Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

DOI: 10.1148/radiol.2433060080
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Ferencik, M.
Right arrow Articles by Achenbach, S.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Ferencik, M.
Right arrow Articles by Achenbach, S.
(Radiology 2007;243:696-702.)
© RSNA, 2007

Cardiac Imaging

Diagnostic Accuracy of Image Postprocessing Methods for the Detection of Coronary Artery Stenoses by Using Multidetector CT1

Maros Ferencik, MD, PhD, Dieter Ropers, MD, Suhny Abbara, MD, Ricardo C. Cury, MD, Udo Hoffmann, MD, Koen Nieman, MD, PhD, Thomas J. Brady, MD, Fabian Moselewski, MD, Werner G. Daniel, MD, and Stephan Achenbach, MD

1 From the Department of Radiology, Massachusetts General Hospital and Harvard Medical School, Boston, Mass (M.F., S.A., R.C.C., U.H., K.N., T.J.B., F.M., S.A.); and Department of Internal Medicine II, University of Erlangen, Ulmenweg 18, Erlangen 91054, Germany (D.R., W.G.D., S.A.). Received January 15, 2006; revision requested March 14; revision received May 24; accepted June 8; final version accepted October 4. Supported in part by Staedtler-Foundation, Nuremberg, Germany. M.F. supported in part by National Institutes of Health grant 1 T32 HL076136-02. K.N. supported by Interuniversity Cardiology Institute of the Netherlands (Utrecht, the Netherlands). F.M. supported by Daniela und Juergen Westphal-Stiftung, Flensburg, Germany. Address correspondence to S.A. (e-mail: stephan.achenbach{at}med2.med.uni-erlangen.de).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCE IN KNOWLEDGE
 References
 
Purpose: To retrospectively evaluate the diagnostic accuracy of multidetector computed tomography (CT) coronary angiography for detection of hemodynamically significant (≥50%) stenoses by using various image postprocessing methods, with conventional coronary angiography as the reference standard.

Materials and Methods: The analysis used data from previous studies, use of which had been approved by the Institutional Review Board. Sixteen-section multidetector CT data sets for 40 patients (30 men, 10 women; mean age 56 years ± 8; mean heart rate, 61 beats per minute ± 6) were evaluated. Six independent investigators evaluated the data sets for the presence of stenoses with diameter reduction of 50% or more, by using either exclusively transverse images, free oblique multiplanar reconstructions (MPRs), free oblique maximum intensity projections (MIPs, 5 mm thick), prerendered curved MPRs, prerendered curved MIPs, or prerendered three-dimensional volume rendered reconstructions (VRTs). Evaluation results were compared with conventional coronary angiography for each artery in a blinded fashion ({chi}2 test).

Results: Overall, 35 coronary artery stenoses were present. Percentage of evaluable arteries and accuracy for detecting stenosis (percentages of accurately classified arteries were, respectively, 99% and 88% for transverse, 99% and 91% for oblique MPR, 94% and 86% for oblique MIP, 94% and 83% for curved MIP, 93% and 81% for curved MPR, and 91% and 73% for VRT). Accuracy was significantly higher for oblique MPR than for curved MPR (P = .01), curved MIP (P = .03), and VRT (P < .001).

Conclusion: The evaluation of multidetector CT coronary angiography with interactive image display methods, especially interactive oblique MPRs, permits higher diagnostic accuracy than evaluation of prerendered images (curved MPR, curved MIP, or VRT images).

© RSNA, 2007


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCE IN KNOWLEDGE
 References
 
Contrast material–enhanced multidetector computed tomography (CT) is a reliable method for the detection and exclusion of hemodynamically significant coronary stenosis (1–12). Multidetector CT data sets for coronary artery visualization typically consist of 200–400 thin (0.5 mm–1.0 mm) transverse cross-section images. Diagnostic evaluation of the data sets is performed off line and includes review of the originally reconstructed transverse images as well as various postprocessing techniques. Some of these require the reader to interactively manipulate the data on workstations (eg, oblique multiplanar reconstructions [MPRs] or oblique maximum intensity projections [MIPs]), others can be rendered by a third party (eg, an image postprocessing laboratory) and reviewed without further data manipulation (eg, curved MPR or three-dimensional [3D] reconstructions).

The use of images that are prerendered and provide only a limited number of views could obviate the use of workstations for reading and shorten the time necessary for diagnostic evaluation of multidetector CT coronary angiograms. This could either be an interesting alternative for clinical practice, or it could reduce diagnostic accuracy of the technique. Therefore, the purpose of our study was to retrospectively evaluate the diagnostic accuracy of multidetector CT coronary angiography for detection of hemodynamically significant (≥50%) stenoses by using various image postprocessing methods, with conventional coronary angiography as the reference standard.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 ADVANCE IN KNOWLEDGE
 References
 
Forty multidetector CT data sets of patients enrolled in an original research study comparing multidetector CT coronary angiography and conventional coronary angiography were evaluated (Fig 1). The data sets were obtained from consecutive patients (30 men, 10 women; mean age, 56 years ± 8) with a heart rate of less than 65 beats per minute during the scan (mean heart rate, 61 beats per minute ± 6) and absence of major artifacts caused by motion or inconsistent breath holding in the data set. The original study was approved by the Institutional Review Board of the University of Erlangen (Erlangen, Germany), and written informed consent was obtained from the study participants. For the purpose of the retrospective analysis in our current study, the Institutional Review Boards of the University of Erlangen and Massachusetts General Hospital (Boston, Mass) approved the secondary use of data with informed consent waiver. Those parts of data handling and analysis that were performed in the United States were carried out in compliance with Health Insurance Portability and Accountability Act regulations.


Figure 1
View larger version (16K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1: Flow diagram describes number of subjects considered for the study and number of subjects excluded due to high heart rate or motion artifacts. bpm = Beats per minute.

