| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Radiology, Asan Medical Center, University of Ulsan, College of
Medicine, 388-1 Poongnap-dong, Songpa-gu, Seoul 138-736, South Korea.
2 Department of Radiology, Inje University, Sanggyepaik Hospital, Seoul 139-707,
South Korea.
Received November 7, 2003;
accepted after revision February 2, 2004.
Address correspondence to J. K. Kim
(rialto{at}smc.seoul.kr).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. In 76 patients who underwent donor nephrectomy, vascular phase CT data were obtained using an MDCT scanner (detector array, 1.25 mm x 4; beam pitch, 1.5). Two radiologists independently reconstructed CT angiographic images using thick-slab volume rendering, thick-slab maximum intensity projection (MIP), sliding thin-slab volume rendering, and sliding thin-slab MIP. The radiologists counted the number of renal arteries, early branching arteries, and renal veins. We compared the accuracy rates for the detection of vessels achieved with the four types of reconstructed images, using the surgical findings as the gold standard. Agreement between the two observers and between the surgical and CT angiographic findings was evaluated.
RESULTS. The sensitivity for detecting the supernumerary artery was
significantly greater with sliding thin-slab volume rendering and sliding
thin-slab MIP (97%) than with thick-slab volume rendering (59%) (p =
0.039). No significant difference between the other comparison pairs of
reconstruction methods was found. The interobserver agreement for detecting
supernumerary and early branching arteries with sliding thin-slab volume
rendering and MIP was excellent (
= 0.820-0.859) and good for renal
veins ( = 0.698-0.724), whereas the agreement of thick-slab volume
rendering and MIP was good for arteries ( = 0.706-0.791) and moderate
for veins ( = 0.443-0.579). The agreement between CT angiographic
reconstructed images and surgical findings for detection of vessels was better
with sliding thin-slab volume rendering and MIP ( = 0.793-1.000) than
in thick-slab volume rendering and MIP ( = 0.306-0.613).
CONCLUSION. For CT angiographic evaluation of living renal donors, sliding thin-slab reconstruction is superior to thick-slab reconstruction.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In this study, we compared various reconstruction methods for CT angiography in evaluating living renal donors. The purpose of our study was to investigate the adequacy of reconstruction methods for CT angiography in assessing living donors.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Study Design and Patient Population
Between August 2001 and November 2002, 76 consecutive patients (45 men and
31 women; age range, 27-52 years; mean age ± standard deviation, 39
± 9 years) underwent CT angiography and then open extraperitoneal donor
nephrectomy at our institution. In those patients, we retrospectively
evaluated CT angiographic images reconstructed by various methods.
CT Image Acquisition
All CT data were obtained using a 4-MDCT scanner (LightSpeed QX/i, GE
Healthcare). Patients received IV contrast material ([iopamidol] Iopamiro 300,
Bracco) in an antecubital vein via a mechanical injector; 130-140 mL was
administered at a rate of 3.5-4.0 mL/sec. Using an automatic bolus tracking
method (Smart Prep, GE Healthcare), scanning was initiated, with a delay of 5
sec after triggering at a threshold of 80 H in the region of interest in the
abdominal aorta at the level of the renal arteries. With this technique, the
scanning began 24-30 sec (mean, 27 sec) after the start of IV contrast
material injection and covered the region from the diaphragm to the midpelvis.
The scanning parameters included a detector array of 1.25 mm x 4, a beam
pitch of 1.5 (equivalent to a slice pitch of 6, high-speed mode), a table
speed of 7.5 mm per rotation (9.38 mm/sec), and a reconstruction increment of
0.5 mm.
Image Processing
To reconstruct the CT angiographic images, we transferred the source CT
data to a workstation (Advantage Windows version 3.0, GE Healthcare) that
allows real-time interactive manipulation of images. Two experienced
radiologists who were unaware of surgical findings, results of initial CT
angiography reports, and patient-identifying information independently
performed image reconstruction and interpretation. They used four different
reconstruction methods in the following order: thick-slab volume rendering,
thick-slab maximum intensity projection (MIP), sliding thin-slab volume
rendering, and sliding thin-slab MIP. In thick-slab reconstruction, the
reconstructed images contained the entire thickness of renal vessels and
kidneys; the slab thickness was 6-8 cm. By contrast, in thin-slab
reconstruction, the reconstructed images covered only part of the thickness of
the renal vessels to emphasize the vascular visualization; the slab thickness
was approximately 2-10 mm. Both radiologists evaluated renal vessels in the
oblique transverse and oblique coronal planes while rotating images of the
kidneys in either plane. In the assessments of each reconstruction technique,
the order of patients was randomized.
