Published online August 7, 2003, 10.1148/rg.e16
(Radiographics. 2003;23:e16.)
© RSNA, 2003
Coronary Artery Imaging with Multidetector CT: Visualization Issues1
Peter M. A. van Ooijen, MSc,
Kai Yiu Ho, MD, PhD,
Joost Dorgelo, MD and
Matthijs Oudkerk, MD, PhD
1 From the Department of Radiology, Groningen University Hospital, Hanzeplein 1, 9700 RB, Groningen, The Netherlands. Presented as an educational exhibit at the 2002 RSNA scientific assembly. Received March 7, 2003, revision requested May 5, final revision received and accepted June 20. Address correspondence to P.M.A.v.O. (e-mail: p.m.a.van.ooyen@rad.azg.nl).
 |
Abstract
|
|---|
Noninvasive imaging of the coronary arteries has attracted growing interest in the past few years. One of the possible acquisition techniques is multidetector computed tomography (CT) that produces large three-dimensional (3D) data sets that require visualization techniques for data evaluation. The objective of this article is to increase knowledge of possible 3D visualization techniques together with their advantages and disadvantages for the routine evaluation of cardiac data sets. Common imaging techniques available to the radiologist at standard workstations are multiplanar reformation (MPR), oblique MPR, curved MPR, maximum-intensity projection (MIP), shaded-surface display, and direct volume rendering. Each of these techniques has its advantages and disadvantages for the visualization of the coronary artery tree. Several additions to the basic techniques have been developed to overcome some of their shortcomings. Different clinical examinations, such as stent evaluation, stenosis evaluation, and bypass evaluation, require different visualization techniques. The choice of preferred technique for each clinical study depends on the advantages and disadvantages of the various techniques as described in the literature. Because of the large number of possible settings and projection angles, it is important for users to interactively manipulate the images and review the whole vessel volume rather than just looking at static reformatted images. Errors such as findings of false stenoses can be avoided by means of accurate and appropriate use of software features. This requires training of users both with regard to the capabilities of the software and the background of the different techniques and their possible pitfalls. The authors believe that volume rendering of the whole heart is useful for anatomic evaluation of the coronary arteries. For more detailed observation of specific lesions, slab imaging with volume rendering or MIP is required.
© RSNA, 2003
Index Terms: Computed tomography (CT), angiography, 51.12116 • Compted tomography (CT), multi-detector row, 54.12115 • Computed tomography (CT), three-dimensional, 54.12117 • Coronary vessels, CT, 54.12115, 54.12117 • Heart, CT, 51.12116, 51.12117
|
Introduction
|
|---|
Noninvasive imaging of the coronary arteries has attracted growing interest in the past few years. One of the possible acquisition techniques is multidetector computed tomography (CT) that produces large three-dimensional (3D) data sets that require visualization techniques for data evaluation. Common imaging techniques available to the radiologist at standard workstations are multiplanar reformation (MPR), oblique MPR, curved MPR, maximum-intensity projection (MIP), shaded-surface display (SSD), and direct volume rendering (DVR) (1–9). Each of these techniques has its advantages and disadvantages for the visualization of the coronary artery tree. Several additions to the basic techniques have been developed to overcome some of their shortcomings. The objective of this article is to increase knowledge of possible 3D visualization techniques together with their advantages and disadvantages for the routine evaluation of cardiac data sets obtained by means of noninvasive coronary artery imaging with 16-section CT.
|
Materials and Methods
|
|---|
Cardiac data sets were acquired with a multidetector CT scanner (Sensation 16; Siemens, Erlangen, Germany) and an intravenous contrast material injection of 165 mL of iomeprol (400 mg/mL) (Iomeron; Bracco-Byk, Italy) at 4.0 mL/sec. Scan parameters were defined as shown in Table 1.