 
Multidetector CT Data Acquisition
Patient preparation included the oral administration of a ß-blocker (atenolol 100 mg) 1 hour prior to the scan for all patients with a heart rate of more than 60 beats per minute. Nitroglycerin spray (0.4 mg, glycerol trinitrate) was administered sublingually immediately before the scan.

Multidetector CT was performed with a 16-section scanner (Sensation 16; Siemens Medical Solutions, Frochheim, Germany). The scan protocol included the acquisition of a low-energy tomogram and bolus timing scan. Multidetector CT coronary angiography was performed with acquisitions of 16 sections at 0.75-mm collimation, 120 kVp, 550 mAs, and 375-msec gantry rotation time. Multidetector CT images were acquired during injection of 80–90 mL of iopromide (370 mg of iodine per milliliter) followed by 50 mL of saline, both at a flow rate of 5 mL/sec. Retrospective electrocardiographically gated half-scan reconstruction was applied to create a stack of contiguous transverse cross-section images with a section thickness of 1.0 mm and an increment of 0.5 mm. Images were reconstructed by using a medium smooth reconstruction kernel (B35f). The initial data set was reconstructed with the reconstruction window started at 65% of the cardiac cycle. If motion artifacts were present in any coronary artery, image reconstruction was repeated with the reconstruction window offset by 5% toward the beginning or end of the cardiac cycle, and multiple reconstructions were obtained until all arteries were depicted free of motion artifact or until reconstructions in 5% intervals throughout the cardiac cycle had been obtained.

Multidetector CT Image Evaluation
Transverse images were transferred to an off-line workstation (Leonardo; Siemens Medical Solutions). One observer (F.M., with 3 years experience in cardiac CT imaging) who was blinded to the results of conventional coronary angiography prepared all multidetector CT data sets for further evaluation by rendering curved MPRs, curved MIPs, and segmented datasets for the 3D volume rendering technique (VRT) reconstructions. Curved MPRs were rendered with 1.0-mm section thickness, displaying the course of the left main and left anterior descending arteries, left main and left circumflex arteries (the obtuse marginal branch was used if it was larger than the left circumflex coronary artery), and the right coronary artery, each in two orientations (with transverse and coronal images as reference) (13). Curved MIPs were rendered by using a 3-mm-thick MIP display. Data sets for 3D display were prepared by manually segmenting the transverse images so that all structures overlapping the coronary arteries (chest wall, pulmonary trunk, atrial appendages, inferior vena cava, liver) were removed. The data sets were then stored for further analysis.

Subsequently, six investigators experienced in cardiac CT imaging (M.F., with 4 years experience, D.R., with 9 years experience, S.A., with 5 years experience, R.C.C., with 4 years experience, U.H., with 7 years experience, and K.N., with 7 years experience) independently evaluated the multidetector CT data sets on the same workstation by exclusively using either transverse images, free oblique MPRs (1.0 mm thick in any desired plane), free oblique MIPs (5.0 mm thick in any desired plane), the prerendered curved MPRs, curved MIPs, or prerendered 3D VRT images (Fig 2). Transverse, oblique MPR, oblique MIP, curved MPR, and curved MIP images were initially displayed with a default window setting (level, 250 HU; window, 750 HU). The window and level of the evaluated images could then be adjusted by the observer. VRT images could be manipulated to show the heart and coronary arteries from any desired angle. They were initially displayed with preset threshold values and color schemes that could be modified by the observer as desired.


Figure 2A
View larger version (109K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2a: Image postprocessing methods used in analysis of right coronary artery. (a) Transverse images. Operators could interactively move up and down stack of transverse images, which were reconstructed with 1.0-mm section thickness and 0.5-mm increment. Shown here is origin of the right coronary artery (arrow). (b) Oblique MPR. Operators could interactively and freely manipulate position and orientation of imaging plane. Right coronary artery image section was 1.0 mm thick (arrows). (c) Oblique MIP. Operators could interactively and freely manipulate position and orientation of image, consisting of a 5-mm section, displaying attenuation for each image pixel. This form of reconstruction allows depiction of longer segments of a given coronary artery. Section position and orientation of c are identical to those of b (arrow = right coronary artery). (d, e) Curved MPR. By using transverse and coronary images as reference, the artery course was interactively traced by an independent operator. A 1-mm MPR image was rendered, the plane of which followed the outlined trace. Images d and e show the resulting display of the right coronary artery (arrow). (f, g) Curved MIP. As with d and e, the course of the right coronary artery is traced. The image is then displayed as an MIP (3 mm thick). Images f and g show the resulting display of the right coronary artery (arrow). (h) Three-dimensional reconstruction. Image data set was segmented by an independent operator and subsequently displayed as 3D surface-weighted VRT image. The images could be viewed from any desired angle to assess the course of all arteries (arrows = right coronary artery). (i) Corresponding conventional coronary angiogram (arrow = right coronary artery) of the same patient.

 

Figure 2B
View larger version (118K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2b: Image postprocessing methods used in analysis of right coronary artery. (a) Transverse images. Operators could interactively move up and down stack of transverse images, which were reconstructed with 1.0-mm section thickness and 0.5-mm increment. Shown here is origin of the right coronary artery (arrow). (b) Oblique MPR. Operators could interactively and freely manipulate position and orientation of imaging plane. Right coronary artery image section was 1.0 mm thick (arrows). (c) Oblique MIP. Operators could interactively and freely manipulate position and orientation of image, consisting of a 5-mm section, displaying attenuation for each image pixel. This form of reconstruction allows depiction of longer segments of a given coronary artery. Section position and orientation of c are identical to those of b (arrow = right coronary artery). (d, e) Curved MPR. By using transverse and coronary images as reference, the artery course was interactively traced by an independent operator. A 1-mm MPR image was rendered, the plane of which followed the outlined trace. Images d and e show the resulting display of the right coronary artery (arrow). (f, g) Curved MIP. As with d and e, the course of the right coronary artery is traced. The image is then displayed as an MIP (3 mm thick). Images f and g show the resulting display of the right coronary artery (arrow). (h) Three-dimensional reconstruction. Image data set was segmented by an independent operator and subsequently displayed as 3D surface-weighted VRT image. The images could be viewed from any desired angle to assess the course of all arteries (arrows = right coronary artery). (i) Corresponding conventional coronary angiogram (arrow = right coronary artery) of the same patient.

