SSD images indicated that 21 patients (52.5%) had conventional (Fig.
2B), 16 (40%) had aberrant
(Figs. 3B,
4B, and
5B), and one (2.5%) had
unclassified biliary anatomy (Fig.
6B). SSD images were
inadequate to determine biliary anatomy in two patients (5%). In four
patients, intraoperative cholangiography showed type A, whereas SSD revealed
type B, and one patient with type A was classified as type C. One patient with
type A was not classified with SSD images. SSD correctly predicted seven of
eight type C. In one patient with type C, we were unable to define the biliary
anatomy with SSD. SSD correctly predicted one patient with unclassified
biliary anatomy. The sensitivity, specificity, accuracy, PPV, and NPV of SSD
in detecting variant donor biliary anatomy were 100%, 77.8%, 85%, 68.4%, and
100%, respectively.
A type 1 choledochal cyst was observed in one patient with trifurcation. No donor had aberrant drainage of a segment IV sectoral branch. A segment IV sectoral branch was observed in MRCP source data in 29 patients (72.5%) and in MRCP source data and MIP views in four patients (10%). A segment IV sectoral branch was not visualized with MRCP source data and reconstructed images in seven patients.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Accurate preoperative radiologic imaging is essential to assess the vascular and biliary anatomies of a living donor candidate. A correct understanding of a donor's biliary anatomy is crucial for safe donor hepatectomy and to minimize recipient biliary complications [15]. Anatomic variation of the biliary tree of an LDLT donor is not a contraindication for donor hepatectomy. Thus, if multiple donor candidates are available, the optimal donor can be selected according to the biliary anatomy [16].
MRCP has potential as a noninvasive, non-biohazardous diagnostic technique for evaluating LDLT donors. Various MRCP techniques have been reported in the literature. TSE sequences are superior to gradient-echo sequences owing to higher signal-to-noise ratios, higher spatial resolution, and fewer motion artifacts. The application of TSE imaging, based on rapid acquisition with the relaxation enhancement technique introduced by Hennig and associates [17], significantly improved image quality. TSE sequences can be performed as 2D or 3D acquisitions: A single-shot projectional technique and a multislice technique are available for MRCP. However, some limitations compromise the diagnostic potential of MRCP. Among these, respiratory motion significantly influences image quality. Breath-hold imaging effectively eliminates respiratory motion artifacts. Nevertheless, some patients are unable to hold their breath sufficiently. Spatial resolution can be improved in a longer acquisition time. The breath-hold multislice technique also has a relatively short acquisition time; image quality is hampered by a low signal-to-noise ratio and low spatial resolution. The respiratory-triggered multislice technique offers the possibility of extending the acquisition time, which can be invested into a higher spatial resolution. Respiration-triggered liver imaging has been shown to lead to sharper anatomic details, improved liver-lesion contrast, and fewer artifacts; it also improves detection of focal liver lesions [18]. Several recent reports, as shown in Table 1, have provided encouraging results about standard MRCP using thick- and thin-slab single-shot fast spin-echo (SSFSE) sequences.
|
According to Goldman and associates [15], standard MRCP seems inadequate for identifying the intrahepatic biliary ducts. Thick- and thin-slab half-Fourier SSFSE imaging showed only a small percentage of the right anterior and posterior branches (35% and 12%, respectively) and did not depict the left medial and lateral branches. A conventional MRCP technique (half-Fourier acquisition) has several characteristics that limit detailed depiction of the biliary tree, especially in nondilated ducts.
According to another study [19], standard MRCP using a T2 SSFSE or a T2 HASTE sequence was highly accurate in preoperatively mapping the biliary tracts of 26 LDLT donors. In that study, MRCP was 84.6% accurate in preoperative biliary mapping, although the study enrolled a small number of subjects. In another study, Song and associates [16] reported a large number of LDLT donors and used the same MR technique (half-Fourier rapid acquisition with relaxation enhancement) and the same standard for comparison (intraoperative cholangiography). For MRCP, 95.5% sensitivity, 95.2% specificity, 96.8% PPV, and 93.3% NPV were reported. These findings indicate that conventional unenhanced MRCP has potential for the preoperative assessment of nondilated biliary systems in LDLT donors, although it may not provide adequate information for accurate assessment in a minority of patients with delicate bile ducts.
Mangafodipir trisodium (MnDPDP)-enhanced MRCP has been reported to be as accurate as intraoperative cholangiography, suggesting its use for preoperative planning of biliary anastomoses. Ayuso and associates [7] compared MnDPDP-enhanced MRCP with intraoperative cholangiography in the preoperative evaluation of the biliary anatomy of adult living liver donors. The sensitivity, specificity, accuracy, PPV, and NPV were 93.7%, 100%, 86.4%, 100%, and 90%, respectively. However, the MR protocol using MnDPDP has limitations including the need for additional table time owing to the requirement of a 10- to 15-minute delay after slow injection of contrast agent for biliary excretion. Furthermore, the contrast agent is expensive, is not widely used, and is presently unavailable in some countries [7, 8, 15, 19].