Postprocessing of data was performed on a commercially available workstation (Vitrea2; Vital Images, Plymouth, Minn). This workstation allowed evaluation of the coronary arteries with MPR, MIP, or DVR. SSD was not supported by this software package and hence not used or described.
|
Multiplanar Reformation
|
|---|
In MPR, a plane is defined inside the 3D volume and only the data in this plane are displayed. An MPR can be performed by using either straight planes (Fig 1) or curved planes (Fig 2). When performing an MPR with straight planes, the thickness of the selection is set to 0 as the default. When a greater thickness is selected, a slab MPR is created. Usually this slab is rendered with MIP or DVR; the application of this technique is discussed in the corresponding sections below.
View larger version (86K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1. Oblique MPR of the right coronary artery performed with only straight planes. Resulting image of the right coronary artery is shown in the upper right image. The upper left and lower right images show the positioning of the MPR planes in two dimensions. The lower left image shows the positioning of the MPR planes in 3D.
|
|
View larger version (88K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2. Curved MPR of the right coronary artery (curved green line in the lower right image). The upper right image shows the resulting visualization of the right coronary artery. The upper left and lower right images show the positioning of the MPR planes in two dimensions. The lower left image shows the positioning of the MPR planes in 3D.
|
|
More advanced methods of performing curved MPR allow localization of a vessel by selection of a point within the lumen of the vessel, after which automatic segmentation of the centerline of the vessel is used to perform the curved MPR (Fig 3).
View larger version (124K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3. Curved MPR after automated vessel detection combined with display of sections of the vessel perpendicular to the centerline. The centerline of the vessel is represented by the green line in the 3D image. The two rectangles to the far right show two perpedicular cross sections of the right coronary artery along the (green) centerline. The large square shows a cross section of the artery at the level of the pointer in the rectangular images. The smaller squares show the cross sections along the centerline, both before and after the position of the larger square.
|
|
A major disadvantage of the MPR methods is the high dependence on the manual orientation of the planes. Therefore, MPR is prone to introduce false-positive or false-negative stenoses (Fig 4). Interactive viewing of these type of images from multiple viewing angles is therefore required (10). Furthermore, only one branch of a vessel can be displayed at a time. For each branch, a separate MPR image is required.
View larger version (8K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4. Schematics show two cross sections of a vessel, one normal (left) and one with a stenosis (right). The horizontal lines denote the selected image plane; the small circles denote the user-selected points within the vessel. False-positive stenoses can be introduced by incorrect placement of the point within the vessel; on the left, the top point is incorrectly placed and results in a smaller vessel diameter than the bottom point. False-negative stenoses can be introduced by incorrect orientation of the plane; on the right, a plane perpendicular to the current one should be selected to visualize the stenosis.
|
|
Advantages of MPR are the ease of use and speed of the MPR algorithm. Furthermore, MPR provides images containing all available information (all Hounsfield unit values are retained).
|
Maximum-Intensity Projection
|
|---|
With MIP, only the highest-attenuation voxels are preserved. This algorithm casts a ray through the 3D data for each pixel in the resulting image, and only the highest-attenuation voxels found on a ray are preserved. Because of the general infusion of contrast medium, not only the coronary arteries are filled with contrast medium but also the cavities of the heart and other vascular structures. These cavities and other vascular structures will overlap the coronary arteries when full-volume MIP is applied. Therefore, a so-called "sliding-thin-slab" MIP is used (11). To obtain this MIP, a thin-slab MPR is selected from which an MIP image is reconstructed. This slab is moved through the volume, with the slab movement distance smaller than the slab thickness, and at each step an MIP is created (Movie 1).
Another way of applying MIP to visualize the coronary artery tree is by segmentation of all contrast-enhanced structures except the coronary arteries (12). This requires manual or semiautomatic segmentation of all cavities and vascular structures, after which a good representation of the coronary arteries can be obtained (Fig 5).
View larger version (102K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5. Fully segmented MIP. Images obtained after (left) and before (right) segmentation of contrast-enhanced cavities and vascular structures other than the coronary arteries. Top images are in anteroposterior view, bottom images in left-to-right view.
|
|
Disadvantages of MIP are that only a fraction of the available data is used and that the algorithm results in overestimation of stenoses (4,13). Furthermore, many artifacts are known to exist in MIP images and no 3D depth is obtained.