 

Figure 2C
View larger version (121K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2c: Image postprocessing methods used in analysis of right coronary artery. (a) Transverse images. Operators could interactively move up and down stack of transverse images, which were reconstructed with 1.0-mm section thickness and 0.5-mm increment. Shown here is origin of the right coronary artery (arrow). (b) Oblique MPR. Operators could interactively and freely manipulate position and orientation of imaging plane. Right coronary artery image section was 1.0 mm thick (arrows). (c) Oblique MIP. Operators could interactively and freely manipulate position and orientation of image, consisting of a 5-mm section, displaying attenuation for each image pixel. This form of reconstruction allows depiction of longer segments of a given coronary artery. Section position and orientation of c are identical to those of b (arrow = right coronary artery). (d, e) Curved MPR. By using transverse and coronary images as reference, the artery course was interactively traced by an independent operator. A 1-mm MPR image was rendered, the plane of which followed the outlined trace. Images d and e show the resulting display of the right coronary artery (arrow). (f, g) Curved MIP. As with d and e, the course of the right coronary artery is traced. The image is then displayed as an MIP (3 mm thick). Images f and g show the resulting display of the right coronary artery (arrow). (h) Three-dimensional reconstruction. Image data set was segmented by an independent operator and subsequently displayed as 3D surface-weighted VRT image. The images could be viewed from any desired angle to assess the course of all arteries (arrows = right coronary artery). (i) Corresponding conventional coronary angiogram (arrow = right coronary artery) of the same patient.

 

Figure 2D
View larger version (114K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2d: Image postprocessing methods used in analysis of right coronary artery. (a) Transverse images. Operators could interactively move up and down stack of transverse images, which were reconstructed with 1.0-mm section thickness and 0.5-mm increment. Shown here is origin of the right coronary artery (arrow). (b) Oblique MPR. Operators could interactively and freely manipulate position and orientation of imaging plane. Right coronary artery image section was 1.0 mm thick (arrows). (c) Oblique MIP. Operators could interactively and freely manipulate position and orientation of image, consisting of a 5-mm section, displaying attenuation for each image pixel. This form of reconstruction allows depiction of longer segments of a given coronary artery. Section position and orientation of c are identical to those of b (arrow = right coronary artery). (d, e) Curved MPR. By using transverse and coronary images as reference, the artery course was interactively traced by an independent operator. A 1-mm MPR image was rendered, the plane of which followed the outlined trace. Images d and e show the resulting display of the right coronary artery (arrow). (f, g) Curved MIP. As with d and e, the course of the right coronary artery is traced. The image is then displayed as an MIP (3 mm thick). Images f and g show the resulting display of the right coronary artery (arrow). (h) Three-dimensional reconstruction. Image data set was segmented by an independent operator and subsequently displayed as 3D surface-weighted VRT image. The images could be viewed from any desired angle to assess the course of all arteries (arrows = right coronary artery). (i) Corresponding conventional coronary angiogram (arrow = right coronary artery) of the same patient.

 

Figure 2E
View larger version (116K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2e: Image postprocessing methods used in analysis of right coronary artery. (a) Transverse images. Operators could interactively move up and down stack of transverse images, which were reconstructed with 1.0-mm section thickness and 0.5-mm increment. Shown here is origin of the right coronary artery (arrow). (b) Oblique MPR. Operators could interactively and freely manipulate position and orientation of imaging plane. Right coronary artery image section was 1.0 mm thick (arrows). (c) Oblique MIP. Operators could interactively and freely manipulate position and orientation of image, consisting of a 5-mm section, displaying attenuation for each image pixel. This form of reconstruction allows depiction of longer segments of a given coronary artery. Section position and orientation of c are identical to those of b (arrow = right coronary artery). (d, e) Curved MPR. By using transverse and coronary images as reference, the artery course was interactively traced by an independent operator. A 1-mm MPR image was rendered, the plane of which followed the outlined trace. Images d and e show the resulting display of the right coronary artery (arrow). (f, g) Curved MIP. As with d and e, the course of the right coronary artery is traced. The image is then displayed as an MIP (3 mm thick). Images f and g show the resulting display of the right coronary artery (arrow). (h) Three-dimensional reconstruction. Image data set was segmented by an independent operator and subsequently displayed as 3D surface-weighted VRT image. The images could be viewed from any desired angle to assess the course of all arteries (arrows = right coronary artery). (i) Corresponding conventional coronary angiogram (arrow = right coronary artery) of the same patient.