First, CT angiographic images were generated using thick-slab volume rendering. Each radiologist segmented kidneys, renal arteries, renal veins, the aorta, the inferior vena cava, and perinephric fat on the source CT scans by manually drawing a 3D region of interest. The radiologists then deleted all the other structures while attempting to avoid deleting the regions of interest. Using this technique, additional detailed editing was performed to eliminate irrelevant structures such as the mesenteric and lumbar arteries and residual bones. From these edited CT scans, CT angiographic images were subsequently reconstructed.
Two weeks after obtaining CT angiographic images with thick-slab volume rendering, the radiologists reconstructed the images using thick-slab MIP. The segmentation and generation of the regions of interest were performed in the same manner that had been used in thick-slab volume rendering. Two weeks after obtaining the thick-slab MIP images, the radiologists reconstructed CT angiographic images using sliding thin-slab volume rendering. Adjunctive image segmentation was not used. Finally, 2 weeks after obtaining sliding thin-slab volume-rendered images, the radiologists reconstructed the CT angiographic images with sliding thin-slab MIP in the same manner as they had obtained images reconstructed with sliding thin-slab volume rendering. The time spent on CT angiographic image reconstruction and interpretation was limited to 15 min for each technique in each patient.
Image Interpretation
In each of the four reconstruction sessions, the two observers
independently recorded the number of renal arteries, early branching arteries,
and veins in the kidneys. If the kidney had two or more arteries or veins, the
vessel with the greatest diameter was considered to be dominant and the others
to be supernumerary. An early branching artery was determined to be present if
any branch diverged from the aorta by 1 cm or less from the aorta. The
presence or absence of arterial abnormalities, such as stenosis or calcified
plaque, and venous abnormalities, such as a retroaortic or circumaortic vein,
was also evaluated.
Surgical Correlation
Seventy-six of 152 kidneys were donated for transplantation. The donated
kidneys were selected on the basis of the CT angiography reports. The presence
or absence of complex vascular anatomy was the principal consideration, and if
the vascular anatomy was simple, the left kidney was preferred for donation
because of the greater length of its veins. The surgeons recorded the number
of arteries, early branching arteries, and veins in the surgical field as well
as vessel abnormalities in each kidney, applying the same criteria used in CT
angiography.
In cases in which the observers' evaluations differed, the final CT angiographic results for each reconstruction method in each of the 76 donated kidneys were decided by consensus. Then, the final CT angiographic findings were compared with the surgical findings.
Statistical Analysis
The agreement between the two observers and between the CT angiographic and
surgical findings was quantified by using kappa statistics: a kappa value of
less than 0.20 was considered to indicate poor agreement; 0.21-0.40, fair;
0.41-0.60, moderate; 0.61-0.80, good; and 0.81-1.00, excellent.
The sensitivity, specificity, positive predictive value, negative predictive value, and overall accuracy were calculated per donated kidney, and the surgical findings were regarded as the gold standard. Thereafter, we compared the diagnostic accuracy for the four different reconstruction methods using the McNemar test.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
On CT angiography with sliding thin-slab volume rendering, the first observer found 192 renal arteries with 35 supernumerary arteries and 18 early branching arteries, whereas the second observer identified 192 arteries with 35 supernumerary arteries and 19 early branching arteries. On CT angiography with sliding thin-slab MIP, the first observer noted 193 renal arteries with 35 supernumerary arteries and 20 early branching arteries, and the second observer found 195 renal arteries with 37 supernumerary arteries and 21 early branching arteries.
