| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 All authors: Department of Clinical Science and Bioimaging, Section of Radiology, University "G. D'Annunzio," Chieti, via dei Vestini, Chieti, Italy.
Received August 21, 2006;
accepted after revision January 15, 2007.
Address correspondence to R. Iezzi
(r.iezzi{at}rad.unich.it).
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Fifty patients with abdominal aortic aneurysm treated with endovascular repair who had undergone follow-up MDCT were enrolled in this study. Contrast-enhanced images were obtained with a 4-MDCT scanner (1-mm collimation). Images were reconstructed using a 1-mm (set A), 3-mm (set B), or 5-mm (set C) slice width. Each image set was interpreted by two independent readers for the presence of endoleaks and for image quality on a dedicated workstation. Sensitivity, specificity, and positive predictive values of each reading session were compared.
RESULTS. The statistical values obtained with sets A and B were significantly higher (p < 0.001) than those obtained with set C. No statistically significant differences were found between the values obtained with sets A and B.
CONCLUSION. For the detection of endoleaks at MDCT, the sensitivity of 1- and 3-mm-thick images was significantly higher than that of 5-mm-thick slices. However, no statistically significant differences were found between the 1- and 3-mm image sets; moreover, the use of thinner reconstruction images (1 mm) has the disadvantage of increasing the number of images that must be interpreted and archived.
Keywords: abdominal aortic aneurysms • aorta • CT angiography • CT technique • grafts • stents • vascular imaging
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The introduction of MDCT technology allows noninvasive imaging of the abdominal aorta and iliac arteries with improved spatial resolution and temporal resolution, thanks to the use of a thinner collimation. Moreover, MDCT allows retrospective reconstruction of the original CT data with different slice thicknesses. The higher spatial resolution obtained with the use of a thinner collimation would seem to offer improved sensitivity in the detection of endoleaks. However, the main drawback of routine scanning with a thin collimation is the increase of radiation exposure to the patient; furthermore, the use of a thinner collimation results in increased image noise, in terms of signal-to-noise ratio (SNR), with a potential subsequent decrease in image quality. Moreover, the combination of a thin collimation and thin reconstruction thickness produces an enormous number of images that radiologists have to interpret.
To the best of our knowledge, there is no agreement in the literature about the optimal reconstruction slice thickness for the detection of endoleaks in patients who have undergone EVAR. Thus, the aim of our study was to compare the diagnostic impact of different slice thicknesses on the detection of endoleaks at MDCT in the follow-up of patients who had undergone EVAR.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Each patient provided written informed consent for MDCT angiography. This study was approved by our departmental review board and was performed in agreement with the 1990 Declaration of Helsinki principles of human rights.
Patients
A total of 50 consecutive patients (three women, 47 men; mean age, 73
years; age range, 61–86 years) who had undergone endovascular repair of
an unruptured infrarenal abdominal aortic aneurysm were enrolled from March to
December 2004.
The endovascular grafts used in these patients included 49 bifurcated grafts (Talent, Medtronic AVE, n = 12; Excluder, WL Gore, n = 21; AneuRx, Medtronic AVE, n = 6; Zenith, Cook Imaging, n = 10) and one aortomonoiliac graft with contralateral iliac occlusion and crossed femorofemoral bypass grafting (Talent, Medtronic AVE).
CT
CT scans were obtained using a 4-MDCT scanner (Somatom Volume Zoom, Siemens
Medical Solutions).
All examinations consisted of unenhanced CT scans followed by two CT acquisitions during the arterial and delayed phases after IV injection of 120 mL of iodinated nonionic contrast medium (iomeprol 300 mg I/mL [Iomeron, Bracco]) at a flow rate of 3 mL/s via an antecubital vein. The scanning delay was individualized to each patient using proprietary bolus-tracking software (CARE Bolus, Siemens). Delayed phase CT scans were obtained 60 seconds after contrast material administration.
Unenhanced CT images were obtained from the level of the diaphragm to the symphysis pubis with a 4 x 2.5 mm slice collimation, 5-mm slice width and reconstruction increment, table speed of 15 mm per rotation, and 0.5-second gantry rotation time. Contrast-enhanced CT images were obtained with a 4 x 1 mm collimation, table speed of 6 mm per rotation, 0.5-second rotation time, and 130 mAs and 120 kV. Arterial phase acquisition was performed from the suprarenal abdominal aorta to the common femoral artery, whereas delayed phase acquisition was focused on the endovascular graft.