MIP also has some advantages: The algorithm is fast and easy to configure, images show a good differentiation between vessels and background, and calcifications are well visualized.
|
Direct Volume Rendering
|
|---|
With DVR, all available data can be used. Certain properties (opacity, color, etc) are assigned to each voxel value. The most important property is opacity, which enables the user to hide certain objects or make them transparent. Figures 6 and 7 show a simplified example of the application of opacity. In DVR, a ray is cast through the volume and an equation is used on the voxels traversed by the ray to compute the resulting pixel value. The squares at the top of Figure 7 represent the voxel values traversed along one line through a volume; the square at the far right is the resulting pixel. Each of the squares has a certain value, and for each value an opacity is defined (Fig 6). For example, voxels with a value of 100 have 25% opacity (which equals 75% transparency), and voxels with a value of 20 have 100% opacity. In Figure 7, the top row of numbers represents the voxel values and the second row represents the corresponding opacity percentages. To compute the resulting pixel value, we use the following equation: C(sum[n]) = A + B, where A is opacity x voxel value, which is the contribution of the current voxel to the pixel value, and B is the contribution of the previously traversed voxels and is computed by multiplying 1 - opacity (transparency) by the previously computed pixel value, or sum[n - 1]. Note that when a pixel with opacity of 100% is traversed, the contribution of previously traversed voxels will be zero because they are not visible.
To demonstrate the effect of transparency, different settings are applied to the same data set of a coronary calcification phantom. This phantom consists of a large cylinder representing the heart and a smaller one representing the spine. Different tissues can be distinguished within this phantom on the basis of their voxel values. In this case, we have the soft tissue of the heart, bone marrow, bone of the spine, and calcified lesions in different sizes and densities (Fig 8).
View larger version (82K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 8. Different opacity settings in the same phantom. Upper left: The soft tissues are opaque (large cylinder). Upper right: The soft tissues are made transparent, thus revealing the calcified lesions within the volume of the heart (large cylinder). Lower left: The soft tissue is made fully transparent, clearly showing the full boundary of the spine (small cylinder) and the calcified lesions. Lower right: The spine is made partly transparent.
|
|
The use of opacity and transparency is one of the great advantages of DVR. In addition, DVR provides good 3D depth and can use all the data available. The disadvantages are the relatively time-consuming computations required and the multitude of user-definable parameters and settings. Because of upgrades in performance of both hardware and software, the computational time is becoming a minor issue. However, especially with large data sets, interactive manipulation with DVR will be slower than with one of the other techniques. The multitude of user-definable parameter settings impairs the accuracy of DVR for the evaluation of arterial stenoses, as the selection of window width, window level, brightness, and opacity differs for each observer (14).
|
Coronary Artery Evaluation
|
|---|
For the evaluation of a coronary data set with the described visualization techniques, data editing is required. The extensiveness of the editing depends on the visualization technique used.
For each clinical question, a different visualization technique should be used to obtain an optimal result. Possible clinical questions include coronary artery disease status (plaque, stenoses), coronary artery calcification burden, stent status (in-stent restenosis), and bypass-graft status.
For the detection of plaque and stenoses, optimal results are achieved by using DVR, MPR, or MIP (Fig 9). Calcified and soft plaque deposits should be evaluated with either a slab or fully segmented MIP or with a curved MPR through the centerline of the artery with a plane perpendicular to the centerline. These same views can be used for the evaluation of stenoses, although an overall 3D DVR is preferred for localization of the area of interest.
View larger version (83K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 9. Soft plaque and calcified plaque in stenosis. Top image shows the conventional angiogram (no indication of calcified plaque), middle image shows the DVR, and bottom image shows the thin-slab MIP. Ao = aorta, LAD = left anterior descending artery.
|
|
For evaluation of aberrant anatomy, 3D DVR is the preferred technique because it provides a good insight into the 3D relationships and course of the coronary artery tree (Movie 2).
Evaluation of anastomosis of coronary arteries or venous bypass grafts is best performed with an oblique or curved MPR through the site of anastomosis, although the correct positioning of this MPR may be difficult (Fig 10).