The interobserver agreements for the number of supernumerary arteries and
early branching arteries are shown in Table
1. The agreements for both supernumerary and early branching
arteries were good ( = 0.706-0.791) in thick-slab volume rendering and
thick-slab MIP and excellent in sliding thin-slab volume rendering and sliding
thin-slab MIP ( = 0.820-0.859). The two observers agreed on the number
of supernumerary and early branching arteries in more than 90% of 152 kidneys
viewed on the four types of reconstructions. Neither observer found arterial
abnormalities such as strictures or calcified plaque on CT angiography with
any of the four reconstruction methods.
|
Renal Vein
In the 152 kidneys, the first observer noted 162 renal veins on the images
reconstructed with thick-slab volume rendering; 165, with thick-slab MIP; 174,
with sliding thin-slab volume rendering; and 171, with sliding thin-slab MIP.
The second observer found 166 renal veins in the images reconstructed with
thick-slab volume rendering; 164, with thick-slab MIP; 174, with sliding
thin-slab volume rendering; and 169, with sliding thin-slab MIP.
The interobserver agreement for the number of renal veins is shown in
Table 1. The interobserver
agreement was moderate for thick-slab volume rendering and thick-slab MIP
( = 0.443-0.579) and good for sliding thin-slab volume rendering and
sliding thin-slab MIP ( = 0.698-0.724). The observers agreed on the
number of renal veins in more than 90% of 152 kidneys viewed on the four types
of reconstructions. Neither observer found venous abnormalities on CT
angiography with any of the four reconstruction methods.
Surgical Correlation
On the basis of the CT angiographic findings, 59 left and 17 right kidneys
were chosen for donor nephrectomy. In those kidneys, surgeons identified 93
arteries: 62 kidneys had single arteries; 12 kidneys had two arteries in each
kidney; one kidney had three arteries; and one kidney had four arteries.
Therefore, 14 kidneys had 17 supernumerary arteries.
The detection rates for renal arteries on CT angiography were 92% (86/93) with thick-slab volume rendering, 94% (87/93) with thick-slab MIP, 99% (92/93) in sliding thin-slab volume rendering, and 98% (91/93) in sliding thin-slab MIP (Fig. 1A, 1B, 1C, 1D).
|
|
|
|
Comparisons between surgical and CT angiographic findings with four reconstruction methods for the number of supernumerary arteries seen in 76 donated kidneys are summarized in Table 2. The sensitivity for detecting supernumerary arteries was significantly greater with sliding thin-slab volume rendering (93%) and sliding thin-slab MIP (93%) than with thick-slab volume rendering (50%) (p = 0.038). The specificity for detection was greater with sliding thin-slab volume rendering and sliding thin-slab MIP than with thick-slab MIP, but the difference was not statistically significant (p > 0.05). The sensitivity for detection of supernumerary arteries with volume rendering and MIP in the same slab thickness was similar (p > 0.05). The specificity, positive predictive value, negative predictive value, and overall accuracy were similar for the four reconstruction methods (p > 0.05).
|
The agreement between CT angiographic and surgical findings was excellent
for sliding thin-slab volume rendering and sliding thin-slab MIP ( =
0.819 and 0.869, respectively), good for thick-slab MIP ( = 0.613), and
moderate for thick-slab volume rendering ( = 0.485).
Overall, the number of supernumerary arteries detected was correct in 65 kidneys (86%) with thick-slab volume rendering, in 68 (89%) with thick-slab MIP, in 73 (96%) with sliding thin-slab volume rendering, and in 72 (95%) with sliding thin-slab MIP (p > 0.05).
At surgery, surgeons identified six early branching arteries. Comparisons between CT angiographic and surgical findings for the number of early branching arteries are summarized in Table 3. The sensitivity and positive predictive value for detecting early branching arteries were greater with sliding thin-slab reconstruction than with thick-slab reconstruction, but the difference was not statistically significant (p > 0.05) (Fig. 2A, 2B, 2C, 2D). The specificity, negative predictive value, and overall accuracy were similar for the four reconstruction methods (p > 0.05).
|
|
|
|
|
The agreement between CT angiographic and surgical findings was excellent
with sliding thin-slab volume rendering and sliding thin-slab MIP ( =
0.819 and 1.000, respectively) but moderate with thick-slab volume rendering
and thick-slab MIP ( = 0.510 and 0.573, respectively).