 

Figure 2F
View larger version (124K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2f: Image postprocessing methods used in analysis of right coronary artery. (a) Transverse images. Operators could interactively move up and down stack of transverse images, which were reconstructed with 1.0-mm section thickness and 0.5-mm increment. Shown here is origin of the right coronary artery (arrow). (b) Oblique MPR. Operators could interactively and freely manipulate position and orientation of imaging plane. Right coronary artery image section was 1.0 mm thick (arrows). (c) Oblique MIP. Operators could interactively and freely manipulate position and orientation of image, consisting of a 5-mm section, displaying attenuation for each image pixel. This form of reconstruction allows depiction of longer segments of a given coronary artery. Section position and orientation of c are identical to those of b (arrow = right coronary artery). (d, e) Curved MPR. By using transverse and coronary images as reference, the artery course was interactively traced by an independent operator. A 1-mm MPR image was rendered, the plane of which followed the outlined trace. Images d and e show the resulting display of the right coronary artery (arrow). (f, g) Curved MIP. As with d and e, the course of the right coronary artery is traced. The image is then displayed as an MIP (3 mm thick). Images f and g show the resulting display of the right coronary artery (arrow). (h) Three-dimensional reconstruction. Image data set was segmented by an independent operator and subsequently displayed as 3D surface-weighted VRT image. The images could be viewed from any desired angle to assess the course of all arteries (arrows = right coronary artery). (i) Corresponding conventional coronary angiogram (arrow = right coronary artery) of the same patient.

 

Figure 2G
View larger version (120K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2g: Image postprocessing methods used in analysis of right coronary artery. (a) Transverse images. Operators could interactively move up and down stack of transverse images, which were reconstructed with 1.0-mm section thickness and 0.5-mm increment. Shown here is origin of the right coronary artery (arrow). (b) Oblique MPR. Operators could interactively and freely manipulate position and orientation of imaging plane. Right coronary artery image section was 1.0 mm thick (arrows). (c) Oblique MIP. Operators could interactively and freely manipulate position and orientation of image, consisting of a 5-mm section, displaying attenuation for each image pixel. This form of reconstruction allows depiction of longer segments of a given coronary artery. Section position and orientation of c are identical to those of b (arrow = right coronary artery). (d, e) Curved MPR. By using transverse and coronary images as reference, the artery course was interactively traced by an independent operator. A 1-mm MPR image was rendered, the plane of which followed the outlined trace. Images d and e show the resulting display of the right coronary artery (arrow). (f, g) Curved MIP. As with d and e, the course of the right coronary artery is traced. The image is then displayed as an MIP (3 mm thick). Images f and g show the resulting display of the right coronary artery (arrow). (h) Three-dimensional reconstruction. Image data set was segmented by an independent operator and subsequently displayed as 3D surface-weighted VRT image. The images could be viewed from any desired angle to assess the course of all arteries (arrows = right coronary artery). (i) Corresponding conventional coronary angiogram (arrow = right coronary artery) of the same patient.

 

Figure 2H
View larger version (104K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2h: Image postprocessing methods used in analysis of right coronary artery. (a) Transverse images. Operators could interactively move up and down stack of transverse images, which were reconstructed with 1.0-mm section thickness and 0.5-mm increment. Shown here is origin of the right coronary artery (arrow). (b) Oblique MPR. Operators could interactively and freely manipulate position and orientation of imaging plane. Right coronary artery image section was 1.0 mm thick (arrows). (c) Oblique MIP. Operators could interactively and freely manipulate position and orientation of image, consisting of a 5-mm section, displaying attenuation for each image pixel. This form of reconstruction allows depiction of longer segments of a given coronary artery. Section position and orientation of c are identical to those of b (arrow = right coronary artery). (d, e) Curved MPR. By using transverse and coronary images as reference, the artery course was interactively traced by an independent operator. A 1-mm MPR image was rendered, the plane of which followed the outlined trace. Images d and e show the resulting display of the right coronary artery (arrow). (f, g) Curved MIP. As with d and e, the course of the right coronary artery is traced. The image is then displayed as an MIP (3 mm thick). Images f and g show the resulting display of the right coronary artery (arrow). (h) Three-dimensional reconstruction. Image data set was segmented by an independent operator and subsequently displayed as 3D surface-weighted VRT image. The images could be viewed from any desired angle to assess the course of all arteries (arrows = right coronary artery). (i) Corresponding conventional coronary angiogram (arrow = right coronary artery) of the same patient.

 

Figure 2I
View larger version (177K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2i: Image postprocessing methods used in analysis of right coronary artery. (a) Transverse images. Operators could interactively move up and down stack of transverse images, which were reconstructed with 1.0-mm section thickness and 0.5-mm increment. Shown here is origin of the right coronary artery (arrow). (b) Oblique MPR. Operators could interactively and freely manipulate position and orientation of imaging plane. Right coronary artery image section was 1.0 mm thick (arrows). (c) Oblique MIP. Operators could interactively and freely manipulate position and orientation of image, consisting of a 5-mm section, displaying attenuation for each image pixel. This form of reconstruction allows depiction of longer segments of a given coronary artery. Section position and orientation of c are identical to those of b (arrow = right coronary artery). (d, e) Curved MPR. By using transverse and coronary images as reference, the artery course was interactively traced by an independent operator. A 1-mm MPR image was rendered, the plane of which followed the outlined trace. Images d and e show the resulting display of the right coronary artery (arrow). (f, g) Curved MIP. As with d and e, the course of the right coronary artery is traced. The image is then displayed as an MIP (3 mm thick). Images f and g show the resulting display of the right coronary artery (arrow). (h) Three-dimensional reconstruction. Image data set was segmented by an independent operator and subsequently displayed as 3D surface-weighted VRT image. The images could be viewed from any desired angle to assess the course of all arteries (arrows = right coronary artery). (i) Corresponding conventional coronary angiogram (arrow = right coronary artery) of the same patient.

 
Each observer evaluated the CT images of all 40 patients, using only one postprocessing method for each case to avoid an influence of prior knowledge based on having evaluated the same data set and another postprocessing method. To minimize the influence of individual experience in reading CT data sets, each observer used one of the six postprocessing methods in six or seven patients. The images were presented to the readers in random order over a period of 3 months.