The projection data from the arterial phase were reconstructed with a 1-mm slice thickness and 1-mm increment (set A), for a mean number of 258 ± 28 (± standard error [SE]) images per patient (range, 212–296 images); a 3-mm slice thickness and 2-mm increment (set B), for a mean number of 158 ± 24 images per patient (range, 112–193 images); and a 5-mm slice thickness and 5-mm increment (set C), for a mean number of 82 ± 13 images per patient (range, 61–102 images).
Image Analysis
Both unenhanced and arterial phase images were independently evaluated by
two experienced blinded readers with 15 and 5 years of experience in body CT,
respectively, in three separate reading sessions. In detail, each reading
session included 5-mm-thick unenhanced CT images and one set of arterial phase
images (1-, 3-, or 5-mm-thick images). The interval time among the three
reading sessions was at least 1 month; cases were presented in a different
order in each reading session. Axial images were reviewed on a dedicated
workstation (Leonardo, Siemens) and were made anonymous in terms of patient
information and reconstruction parameters. Readers were unaware of previous
imaging findings (presence or absence of endoleak) and aneurysmal sac
evolution in comparison with pretreatment or previous follow-up CT aneurysmal
sac diameter. In cases of significant disagreement between the two readers,
modifying the diagnosis, the final decision was based on a second evaluation
performed in consensus.
Delayed phase images were not included in the analysis but were used in a consensus reading that served as our standard of reference.
The readers assessed the images for the presence of an endoleak by using a 5-point confidence level scale as follows: 1, endoleak certainly absent; 2, probably absent; 3, possibly present; 4, probably present; and 5, certainly present. Before evaluating the images, the readers were informed that a confidence level of 3 or higher represented a positive diagnosis of an endoleak.
For objectivity and reproducibility of the image analyses performed in this study, standard criteria for endoleak diagnosis were provided. During the reading sessions including unenhanced plus arterial phase images, the presence of an endoleak was considered probable or certain if a high-attenuation area was present beyond the graft but within the aneurysmal sac in the arterial phase but was absent on the unenhanced phase images. The evaluation was based on visual assessment, without measurements of attenuation. The reading time for each patient in each session was recorded.
A qualitative analysis was also performed on arterial phase images to visually define differences among the three reconstruction protocols. A 4-point scale was used to score image quality as follows: 1, nondiagnostic quality (poor diagnostic information, impossible to detect or exclude vascular lesions, with beam-hardening artifacts affecting image interpretation); 2, moderate diagnostic quality (inhomogeneous enhancement in vessel lumen, evaluation of vascular lesions possible with low diagnostic confidence, with beam-hardening artifacts not affecting image interpretation); 3, good visualization (good and almost completely homogeneous enhancement in vessel lumen, evaluation of vascular lesions possible with satisfactory diagnostic confidence, without beam-hardening artifacts); and 4, excellent visualization (optimal and completely homogeneous enhancement in vessel lumen, evaluation of vascular lesions possible with high diagnostic confidence, without beam-hardening artifacts).
Image noise, defined as "graininess" of the image, was also evaluated according to a 5-point scale as follows: 1, minimum or no image noise; 2, less-than-average noise; 3, average noise in an acceptable image; 4, above-average increase of noise; and 5, unacceptable noise.
Standard of Reference
Our standard of reference for both detection and exclusion of an endoleak
was represented by evaluation of the triple-phase CT acquisition, including
unenhanced and 1-mm slice thickness arterial and delayed phase images. All CT
images were evaluated in consensus by two experienced readers with 15 and 5
years of experience in body CT, respectively, not involved in image analysis
who were aware of patient clinical history and previous CT findings. These two
readers were also invited to classify the endoleak, as proposed by White et
al.
[11–13],
as follows: type I, due to an incomplete sealing at the proximal or distal
anchor site; type II, due to a retrograde filling via aortic collateral
arteries; type III, due to a graft defect or a graft modules disconnection;
type IV, due to a graft wall porosity; and type V, or endotension, increase of
the aneurysmal sac diameter in the absence of a visible leakage. All endoleaks
detected only on delayed phase images were classified as low-flow leaks
[2].