Surgeons identified a total of 79 veins from the 76 donated kidneys; three
kidneys had two veins each. The detection rates for the renal vein on CT
angiography were similar with all four reconstruction methods (97-99%;
p > 0.05). The agreement between CT angiographic and surgical
findings was good with sliding thin-slab volume rendering and sliding
thin-slab MIP ( = 0.793), moderate with thick-slab volume rendering
( = 0.490), and fair with thick-slab MIP ( = 0.306).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Sliding thin-slab reconstruction was first introduced in 1993 as a form of sliding thin-slab MIP and has recently become available in commercial software [16]. Investigators in previous studies have shown that this reconstruction method can provide high contrast of the thin original section and reduce the influence of the partial volume effects while retaining the benefits of improved anatomic orientation. In the radiology literature, use of sliding thin-slab reconstruction has most often been reported in thoracic CT imaging [16-20]. However, the usefulness of sliding thin-slab MIP for CT angiographic evaluation of living renal donors has not been evaluated, and sliding thin-slab volume rendering has not previously been introduced in the radiology literature. To our knowledge, our study is the first attempt to evaluate the value of sliding thin-slab reconstruction in CT angiography of living renal donors.
In thick-slab reconstruction, redundant voxels are processed, thereby leading to a greater expenditure of time and the depiction of unnecessary information. In contrast, sliding thin-slab reconstruction requires only minimal computational complexity because the volume is restricted to a thin slab with a depth of only several voxel widths, and segmentation of the organ of interest is not necessary. Therefore, with this technique, less time is required, and the risk of adjacent structures obscuring or mimicking renal vessels can be reduced.
In interpreting CT angiography, vessel distinction is the most important factor. In thick-slab reconstruction, because many voxels with various attenuations are grouped together, vessels that are inadequately opacified may not be depicted. In MIP reconstruction, high-attenuation voxels representing calcifications or bones may obscure important structures. Sliding thin-slab reconstruction has fewer possible limitations than thick-slab reconstruction because sliding thin-slab reconstruction includes a few voxels representing only vessels and adjacent connective tissues. Usefulness of sliding thin-slab reconstruction may be limited for evaluation of vessels with complex courses, such as hepatic or mesenteric arteries, because the slab cannot include the entire vessel length in a single projection. However, for renal vessels that usually have a straight course, this limitation is less significant.
In our study, the performance of CT angiography with sliding thin-slab reconstruction appeared to be superior to its reported performance in previous studies. According to our data, findings of CT angiography with sliding thin-slab reconstruction agreed with the surgical findings on the number of arteries in 95-96% of the donated kidneys, compared with previously reported values of 90-95% [1-5, 21-24]. Both the sensitivity and specificity for detecting supernumerary arteries were also greater in our study (93% and 98%, respectively) than in an earlier large study (77% and 89%, respectively) [4]. In contrast, our results for thick-slab reconstruction (sensitivity for detecting accessory arteries, 50-57%) seemed to be even worse than those for the review of source CT scans (> 90%) reported in recent studies [25]. These results indicate that sliding thin-slab reconstruction is superior to thick-slab reconstruction.
Although studies have compared the accuracy of volume rendering and MIP for CT angiography in various organs, controversy persists over which reconstruction method is superior. Hong and Freeny [26] reported that volume rendering was superior to MIP for depicting small peripancreatic arteries. Johnson et al. [27] showed that volume rendering was more accurate than MIP for depicting splanchnic vascular anatomy, vascular and visceral spatial interrelations, variant vasculature, tumor encasement, and hepatic tumor localization. In contrast, Byun et al. [8] recently compared volume rendering and MIP for CT angiography of the hepatic artery in potential liver transplant donors and observed that MIP was superior to volume rendering for depicting anatomic variations. In our study, no significant difference was found between volume rendering and MIP in images with the same slab thickness. Furthermore, comparison of volume rendering and MIP may not be critical because it is easy to switch between the two techniques using the current image reconstruction software.