In another study [20], T1-weighted 3D gradient-echo images that were obtained 1 hour after injection of gadobenate dimeglumine revealed contrast excretion into the biliary tree and provided cholangiographic images. The sensitivity and specificity for detecting variant biliary anatomy were 50% and 100%, respectively. In that study, combined interpretation of both T1-weighted 3D gradient-echo images and T2-weighted HASTE images resulted in defining the biliary anatomy for a sensitivity of 90%, specificity of 93%, and diagnostic accuracy of 92% for detecting the variations.
To date, breath-holding sequences dominate MRCP; respiratory-triggered MRCP is performed infrequently [21]. The respiratory-triggering technique used for our study is based on a completely different technique, making no additional hardware components necessary. The T2-weighted PACE TSE is interleaved with a navigator sequence tracking the movement of the right diaphragm, identifying the respiratory phase with the least motion to place data acquisition into the respiratory cycle. This technique has been named "PACE"; it can be applied to any kind of T2-weighted MR sequence. The application for MRCP is realized with a heavily T2-weighted TSE sequence with a 3D acquisition, which allows projection reformation—in particular, the MIP algorithm [22].
Three-dimensional algorithms used to view images in multiple orthogonal planes are also helpful to evaluate the relationships of the right, left, and common hepatic ducts. Several rendering algorithms have been described for processing 3D MRI data including MIP, volume rendering (VR), and SSD. MIP is the simplest and is still the most commonly used image-processing algorithm. In this algorithm, a "virtual beam" is projected through the stack of image planes from a defined direction. Only the pixels with the highest signal intensity along the projection line amount to a pixel in the final projection. One reason for the broad application of MIP is its short acquisition time and computational requirements. The major disadvantage of MIP, however, is its substantial loss of image information, especially of low-signal-intensity features. This loss may cause a loss of small ducts in the final projection, especially when structures of high signal intensity are present in the route of the projection.
Surface rendering refers to a technique in which data are segmented according to user-defined signal thresholds and the processed image displays of the structure of the surface. With additional depth encoding by adding shades resulting from a virtual light source, this rendering method is referred to as SSD. The advantage of SSD compared with other rendering algorithms is the need for minimal computational power. Even more than MIP, SSD results in a substantial loss of image information because the segmentation uses a simple binary classification. Consequently, narrow threshold selection may result in artificial narrowing or even nonvisualization of the small ducts in the final 3D projection.
VR is an algorithm that makes use of different techniques such as segmentation, gradient calculation, resampling, classification, shading, and illumination models. The segmentation technique allows the subdivision of data and allows a specific object within the volume to be identified. To identify the surfaces within the volume, a gradient calculation is necessary. The main feature of VR is the use of all the information set inside the volume to reconstruct 3D images; no threshold values are used and there is no data loss. Voxels are not limited to an all-or-nothing contribution to the rendered image as they are in SSD and MIP. VR is the more recently developed technique and produces high-quality images that enclose both spatial and signal intensity information [23].
In our study, a T2-weighted PACE TSE sequence was used for the first time, to our knowledge, to evaluate preoperatively the biliary anatomy of potential LDLT donors and MRCP data source images. Different rendering algorithms such as MIP and SSD were compared with intraoperative cholangiography. When our study results were compared with those of other recent studies, MRCP source data, MIP, and SSD images using a T2-weighted PACE TSE sequence showed the highest sensitivities and accuracies for preoperatively evaluating the biliary anatomy. In this study, the accuracy of MIP for assessing preoperative biliary anatomy was equal to that of MRCP source data; however, the accuracy of MIP was superior to that of SSD.
A major limitation of MRCP previously was depicting the peripheral intrahepatic biliary tree. In our study, T2-weighted PACE TSE improved visualization of the peripheral intrahepatic biliary tree to the level of the interlobular bile duct, especially in source images and MIP images. The peripheral intrahepatic biliary tree was not evaluated by rating the quality of duct visualization, which should be subject to future studies.
A major limitation of the free-breathing T2-weighted PACE TSE sequence was that patients had to breathe slowly and rhythmically (ideal for respiratory triggering). Therefore, free breathing is not a good option in patients who breathe rapidly and irregularly.
In conclusion, in the absence of a deceased-donor liver for transplantation, LDLT is an option for patients with end-stage liver disease. In evaluating living donors for a liver transplant, as correlated with intraoperative cholangiography, our study shows that the T2-weighted PACE TSE technique allows significantly good visualization of the peripheral intrahepatic biliary tree with MIP and SSD 3D-rendering algorithms. The T2-weighted PACE TSE technique should be included in the standard protocol of preoperative radiologic evaluation of living donor candidates.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