All investigators were blinded to conventional coronary angiographic data. Each artery (four arteries per patient: left main, left anterior descending, left circumflex, and right coronary artery) was classified as "evaluable" or "nonevaluable." Subjective assessment was used to classify an artery as "nonevaluable" if motion artifacts, severe calcification, excessive image noise, and/or overlap of adjacent structures prevented assessment concerning the presence of stenosis. Evaluable arteries were categorized as to the presence or absence of a stenosis of 50% diameter reduction or more in its entire length. Side branches were excluded from analysis. Multidetector CT evaluation results were documented in writing and then compared with the results of conventional coronary angiography.

Conventional Coronary Angiography
In all patients, conventional coronary angiography had been performed 1 day after multidetector CT. Acquisitions were obtained after intracoronary administration of 0.2 mg of isosorbide dinitrate (Isoket; Schwarz Pharma, Moneheim, Germany) in standard conventional coronary angiography planes (minimum of four projections: four for the left coronary artery and two for the right coronary artery) during selective injection of contrast material. All conventional coronary angiograms were evaluated by a blinded independent observer (S.A., with 8 years experience in conventional coronary angiography) with the use of quantitative coronary angiography (QuantCor.QCA; Pie Medical Imaging, Maastricht, the Netherlands) and were used as the reference standard for stenosis detection. Lesions with a diameter reduction of 50% or more were considered representative of hemodynamically significant stenoses. The presence of stenosis was reported for each artery. All lesions with a reference diameter (vessel diameter in nondiseased artery immediately proximal to the lesion) of 1.5 mm or more were included in the comparison with multidetector CT findings.

Statistical Analysis
The results are expressed as mean ± standard deviation unless stated otherwise. The number of evaluable arteries and their sensitivity, specificity, positive predictive value, and negative predictive value of multidetector CT for detection of hemodynamically significant (≥50%) coronary stenosis as compared with the reference standard (quantitative coronary angiography) were calculated for each artery by using each evaluated postprocessing method. The overall accuracy of multidetector CT to enable detection of hemodynamically significant stenosis by using different postprocessing methods was calculated by counting nonevaluable arteries as "not accurately classified." As four coronary arteries have been evaluated in each patient, data on the arterial level may not be independent. We accounted for this by using generalized estimation equations methods. Ninety-five percent confidence intervals were calculated according to the approach of Zhou et al (14). We compared the accuracy of the different postprocessing methods for the detection of stenosis by using the {chi}2 test. A P value of less than .05 was considered to indicate significant difference. Statistical analysis was performed by using SPSS (version 13.0; SPSS, Chicago, Ill).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 ADVANCE IN KNOWLEDGE
 References
 
Overall, 35 stenoses of 50% diameter reduction or more were present in the patient group on conventional coronary angiography. Two stenoses were located in the left main coronary artery, 17 in the left anterior descending coronary artery, four in the left circumflex coronary artery or obtuse marginal branch, and 12 in the right coronary artery.

By using the interactive display of transverse images alone, two arteries were classified nonevaluable (one due to image noise and one due to calcification). Thirty-one stenoses were correctly detected and four stenoses were not. This corresponds to both a sensitivity of 89% and specificity of 89% in evaluable arteries and an overall accuracy of 88% (with nonevaluable arteries being classified as inaccurate). If oblique MPRs with the ability to interactively manipulate image orientation into any desired plane were used, one artery was classified as nonevaluable, with sensitivity of 91%, specificity of 92%, and accuracy of 91%. The use of oblique MIP showed a sensitivity of 87%, specificity of 93%, and accuracy of 86% (Table).


View this table:
[in this window]
[in a new window]

  		Accuracy for the Detection of Coronary Artery Stenoses for Various Methods of Image Postprocessing

 
Prerendered Reconstructions
For prerendered reconstructions, lower accuracies were observed: Curved MPRs had a sensitivity of 85%, specificity of 88%, and accuracy of 81%, with 8% nonevaluable arteries. Curved 3-mm-thick MIPs had a sensitivity of 91%, specificity of 88%, and overall accuracy of 83%. with 6% of arteries nonevaluable. Three-dimensional VRT display of the data had a sensitivity of 63%, specificity of 86%, and overall accuracy of 73%, with 10% nonevaluable arteries (Fig 3, Table).


Figure 3A
View larger version (111K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3a: Comparison of image postprocessing methods in patient with stenosis of the left circumflex coronary artery confirmed on conventional coronary angiogram. (a) Series of three transverse images (from left to right) above, at the level of, and distal to the lesion (arrow = left circumflex artery). While the contrast-enhanced lumen is clearly visible proximally and distally to the lesion, the middle image shows absence of contrast-enhanced lumen; only one calcification appears. (b) Visualization of stenosis (arrow) in an oblique MPR. (c) Visualization of stenosis (arrow) in an oblique MIP. (d) Curved MPR of left main and left circumflex coronary artery (arrow = stenosis). (e) Curved MIP (3 mm thick) of left main and left circumflex coronary artery (arrow = stenosis). (f) Three-dimensional reconstruction. Stenosis with a calcification (arrow) can clearly be seen.

 

Figure 3B
View larger version (136K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3b: Comparison of image postprocessing methods in patient with stenosis of the left circumflex coronary artery confirmed on conventional coronary angiogram. (a) Series of three transverse images (from left to right) above, at the level of, and distal to the lesion (arrow = left circumflex artery). While the contrast-enhanced lumen is clearly visible proximally and distally to the lesion, the middle image shows absence of contrast-enhanced lumen; only one calcification appears. (b) Visualization of stenosis (arrow) in an oblique MPR. (c) Visualization of stenosis (arrow) in an oblique MIP. (d) Curved MPR of left main and left circumflex coronary artery (arrow = stenosis). (e) Curved MIP (3 mm thick) of left main and left circumflex coronary artery (arrow = stenosis). (f) Three-dimensional reconstruction. Stenosis with a calcification (arrow) can clearly be seen.