The protocol for this study was selected on the basis of our preliminary experience [28]. We used a 5-sec scanning delay after triggering at a threshold of 80 H in the region of interest in the abdominal aorta at the renal hilar level. With this scanning delay, both the renal arteries and veins are opacified, although arteries are usually more intensely opacified than the veins. In cases in which the difference in the degree of opacification between arteries and veins is not sufficient, performance of thick-slab reconstruction is greatly impaired because the distinction between arteries and veins with similar opacities is significantly limited. In contrast, sliding thin-slab reconstruction may be less affected because differentiation between arteries and veins is easier and more accurate in a thin-slab reconstruction.
Another important factor for vessel distinction is scanning parameters. The MDCT scanner used in our study provides two scanning modes, high-quality mode and high-speed mode, depending on the preferred pitch; a beam pitch of 0.75 is used for high-quality mode and a beam pitch of 1.5 is used for high-speed mode. With the different pitch, high-speed mode provides the faster scanning and thereby can increase the difference in the degree of opacification between arteries and veins, whereas high-quality mode can improve image resolution. We used high-speed mode because we considered the difference in opacification more important for distinction of arteries and veins than improved image resolution.
We did not evaluate the accuracy of the four types of reconstruction methods for detection of renal tumors or calculi that might affect which of the donor's two kidneys could be donated. Such a situation did not occur and was not relevant to our study.
A potential limitation of our study is that the surgeons usually chose to transplant the kidney with the simpler vascular anatomy. Therefore, the accuracy of CT angiographic reconstructions was determined only for less complicated kidneys; the kidneys with more complex vessels had no surgical proof of the number of vessels. This deficiency might have caused a workup bias that led to underestimation of the false-positive or false-negative rate for detecting accessory arteries and early branching arteries.
Another limitation is the possibility that the observers may have involuntarily memorized the renal vasculatures in the kidneys, although image reviews were performed with 2-week time intervals and CT angiographic images were displayed in random order at each session. In addition, thick-slab reconstruction was always interpreted before the sliding thin-slab reconstruction. This fixed order of reconstruction methods could have introduced a bias.
Our study may raise a question concerning the efficacy of reviewing source images because of its similarity to sliding thin-slab reconstruction. A recent study has emphasized the importance of reviewing source images because the investigators found that the source images provided even higher diagnostic accuracy than reconstructed images [27]. Unfortunately, that study did not evaluate the usefulness of sliding thin-slab reconstruction, and our study did not compare source images with reconstructed images. We presume that sliding thin-slab reconstruction has an advantage over source images because sliding thin-slab reconstruction can provide images in any planes that observers need.
In summary, we found that sliding thin-slab reconstructions provided a significantly greater sensitivity for detecting supernumerary arteries than did thick-slab volume rendering, as well as better agreement between observers and greater correlation between the CT angiographic and the surgical findings than did thick-slab reconstructions. We conclude that sliding thin-slab reconstruction is superior to thick-slab reconstruction for CT angiography in the evaluation of living renal donors.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  A. K. Singh, D. V. Sahani, C. R. Kagay, S. P. Kalva, M. C. Joshi, N. Elias, and T. Kawai Semiautomated MIP Images Created Directly on 16-Section Multidetector CT Console for Evaluation of Living Renal Donors Radiology, August 1, 2007; 244(2): 583 - 590. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Rastogi, D. V. Sahani, M. A. Blake, D. C. Ko, and P. R. Mueller Evaluation of Living Renal Donors: Accuracy of Three-dimensional 16-Section CT Radiology, July 1, 2006; 240(1): 136 - 144. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. K. Fishman, D. R. Ney, D. G. Heath, F. M. Corl, K. M. Horton, and P. T. Johnson Volume rendering versus maximum intensity projection in CT angiography: what works best, when, and why. RadioGraphics, May 1, 2006; 26(3): 905 - 922. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. T. Johnson and E. K. Fishman IV Contrast Selection for MDCT: Current Thoughts and Practice Am. J. Roentgenol., February 1, 2006; 186(2): 406 - 415. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. C. Dalrymple, S. R. Prasad, M. W. Freckleton, and K. N. Chintapalli Informatics in Radiology (infoRAD): Introduction to the Language of Three-dimensional Imaging with Multidetector CT RadioGraphics, September 1, 2005; 25(5): 1409 - 1428. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
