| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1
Department of Vascular Surgery, University Medical Center Utrecht,
Heidelberglaan 100, 3584 CX Utrecht, The Netherlands.
2
Department of Radiology, University Medical Center Utrecht, 3584 CX Utrecht,
The Netherlands.
Received October 1, 1999;
accepted after revision April 3, 2000.
Address correspondence to J. D. Blankensteijn.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Thirty CT angiographic data sets were randomly chosen from 91 sets, including preoperative, postoperative, and follow-up CT images. All images were obtained according to a standardized acquisition protocol. On a workstation, three parameters were measured: maximum aneurysm diameter, maximum aneurysm cross-sectional area, and aneurysm volume. All data sets were measured twice by two investigators in a random order. The difference of each pair of measurements was plotted against the mean value. The mean difference and its standard deviation were calculated with a repeatability coefficient.
RESULTS. The intraobserver repeatability coefficient for observer 1 was 3.8 mm for diameter, 201.7 mm2 for cross-sectional area, and 5.6 mL for volume. For observer 2, these figures were 3.0 mm, 219.0 mm2, and 8.1 mL, respectively. The interobserver repeatability coefficients were 3.9 mm, 236.2 mm2, and 10.3 mL.
CONCLUSION. Determination of the repeatability coefficient of aneurysm size measurements obtained after endovascular abdominal aortic aneurysm repair provides a good description of precision.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Helical CT angiography plays an important role in revealing aneurysm size; however, little is known about the variability of size measurements obtained after endovascular abdominal aortic aneurysm repair.
We evaluated the variability of CT angiographic measurements obtained after the endovascular treatment of an abdominal aortic aneurysm.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The study population contained 30 preoperative and 61 postoperative CT angiographic data sets. From these 91 data sets, 30 were randomly allocated to be included in the analysis, resulting in an aselect sample not only among patients but also among individual studies. We analyzed nine preoperative and 21 postoperative scans. All scans were obtained according to a standardized acquisition protocol using a EV-AP CT scanner (Philips Medical Systems, Best, The Netherlands). IV contrast material (140 mL) was administered at a rate of 3 mL/sec. Scanning started 30 sec after the beginning of contrast material injection at the level of the 12th thoracic vertebra, which is the presumed level of the celiac trunk. Fifty to 70 rotations of 1 sec each were obtained. The table speed was set at 5 mm/sec, and the collimation was set at 5 mm (pitch of 1), resulting in a length of the scanned volume of 25-35 cm.
The acquired set of raw CT angiographic data was reconstructed into 123-173 images with a slice thickness of 5 mm and a slice overlap of 3 mm. These images were transferred to an Easy Vision workstation (Philips Medical Systems).
The parameters assessed were maximum aneurysm diameter, maximum aneurysm cross-sectional area, and aneurysm volume. Two investigators independently measured the 30 data sets twice, resulting in two series of paired measurements. Repeated measurements were obtained in a random order without knowledge of the previous values.
Maximum Diameter and Cross-Sectional Area
A central lumen line was drawn manually through the lumen of the aorta.
This was done by positioning points in the center of the lumen in the axial,
sagittal, and coronal planes
[3] (Fig.
1A,1B,1C,1D,1E).
|
|
|
|
|
A complete set of multiplanar reformats was reconstructed by the computer perpendicular to this central lumen line. The maximum aneurysm diameter was measured on this reformatted set of images. At the same level, a contour was drawn manually along the border of the thrombus. This contour was used to calculate the cross-sectional area of the aneurysm at the level of maximum diameter. Repeated measurements were obtained using a new central lumen line.
Volume
The aortic lumen was segmented using the threshold technique. The threshold
was set to highlight only the contrast-enhanced lumen, stent (if present), and
vertebrae [4]. The mean
threshold was 100 H. By positioning a seeding point into the lumen, it was
possible to separate the lumen and stent from the rest of the threshold voxels
(Fig. 1D). In areas of close
contact, the aortic lumen and vertebrae were separated by hand-drawn cutoff
lines. The volume of the contrast-enhanced lumen, including the stent, was
reconstructed from the individually segmented axial slices, starting at the
level immediately below the renal arteries and ending at the level of the
native aortic bifurcation.
The segmentation of the thrombus was fully manual. A contour was drawn along the border of the thrombus on each slice (Fig. 1E). Segmentation of the thrombus started just below the level of the renal arteries and ended at the level of the native bifurcation. No thrombus was found below this point. The procedure resulted in obtaining the volume of the thrombus plus the contrast-filled lumen in the thrombus. The actual thrombus volume was obtained by subtracting the volume of the contrast-filled lumen in the thrombus from the volume of the segmented thrombus (Fig. 1F). This algorithm was necessary because the lumen and thrombus do not end at the same level and are segmented using different techniques.