The size of each endoleak was categorized by comparing the area of the
endoleak (AE) with the maximum cross-sectional area of the
aneurysmal sac (AAS) evaluated on axial images using an
electronic cursor (percentage size of endoleak = AE /
AAS) as follows: small, 
3%; medium, > 3% but <
10%; or large, 
10%.
Readers also assessed any change in the size of the aneurysmal sac
(increment, stability, or reduction) in comparison with previous CT
examinations by measuring the largest diameter of the aneurysm perpendicular
to the aortic axis on the axial images using an electronic cursor. A change in
size of the aneurysmal sac was recorded if it was 2 mm.
Statistical Analysis
Data were reported as mean ± SE for continuous variables, whereas
categoric and ordinal data were reported as frequencies and percentages.
Differences among the three reconstruction protocols in the image quality scores were evaluated by analysis of variance with a generalized linear model.
Interobserver agreement for the evaluation of the CT images was assessed with the intraclass correlation coefficient (ICC), which is used to evaluate rating reliability by comparing the variability of different ratings of the same subject to the total variation across all ratings and all subjects [14].
The diagnostic accuracy of each image set was also estimated by calculating the area under the receiver operating characteristic curve (Az), representing a combined measure of sensitivity and specificity. Because Az is a measure of the overall performance of a diagnostic test, differences in the performance of different tests can be evaluated by comparing their Azs [15–17].
According to DeLong et al. [18], we also evaluated the differences in terms of Az values between readers in the different sets using Mann-Whitney U statistics.
Sensitivity, specificity, and positive predictive values to detect endoleaks for each set of images were calculated and compared using the McNemar test [19].
All two-tailed p values less than 0.05 were considered statistically significant. Statistical analyses were performed using SAS software (SAS release 8.2, SAS Institute).
CT images evaluated with a score of 3 or higher and confirmed as positive for endoleak at the standard of reference were considered true-positive diagnoses, whereas CT images with a confidence level of 1 or 2 evaluated as negative for endoleak at the standard of reference were considered true-negative diagnoses. False-negative diagnoses were represented by CT images with a confidence of 1 or 2 evaluated as positive for endoleak at the standard of reference, whereas false-positive diagnoses were represented by CT images with a score of 3 or higher evaluated as negative for endoleak at the standard of reference.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Standard of Reference
Eighteen patients (18/50, 36%) were found to have endoleaks. The endoleak
was classified as type I in one patient (5.6%), due to incomplete attachment
of proximal end of the prosthesis; type II in 16 patients (88.9%), due to
regressed flow caused by the inferior mesenteric artery in two and lumbar
arteries in 14; and type III in the last patient (5.6%), caused by the
detachment of the prosthesis at various sections. Readers classified endoleaks
as low-flow type II leaks in two of the 18 patients (11.1%).
On the basis of size, endoleaks were classified as small in seven cases (38.9%), medium in eight (44.4%), and large in the last three (16.7%). All small and medium endoleaks were classified as type II, whereas one each of the three large endoleaks were type I, type II, and type III, respectively. Both of the low-flow leaks were classified as small.
An increase in the size of the aneurysmal sac associated with the endoleak was observed in all type I and type III endoleaks and in three type II endoleaks. The remaining 13 type II endoleaks, including the two low-flow leaks, were associated with an unchanged (5/13) or decreased (8/13) aneurysmal sac. Proximal type I and type III endoleaks were successfully treated with a cuff implanted in the proximal end and in the area of partial disconnection of the prosthesis, respectively, with no persistence or recurrence seen on the other CT follow-up examinations. None of the type II endoleaks required treatment; in detail, the three type II endoleaks associated with an increase in size of the aneurysmal sac had spontaneously disappeared at 6-month follow-up CT.
In the 32 patients without an endoleak, as defined by our standard of reference, the diagnosis was confirmed by subsequent follow-up CT scans (at least 1-year follow-up) that showed either a decrease in the size or stability of the aneurysmal sac in the absence of complications.
Image Analysis for Endoleak Detection
The ICC showed excellent interobserver agreement in all reading sessions
for endoleak detection (Table
1); furthermore, no statistical differences were found between
Az values obtained for each reader in all reading sessions
based on the 5-point confidence level scale for endoleak detection. For set A,
the Az value for readers 1 and 2 was 0.97 and 0.97,
respectively, and the ICC was 0.91–1.00 and 0.90–1.00,
respectively (reader 1 vs 2, 
2 = 1.05, p > 0.05).