Finally, stent evaluation is also best done by using an oblique or curved MPR, with the plane situated at the center of the stent (Fig 11).
|
Discussion
|
|---|
Achenbach et al (12) state that in their study MIP showed higher accuracy than SSD for the detection of high-grade coronary artery stenosis and occlusion, and that the evaluation of the original source images permitted almost the same accuracy. Combining these three visualization methods resulted in a slight increase in accuracy. It should be pointed out that to use MIP, extensive and time-consuming manual segmentation is required. Structures other than the coronary artery tree must be removed manually. In contrast, Lu et al (15) demonstrated a high correlation between measured and real vessel diameter for SSD, while MIP showed a lower correlation and MPR was the least accurate method. However, the visualized length of the arteries was longer, and thus more accurate, with MPR than with MIP or SSD. DVR was not included in either of these studies, although it may be more accurate, especially in vessels with a small diameter such as the coronary arteries. This was confirmed by Addis et al (16), who showed more accurate results for DVR than for axial sections, MIP, MPR, or SSD for vessels with a diameter between 2 and 4 mm and statistically more accurate results for vessel diameters of 0.5–1.0 mm. Baskaran et al (17) also showed higher accuracy for DVR compared with MIP in the scoring of stenosis, with an accuracy of 87% and 89% for DVR and MPR, respectively, and an accuracy of 81% for MIP.
Each of the techniques described here has its specific advantages and disadvantages that are inherent to the underlying principle of the technique (Table 2). Although more recently introduced versions of MPR techniques allow automatic definition of the centerline used for the curved MPR, these techniques are still operator dependent in that the resulting centerline depends on the placement of the seed point for selecting the vessel. For MIP, more advanced protocols have been developed to overcome the problem of overprojection, such as the closest-vessel projection algorithm. This algorithm takes not the global maximum found on a ray cast through the volume but a local maximum closest to the viewpoint, thereby selecting the vessel (or other high-intensity structure) closest to the viewer. The speed disadvantages of DVR are largely overcome now with faster computer processors and larger memories, and the dependence on user-defined parameters is increasingly covered by standard protocols, especially for CT.
Because of its advantages and disadvantages, each visualization technique is useful for certain clinical purposes. Good representation of all gray-level information in the original data along a path through a vessel MPR, especially a curved MPR, is suitable for evaluation of stents and stent lumina. Slab imaging with either MIP or DVR covers a larger portion of the data than a simple MPR and is especially suited for evaluation of plaque, bypass-graft anastomosis, and stenosis. Finally, DVR of the full data set is suitable for evaluation of anatomy, clearly showing the distribution and course of the coronary arteries, especially anatomic deviations. Furthermore, DVR can be used to localize stenotic regions, stents, and calcifications.
|
Conclusion
|
|---|
Each visualization technique described here has its advantages and disadvantages, and visualization of the coronary artery tree with these techniques requires careful preparation. Because of the large number of possible settings and projection angles, it is important for users to interactively manipulate the images and review the whole vessel volume rather than just look at static reformatted images. False stenoses can be avoided by accurate and appropriate use of the software features. To achieve this accurate and appropriate use of the software, training of users is crucial. This training should not only focus on the capabilities of the software but on the background of the different techniques and their possible pitfalls. On the basis of our experience, we can state that DVR of the whole heart is useful for anatomic evaluation of the coronary arteries. For more detailed observation of specific lesions, slab imaging with DVR or MIP is required.
|
Footnotes
|
|---|
Abbreviations: DVR = direct volume rendering,
MIP= maximum-intensity projection,
MPR= multiplanar reformation,
SSD= shaded-surface display,
3D= three-dimensional.