 

Figure 3C
View larger version (149K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3c: Comparison of image postprocessing methods in patient with stenosis of the left circumflex coronary artery confirmed on conventional coronary angiogram. (a) Series of three transverse images (from left to right) above, at the level of, and distal to the lesion (arrow = left circumflex artery). While the contrast-enhanced lumen is clearly visible proximally and distally to the lesion, the middle image shows absence of contrast-enhanced lumen; only one calcification appears. (b) Visualization of stenosis (arrow) in an oblique MPR. (c) Visualization of stenosis (arrow) in an oblique MIP. (d) Curved MPR of left main and left circumflex coronary artery (arrow = stenosis). (e) Curved MIP (3 mm thick) of left main and left circumflex coronary artery (arrow = stenosis). (f) Three-dimensional reconstruction. Stenosis with a calcification (arrow) can clearly be seen.

 

Figure 3D
View larger version (131K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3d: Comparison of image postprocessing methods in patient with stenosis of the left circumflex coronary artery confirmed on conventional coronary angiogram. (a) Series of three transverse images (from left to right) above, at the level of, and distal to the lesion (arrow = left circumflex artery). While the contrast-enhanced lumen is clearly visible proximally and distally to the lesion, the middle image shows absence of contrast-enhanced lumen; only one calcification appears. (b) Visualization of stenosis (arrow) in an oblique MPR. (c) Visualization of stenosis (arrow) in an oblique MIP. (d) Curved MPR of left main and left circumflex coronary artery (arrow = stenosis). (e) Curved MIP (3 mm thick) of left main and left circumflex coronary artery (arrow = stenosis). (f) Three-dimensional reconstruction. Stenosis with a calcification (arrow) can clearly be seen.

 

Figure 3E
View larger version (140K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3e: Comparison of image postprocessing methods in patient with stenosis of the left circumflex coronary artery confirmed on conventional coronary angiogram. (a) Series of three transverse images (from left to right) above, at the level of, and distal to the lesion (arrow = left circumflex artery). While the contrast-enhanced lumen is clearly visible proximally and distally to the lesion, the middle image shows absence of contrast-enhanced lumen; only one calcification appears. (b) Visualization of stenosis (arrow) in an oblique MPR. (c) Visualization of stenosis (arrow) in an oblique MIP. (d) Curved MPR of left main and left circumflex coronary artery (arrow = stenosis). (e) Curved MIP (3 mm thick) of left main and left circumflex coronary artery (arrow = stenosis). (f) Three-dimensional reconstruction. Stenosis with a calcification (arrow) can clearly be seen.

 

Figure 3F
View larger version (130K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3f: Comparison of image postprocessing methods in patient with stenosis of the left circumflex coronary artery confirmed on conventional coronary angiogram. (a) Series of three transverse images (from left to right) above, at the level of, and distal to the lesion (arrow = left circumflex artery). While the contrast-enhanced lumen is clearly visible proximally and distally to the lesion, the middle image shows absence of contrast-enhanced lumen; only one calcification appears. (b) Visualization of stenosis (arrow) in an oblique MPR. (c) Visualization of stenosis (arrow) in an oblique MIP. (d) Curved MPR of left main and left circumflex coronary artery (arrow = stenosis). (e) Curved MIP (3 mm thick) of left main and left circumflex coronary artery (arrow = stenosis). (f) Three-dimensional reconstruction. Stenosis with a calcification (arrow) can clearly be seen.

 
The accuracy (number of correctly classified arteries, counting nonevaluable vessels as not accurately classified) was highest for oblique MPRs (91%, 146 of 160 arteries correctly classified). The number of accurately classified arteries was significantly higher when compared with curved MPRs (P = .02), curved MIPs (P = .03), and 3D VRT images (P < .001).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
ADVANCE IN KNOWLEDGE
 References
 
Traditionally, transverse images have been used to assess CT data sets. With the improvement in spatial resolution, particularly along the z-axis, postprocessing and rendering of images by using advanced image postprocessing methods such as MIP, curved MPR, and 3D VRT became part of the diagnostic evaluation of CT images. The original reconstructed transverse sections are the data source for all 2D and 3D postprocessing methods (15–18). Near-isotropic or isotropic spatial resolution in routine examinations permits rendering of oblique MPR and oblique MIP images with a spatial resolution similar to that of the original transverse images (19). However, information from the original data set can be lost as a consequence of processing the images for advanced postprocessing methods.

Some authors (18) have suggested the use of 3D VRT images for the initial assessment, with a subsequent rendering of MPR, curved MPR, and MIP images for diagnostic evaluation. Other authors (16) have suggested that advanced image display methods only supplement evaluation of transverse CT images.

Conflicting data have been reported regarding the diagnostic accuracy of multidetector CT coronary angiography and the use of various postprocessing methods. Herzog et al (20) evaluated the diagnostic accuracy of four-section multidetector CT for the detection of hemodynamically significant coronary stenosis by using transverse CT, virtual endoscopic, MPR, and 3D VRT images. They demonstrated that transverse images had the best sensitivity for the detection of stenosis. Interestingly, in an in vitro phantom study, Addis et al (21) showed that 3D VRT images were superior to transverse, MPR, and MIP images for the quantification of vessel stenosis. Similar results were observed in a study of magnetic resonance angiography of renal arteries (22). Three-dimensional VRT and MPR estimates of stenosis provided better results than did MIP.

Results of a study done by Mahnken et al (23) contradicted these observations when they showed that the quantification of coronary stenosis in CT coronary angiograms by using 3D VRT images was insufficient when compared with quantitative coronary angiography. Based on an ex vivo study performed in an animal model, Lu et al (24) reported that the accurate evaluation of coronary lumina in MPR images was less prone to measurement errors introduced by variations in window and level settings when compared with MIP images and 3D VRT techniques. In a study that compared transverse, MPR, 3D VRT, and virtual angioscopic images obtained with a four-section multidetector CT scanner, the highest sensitivity for stenosis detection was achieved by using transverse images (25). Transverse images were also less susceptible to motion artifacts, while other techniques were often subjected to the false-positive appearance of vascular discontinuity or nonexisting vessel wall irregularities.