For set B, the Az value for readers 1 and 2 was 0.95 and
0.94, respectively, and the ICC was 0.89–1.00 and 0.86–1.00
(reader 1 vs 2, 2= 0.14, p > 0.05). For set C, the
Az value for readers 1 and 2 was 0.71 and 0.67, and the
ICC was 0.55–0.87 and 0.50–0.83 (reader 1 vs 2, 2=
2.03, p > 0.05).
|
No significant disagreement was also found between the two readers in terms of negative versus positive diagnosis of endoleak, so there was no need for a second evaluation performed in consensus. On the basis of these statistical results, the data for the two readers were pooled.
The total number of endoleaks detected by the readers was 17 (17/18, 94.4%) in the thin-section group (set A), 16 (16/18, 88.9%) in the overlapped image group (set B), and 11 (11/18, 61.1%) in the thick-section group (set C) (Table 2). Of the two small low-flow endoleaks, only one was correctly detected on set A images (diagnosed as possibly present, score of 3), whereas neither could be detected on set B or set C images (false-negative cases). Set C images did not allow the detection of all seven small endoleaks (Figs. 1A, 1B, and 1C). All medium and large endoleaks were correctly diagnosed on all sets of images (Figs. 2A, 2B, and 2C). This means that we collected one, two, and seven false-negative cases during review of set A, B, and C images, respectively.
|
|
|
|
|
|
|
|
The Az values for each image set are reported in Figure 3. The Az values of set A (0.97, ICC = 0.91–1.00) and set B (0.95, ICC = 0.89–1.00) were significantly higher than that obtained with set C (0.71, ICC = 0.55–0.87) (p < 0.01), whereas no statistically significant differences were found between Az values of set A and set B. Sensitivity, specificity, and positive predictive values for endoleak detection in the three sets of images are reported in Table 3. No statistically significant differences (p > 0.05) were found when comparing set A and set B; both provided sensitivity, specificity, and positive predictive values that were significantly higher (p < 0.001) than those obtained with set C (5-mm slice thickness) (Table 3).
|
The mean reading time was 6.3 minutes for set A, 4.1 minutes for set B, and 3.2 minutes for set C, with statistically significant differences between set A and sets B and C (p = 0.034), whereas no statistically significant differences were found between sets B and C (p > 0.05).
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
However, MDCT allows original axial images to be reconstructed with different slice thicknesses and consequently offers the possibility of different acquisition protocols.
In the past years, this MDCT feature has been discussed in other fields of interest, such as in the detection of liver lesions, renal masses, or pulmonary nodules [23–29], with alternative results; however, to the best of our knowledge, there is no published work regarding the optimal reconstructed slice thickness for the detection of endoleaks in patients who have undergone EVAR.
On the basis of this background, to identify the slice thickness that may be a good compromise of image noise, partial volume effect, data explosion, and endoleak detection rate in the evaluation of patients who have undergone EVAR, we focused our attention on three reconstruction protocols: 1-mm-thick sections at 1-mm increments (set A: thin-section protocol), 3-mm-thick sections at 2-mm increments (sets B: overlapped image protocol), and 5-mm-thick sections at 5-mm increments (set C: thick-section protocol).
As a consequence of its thinner reconstructed slices, set A images had significantly higher noise than sets B and C images.
Although the use of a low-spatial-frequency reconstruction algorithm could reduce image noise even for thin-section images, this was not the aim of our study. Moreover, we found that the greater degree of anatomic detail obtained with the thin-section protocol outweighed the increase in image noise obtained, revealing similar overall image quality scores for the three reconstruction protocols.
On the other hand, when considering the endoleak detection rate, no statistically significant differences were found between the thin-section protocol (set A) and the overlapped image protocol (set B) in terms of sensitivity, specificity, and positive predictive values. In detail, when compared with set A, set B was not able to obtain a correct diagnosis in only two patients (one false-negative and one false-positive). The false-negative case was a low-flow type II endoleak not associated with an increase of the aneurysmal sac, whereas the false-positive case was an erroneous diagnosis of a small type II endoleak associated with a stable or decreased aneurysmal sac. If diagnosis were based on interpretation of the set B images, neither management nor outcome of these two patients would have changed.