Total aneurysm volume was defined as the complete volume of the lumen and thrombus between the renal arteries and the native aortic bifurcation. This was the volume used when performing statistical comparison.
Analysis
The difference of means analysis described by Bland and Altman
[5] was applied to our results.
This method was used because it is simple to perform and to interpret and can
replace methods like correlation coefficient calculation or regression
analysis, which are misleading methods to use when calculating variability.
The difference of each pair of measurements was plotted against the means. By
analyzing the differences between the paired measurements, the only source of
variability was the measurement error, which was likely to follow a normal
distribution. Also, we calculated the standard deviation of the mean
difference. The repeatability coefficient was defined as 1.96 times the
standard deviation.
The repeatability coefficient was calculated for absolute figures and percentages in which the percentage is the difference between the paired measurements.
Additionally, the Student's t test was used to compare the means of the different series. Correlation between the magnitude of the mean and that of the difference was analyzed using Spearman's rank correlation.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
The differences were plotted against their mean for the interobserver variability of the three parameters (Figs. 2,3,4). Figures 2,3,4 show the mean difference and the range in which 95% of the differences lie.
|
|
|
Student's t test comparison of the means of the different series for maximum diameter, cross-sectional area, and volume is shown in Tables 2,3,4. The tables show that the paired measurements for one observer do not significantly differ among the three parameters, whereas the paired measurements between the observers reveal that observer 2 tends to measure significantly larger values than observer 1.
|
|
|
No correlation was found between the magnitude of the measurement and the measurement error, except in the intraobserver bias for volume measurements for observer 2. Here, a correlation coefficient of -0.38 (p <0.05) was noted.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
When calculating the interobserver variability, we noted that observer 2 measured larger values than observer 1. As these CT angiographic measurements are based upon manipulations of a cursor on a computer screen, they are subject to observer interpretation. Although our measurement protocol is highly standardized, apparently some bias cannot be avoided. For instance, defining the border of thrombus on a calcified wall is in our experience an important source of differences in interpretation.
Sources of variability other than the observer may influence measurements during follow-up examinations; however, these sources are beyond the scope of this article. These sources include patient factors, including blood pressure, obesity, and the abilities to lie still and follow breath-holding instructions. CT angiographic factors, such as the type of CT and acquisition, may also influence measurements. Finally, the way the technician performs CT angiography and directs the patients through the procedure may affect measurements.
To assign statistical significance to the changes measured, it is necessary to define the measurement error in size changes. Calculation of the repeatability coefficient according to Bland and Altman [5] defines a range within which 95% of all size changes lie that are caused by measurement error. If the size change exceeds the repeatability coefficient, the size change can be considered a significant change with a confidence interval of 95%.
The fact that the intraobserver variability is smaller than interobserver variability implies that measurements, and especially trends in measurements during follow-up, will be more reliable when performed by one observer. Therefore, the more observers involved, the more difficult it will be to interpret the impact of measurements on clinical decision-making.
Because we use total volume for follow-up examinations performed after endovascular aneurysm repair, we have defined a 10-mL volume difference between two CT scans as being clinically significant on the basis of inter- and intraobserver variability. This definition has led to clinical decision-making in terms of endovascular reintervention or conversion to open surgery. For maximum diameter measurement, 4 mm would be clinically significant, and a cross-sectional area change of 250 mm2 is the proper cutoff point.
In conclusion, by determining the repeatability coefficient of three different parameters of aneurysm size after endovascular abdominal aortic aneurysm repair, it is possible to define a cutoff value, above which size changes become statistically significant and clinically relevant.
Measurements of total volume show the smallest intraobserver variability compared with maximum diameter and cross-sectional area; therefore, these measurements are the most precise parameters to use during follow-up examinations performed after endovascular abdominal aortic aneurysm repair.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  S. W. Stavropoulos and S. R. Charagundla Imaging Techniques for Detection and Management of Endoleaks after Endovascular Aortic Aneurysm Repair Radiology, June 1, 2007; 243(3): 641 - 655. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Schoder, F. Cartes-Zumelzu, M. Grabenwoger, M. Cejna, M. Funovics, C. G. Krenn, D. Hutschala, F. Wolf, S. Thurnher, G. Kretschmer, et al. Elective Endovascular Stent-Graft Repair of Atherosclerotic Thoracic Aortic Aneurysms:Clinical Results and Midterm Follow-Up Am. J. Roentgenol., March 1, 2003; 180(3): 709 - 715. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