In the majority of reported studies, the evaluation of multidetector CT coronary angiograms for the detection of coronary stenosis has been performed interactively on off-line workstations, by using a combination of transverse, MPR, MIP, and 3D VRT images (1–4,6,8–10,26). Some authors evaluated multidetector CT data sets initially by using MIP images or a prerendered slab of MPR images, and the findings were then confirmed by using MPR, curved MPR, or 3D VRT images (5,7,27). Of note, interactive evaluation was always an integral part of the reading process. The results of our study confirm the importance of the interactive evaluation of multidetector CT coronary angiography data sets. The readers were able to evaluate more coronary arteries by using transverse and oblique MPR images. In addition, the diagnostic accuracy of multidetector CT coronary angiography was significantly higher for oblique MPRs than for curved MPRs, curved MIPs, and 3D VRT images.

Our study had limitations. The sample size was small and no formal power analysis was performed. Analysis of multidetector CT data sets was performed for each artery. Many previous studies used segment- or patient-based analyses (1–10,26). This approach could affect the diagnostic performance of the test in general but should not affect the comparison of the various postprocessing methods. Only data sets without obvious artifacts were included in this study, and all subjects had a low heart rate. Indeed, only two arteries (1.4%) were classified as nonevaluable when transverse images were used for analysis. In clinical practice, the rate of nonevaluable arteries will be higher, which may alter the results. In our study, the rate of nonevaluable arteries was lowest for transverse images and MPRs, which illustrates that interactive manipulation of the data sets helps minimize the number of nonevaluable vessels.

Additionally, the data used for this study were acquired with 16-section CT. Currently, 64-section scanners with faster rotation times are available. When compared with 16-section CT, recent reports of the accuracy of 64-section CT for the detection of coronary stenoses indicate improved sensitivity and specificity, as well as a lower rate of nonevaluable arteries (8–10,12). Since only high-quality data were used in this study, we do not expect a systematic difference when transferring our results to 64-section CT.

The results of our investigation showed that the evaluation of multidetector CT coronary angiograms with interactive image display methods, especially with MPRs, provided higher diagnostic accuracy than did evaluation limited to the analysis of prerendered images (curved MPR, curved MIP, and 3D VRT images). Interactive evaluation of multidetector CT coronary angiography data sets on a workstation should thus be the preferred way of interpretation.


   ADVANCE IN KNOWLEDGE
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCE IN KNOWLEDGE
References
 


   FOOTNOTES
 

Abbreviations: MIP = maximum intensity projection • MPR = multiplanar reconstruction • 3D = three-dimensional • VRT = volume rendering technique reconstruction

Authors stated no financial relationship to disclose.

Author contributions: Guarantors of integrity of entire study, M.F., S.A.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, all authors; clinical studies, all authors; and manuscript editing, all authors


   References
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCE IN KNOWLEDGE
 References
 