However, when considering the number of images obtained and the reading time required, set B data were statistically lower than set A data (mean ± SE, 158 ± 24 vs 258 ± 28 images; mean, 4.1 vs 6.3 minutes, respectively). Furthermore, the lower number of images obtained with set B may also help to lower the storage capacity of data sets stored on archiving systems.
On the other hand, although set C (5 mm) was associated with a lower number of images (82 ± 13 images) than sets A and B and a consequent significant reduction of reading time (mean for set C, 3.2 minutes), it did not allow all seven small type II endoleaks (false-negative cases) to be detected and did not allow the presence of an endoleak in 11 patients of the negative group to be to correctly excluded (false-positive cases), with sensitivity, specificity, and positive predictive values significantly lower than those obtained with both set A and set B.
In our study, we analyzed only unenhanced and arterial phase scans. Delayed phase images were not included in the analysis, first, because the purpose of our study was to analyze how to optimize the arterial phase acquisition and, second, because it is generally accepted that endoleaks are better detected during the arterial phase. Furthermore, the results of some studies have shown that delayed phase imaging does not statistically increase diagnostic sensitivity in detecting endoleaks or change management or outcome of patients because it allows the detection of low-flow endoleaks that generally were not associated with any interrelated complications and that did not require treatment [30–35]. As a matter of fact, in our study, during the standard-of-reference reading session, only two low-flow leaks were detected with the combination of delayed phase and arterial phase images; moreover, one of these two endoleaks was also suspected on 1-mm-thick arterial phase images (diagnosed as possibly present, score of 3).
The main limitation of our study is the relatively small number of patients examined; further investigations with large series of patients are needed to confirm our findings.
A potential limitation of our study could be the lack of a proper gold standard. However, as also reported in literature, triple-phase CT acquisition, including unenhanced and 1-mm slice thickness arterial and delayed phase images, with the addition of clinical data (change in size of aneurysmal sac in comparison with previous CT examinations) seems to be the best gold standard in the follow-up of patients who have undergone EVAR.
Another potential limitation could be that this study was performed on a 4-MDCT unit when the actual tendency is to use higher-detector CT. However, until now, 4-MDCT has been the standard in many departments. Furthermore, by considering that the problem related to the number of images is more critical with the constant progress and updates of CT technology using submillimeter reconstructions, a bigger reconstruction set should be a better compromise also with the new CT machines.
In conclusion, our results indicate that the sensitivity of 1- and 3-mm-thick slices for the detection of endoleaks after EVAR at MDCT was significantly higher than that of 5-mm-thick slices. Moreover, no significant advantages in reducing slice thickness to less than 3 mm were found.
On the basis of these results, we suggest that MDCT images obtained for follow-up after EVAR be reconstructed using a 3-mm slice thickness and 2-mm increment. A second optional set of images should be reconstructed at a thinner thickness only in selected patients or in areas of concern when the standard reconstruction protocol does not allow a definite diagnosis.
Acknowledgments
We gratefully acknowledge the statistical help and advice of Angelo Di
Iorio, Department of Medicine and Sciences of Aging, Postgraduate School of
Physical Medicine and Rehabilitation, University "G. D'Annunzio,"
Chieti, Italy.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
3 cm) renal masses with MDCT: benefits of thin overlapping
reconstructions. AJR 2004;183
: 223-228 1.5 cm) liver metastases:
is thinner collimation better? Radiology2002; 225:137
-142
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  P. T. Johnson, K. M. Horton, S. Kawamoto, J. Eng, M. J. Bean, S. J. Shan, and E. K. Fishman MDCT for Suspected Appendicitis: Effect of Reconstruction Section Thickness on Diagnostic Accuracy, Rate of Appendiceal Visualization, and Reader Confidence Using Axial Images Am. J. Roentgenol., April 1, 2009; 192(4): 893 - 901. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. H. Lin, C. Bechara, P. Kougias, T. T. Huynh, S. A. LeMaire, and J. S. Coselli Assessment of Aortic Pathology and Peripheral Arterial Disease Using Multidetector Computed Tomographic Angiography Vascular and Endovascular Surgery, January 1, 2009; 42(6): 583 - 598. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