|
References
|
|---|
- Fishman EK, Ney DR. Advanced computer applications in radiology: clinical applications. RadioGraphics 1993; 13:463-475.[Abstract]
- Murakami T, Kashiwagi T, Nakamura H, et al. Display of MR angiograms: maximum intensity projection versus three-dimensional rendering. Eur J Radiol 1993; 17:95-100.[CrossRef][Medline]
- Rubin GD, Dake MD, Napel S, et al. Spiral CT of renal artery stenosis: comparison of three-dimensional rendering techniques. Radiology 1994; 190:181-189.[Abstract]
- Heath DG, Soyer PA, Kuszyk BS, et al. Three-dimensional spiral CT during arterial portography: comparison of three rendering techniques. RadioGraphics 1995; 15:1001-1011.[Abstract]
- Meyers SP, Talagala SL, Totterman S, et al. Evaluation of the renal arteries in kidney donors: value of three-dimensional phase-contrast MR angiography with maximum-intensity-projection or surface rendering. AJR Am J Roentgenol 1995; 164:117-121.[Abstract/Free Full Text]
- Soyer P, Heath D, Bluemke DA, et al. Three-dimensional helical CT of intrahepatic venous structures: comparison of three rendering techniques. J Comput Assist Tomogr 1996; 20:122-127.[CrossRef][Medline]
- Hany TF, Schmidt M, Davis CP, Gohde SC, Debatin JF. Diagnostic impact of four postprocessing techniques in evaluating contrast-enhanced three-dimensional MR angiography. AJR Am J Roentgenol 1998; 170:907-912.[Abstract/Free Full Text]
- Rankin SC. CT angiography. Eur Radiol 1999; 9:297-310.[CrossRef][Medline]
- Rensing BJ, Bongaerts AH, van Geuns RJ, van Ooijen PM, Oudkerk M, de Feyter PJ. Intravenous coronary angiography using electron beam computed tomography. Prog Cardiovasc Dis 1999; 42:139-148.[CrossRef][Medline]
- Nakanishi T, Ito K, Imazu M, Yamakido M. Evaluation of coronary artery stenoses using electron-beam CT and multiplanar reformation. J Comput Assist Tomogr 1997; 21:121-127.[CrossRef][Medline]
- Napel S, Rubin GD, Jeffrey RB, Jr. STS-MIP: a new reconstruction technique for CT of the chest. J Comput Assist Tomogr 1993; 17:832-838.[Medline]
- Achenbach S, Ropers D, Regenfus M, Muschiol G, Daniel WG, Moshage W. Contrast enhanced electron beam computed tomography to analyse the coronary arteries in patients after acute myocardial infarction. Heart 2000; 84:489-493.[Abstract/Free Full Text]
- Anderson CM, Saloner D, Tsuruda JS, Shapeero LG, Lee RE. Artifacts in maximun-intensity-projection display of MR angiograms. AJR Am J Roentgenol 1990; 154:623-629.[Free Full Text]
- Ebert DS, Heath DG, Kuszyk BS, et al. Evaluating the potential and problems of three-dimensional computed tomography measurements of arterial stenosis. J Digit Imaging 1998; 11:151-157.
- Lu B, Dai RP, Jiang SL, et al. Effects of window and threshold levels on the accuracy of 3D rendering techniques in coronary artery EBCT angiography. Acad Radiol 2001; 8:754-761.[CrossRef][Medline]
- Addis KA, Hopper KD, Iyriboz TA, et al. CT angiography: in vitro comparison of five reconstruction methods. AJR Am J Roentgenol 2001; 177:1771-1776.
- Baskaran V, Pereles SF, Nemcek AA, 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]
This article has been cited by other articles:
|
|
|
|
 
D. T. Boll, A. S. Bossert, A. J. Aschoff, M. H. Hoffmann, and R. C. Gilkeson
Synergy of MDCT and Cine MRI for the Evaluation of Cardiac Motility
Am. J. Roentgenol.,
June 1, 2006;
186(6_Supplement_2):
S379 - S386.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
E. K. Fishman, D. R. Ney, D. G. Heath, F. M. Corl, K. M. Horton, and P. T. Johnson
Volume rendering versus maximum intensity projection in CT angiography: what works best, when, and why.
RadioGraphics,
May 1, 2006;
26(3):
905 - 922.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
T. Nakanishi, Y. Kayashima, R. Inoue, K. Sumii, and Y. Gomyo
Pitfalls in 16-Detector Row CT of the Coronary Arteries
RadioGraphics,
March 1, 2005;
25(2):
425 - 438.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