  1. Nieman K, Cademartiri F, Lemos PA, Raaijmakers R, Pattynama PM, de Feyter PJ. Reliable noninvasive coronary angiography with fast submillimeter multislice spiral computed tomography. Circulation 2002;106:2051–2054.[Abstract/Free Full Text]
  2. Ropers D, Baum U, Pohle K, et al. Detection of coronary artery stenoses with thin-section multidetector row spiral computed tomography and multiplanar reconstruction. Circulation 2003;107:664–666.[Abstract/Free Full Text]
  3. Hoffmann U, Moselewski F, Cury RC, et al. Predictive value of 16-section multidetector spiral computed tomography to detect significant obstructive coronary artery disease in patients at high risk for coronary artery disease: patient- versus segment-based analysis. Circulation 2004;110:2638–2643.[Abstract/Free Full Text]
  4. Martuscelli E, Romagnoli A, D'Eliseo A, et al. Accuracy of thin-section computed tomography in the detection of coronary stenoses. Eur Heart J 2004;25:1043–1048.[Abstract/Free Full Text]
  5. Mollet NR, Cademartiri F, Krestin GP, et al. Improved diagnostic accuracy with 16-row multi-section computed tomography coronary angiography. J Am Coll Cardiol 2005;45:128–132.[Abstract/Free Full Text]
  6. Kuettner A, Beck T, Drosch T, et al. Diagnostic accuracy of noninvasive coronary imaging using 16-detector slice spiral computed tomography with 188 ms temporal resolution. J Am Coll Cardiol 2005;45:123–127.[Abstract/Free Full Text]
  7. Hoffmann MH, Shi H, Schmitz BL, et al. Noninvasive coronary angiography with multislice computed tomography. JAMA 2005;293:2471–2478.[Abstract/Free Full Text]
  8. Leschka S, Alkadhi H, Plass A, et al. Accuracy of MSCT coronary angiography with 64-section technology: first experience. Eur Heart J 2005;26:1482–1487.[Abstract/Free Full Text]
  9. Leber AW, Knez A, von Ziegler F, et al. Quantification of obstructive and nonobstructive coronary lesions by 64-section computed tomography: a comparative study with quantitative coronary angiography and intravascular ultrasound. J Am Coll Cardiol 2005;46:147–154.[Abstract/Free Full Text]
  10. Raff GL, Gallagher MJ, O'Neill WW, Goldstein JA. Diagnostic accuracy of noninvasive coronary angiography using 64-section spiral computed tomography. J Am Coll Cardiol 2005;46:552–557.[Abstract/Free Full Text]
  11. Achenbach S, Ropers D, Pohle FK, et al. Detection of coronary artery stenoses using multidetector CT with 16 x 0.75 collimation and 375 ms rotation. Eur Heart J 2005;26:1978–1986.[Abstract/Free Full Text]
  12. Mollet NR, Cademartiri F, van Mieghem CA, et al. High-resolution spiral computed tomography coronary angiography in patients referred for diagnostic conventional coronary angiography. Circulation 2005;112:2318–2323.[Abstract/Free Full Text]
  13. Achenbach S, Moshage W, Ropers D, Bachmann K. Curved multiplanar reconstructions for the evaluation of contrast-enhanced electron beam CT of the coronary arteries. AJR Am J Roentgenol 1998;170:895–899.[Abstract/Free Full Text]
  14. Zhou XH, McClish DK, Obuchowski NA. Statistical methods in diagnostic medicine. New York, NY: Wiley, 2002.
  15. Calhoun PS, Kuszyk BS, Heath DG, Carley JC, Fishman EK. Three-dimensional volume rendering of spiral CT data: theory and method. RadioGraphics 1999;19:745–764.[Abstract/Free Full Text]
  16. Pannu HK, Flohr TG, Corl FM, Fishman EK. Current concepts in multidetector row CT evaluation of the coronary arteries: principles, techniques, and anatomy. RadioGraphics 2003;23(Spec Issue):S111–S125.[Abstract/Free Full Text]
  17. van Ooijen PM, Ho KY, Dorgelo J, Oudkerk M. Coronary artery imaging with multidetector CT: visualization issues. RadioGraphics 2003;23:e16. http://radiographics.rsnajnls.org/cgi/content/full/23/6/E16. Accessed November 23, 2005.
  18. Desjardins B, Kazerooni EA. ECG-gated cardiac CT. AJR Am J Roentgenol 2004;182:993–1010.[Free Full Text]
  19. Flohr TG, Schaller S, Stierstorfer K, Bruder H, Ohnesorge BM, Schoepf UJ. Multi–detector row CT systems and image-reconstruction techniques. Radiology 2005;235:756–773.[Abstract/Free Full Text]
  20. Herzog C, Ay M, Engelmann K, et al. Visualization techniques in multislice CT-coronary angiography of the heart: correlations of axial, multiplanar, three-dimensional and virtual endoscopic imaging with the invasive diagnosis [in German]. Rofo 2001;173:341–349.[Medline]
  21. Addis KA, Hopper KD, Iyriboz TA, et al. CT angiography: in vitro comparison of five reconstruction methods. AJR Am J Roentgenol 2001;177:1171–1176.[Abstract/Free Full Text]
  22. Baskaran V, Pereles FS, Nemcek AA Jr, et al. Gadolinium-enhanced 3D MR angiography of renal artery stenosis: a pilot comparison of maximum intensity projection, multiplanar reformatting, and 3D volume-rendering postprocessing algorithms. Acad Radiol 2002;9:50–59.[CrossRef][Medline]
  23. Mahnken AH, Wildberger JE, Sinha AM, et al. Value of 3D-volume rendering in the assessment of coronary arteries with retrospectively ECG-gated multislice spiral CT. Acta Radiol 2003;44:302–309.[CrossRef][Medline]
  24. Lu B, Dai RP, Jiang SL, et al. Effects of window and threshold levels on the accuracy of three-dimensional rendering techniques in coronary artery electron-beam CT angiography. Acad Radiol 2001;8:754–761.[CrossRef][Medline]
  25. Vogl TJ, Abolmaali ND, Diebold T, et al. Techniques for the detection of coronary atherosclerosis: multi–detector row CT coronary angiography. Radiology 2002;223:212–220.[Abstract/Free Full Text]
  26. Dewey M, Laule M, Krug L, et al. Multisegment and halfscan reconstruction of 16-section computed tomography for detection of coronary artery stenoses. Invest Radiol 2004;39:223–229.[CrossRef][Medline]
  27. Kefer J, Coche E, Legros G, et al. Head-to-head comparison of three-dimensional navigator-gated magnetic resonance imaging and 16-section computed tomography to detect coronary artery stenosis in patients. J Am Coll Cardiol 2005;46:92–100.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 J Am Coll CardiolHome page
M. J. Budoff, D. Dowe, J. G. Jollis, M. Gitter, J. Sutherland, E. Halamert, M. Scherer, R. Bellinger, A. Martin, R. Benton, et al.
Diagnostic Performance of 64-Multidetector Row Coronary Computed Tomographic Angiography for Evaluation of Coronary Artery Stenosis in Individuals Without Known Coronary Artery Disease: Results From the Prospective Multicenter ACCURACY (Assessment by Coronary Computed Tomographic Angiography of Individuals Undergoing Invasive Coronary Angiography) Trial
J. Am. Coll. Cardiol., November 18, 2008; 52(21): 1724 - 1732.
[Abstract] [Full Text] [PDF]

Home page
 J Am Coll CardiolHome page
J. A. Rumberger
Coronary Computed Tomography Angiography: Our Time Has Come, But There Are Miles to Go Before We Sleep
J. Am. Coll. Cardiol., November 18, 2008; 52(21): 1733 - 1735.
[Full Text] [PDF]

Home page
 J Am Coll Cardiol ImgHome page
V. Cheng, A. Gutstein, A. Wolak, Y. Suzuki, D. Dey, H. Gransar, L. E.J. Thomson, S. W. Hayes, J. D. Friedman, and D. S. Berman
Moving beyond binary grading of coronary arterial stenoses on coronary computed tomographic angiography insights for the imager and referring clinician.
J. Am. Coll. Cardiol. Img., July 1, 2008; 1(4): 460 - 471.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Ferencik, M.
Right arrow Articles by Achenbach, S.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Ferencik, M.
Right arrow Articles by Achenbach, S.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE