DOI: 10.1148/rg.263055182
RadioGraphics 2006;26:887-904
© RSNA, 2006
Multidetector CT for Visualization of Coronary Stents1
Francesca Pugliese, MD,
Filippo Cademartiri, MD, PhD,
Carlos van Mieghem, MD,
Willem B. Meijboom, MD,
Patrizia Malagutti, MD,
Nico R. A. Mollet, MD, PhD,
Carlo Martinoli, MD,
Pim J. de Feyter, MD, PhD and
Gabriel P. Krestin, MD, PhD
1 From the Departments of Radiology (F.P., F.C., N.R.A.M., G.P.K.) and Cardiology (C.v.M., W.B.M., P.M., P.J.d.F.), Erasmus MC, Dr Molenwaterplein 40, 3015 GD Rotterdam, the Netherlands; and Department of Radiology, University of Genoa, Genoa, Italy (C.M.). Presented as an education exhibit at the 2004 RSNA Annual Meeting. Received September 29, 2005; revision requested October 24 and received November 10; accepted December 15. All authors have no financial relationships to disclose.
Address correspondence to F.P. (e-mail: francesca.pugliese{at}libero.it).
 |
Abstract
|
|---|
Whereas the clinical diagnosis of in-stent thrombosis is straightforward, that of in-stent restenosis remains a problem, because although many patients experience chest pain after coronary stent placement, that symptom is secondary to ischemia in only a few. The use of a noninvasive technique to identify such patients for early invasive intervention versus more conservative management is thus highly desirable. Multidetector computed tomography (CT) performed with 16-section scanners recently emerged as such a technique and has overtaken modalities such as electron-beam CT and magnetic resonance imaging as an alternative to conventional angiography for the assessment of in-stent restenosis. The improved hardware design of the current 64-section CT scanners allows even better delineation of stent struts and lumen. The more reliable criterion of direct lumen visualization thus may be substituted for the presence of distal runoff, which lacks specificity for a determination of in-stent patency because of the possibility of collateral pathways. However, the capability to accurately visualize the in-stent lumen depends partly on knowledge of the causes of artifacts and how they can be compensated for with postprocessing and proper image display settings. In addition, an understanding of the major stent placement techniques used in the treatment of lesions at arterial bifurcations is helpful.
© RSNA, 2006
|
Introduction
|
|---|
Over the past 25 years, catheter-based intervention has become the dominant form of coronary revascularization. Percutaneous coronary interventions are increasingly performed instead of coronary artery bypass graft surgery, even in patients with three-vessel disease or left main coronary artery disease. The most important advance in the field of percutaneous coronary interventions was the introduction of coronary stent implantation in the 1990s, which led to reductions in both the risk of acute major complications and the incidence of restenosis, compared with the risks after balloon angioplasty (1,2). Although the use of recently introduced drug-eluting stents has resulted in even further reductions in the occurrence of restenosis, in-stent thrombosis and neointimal hyperplasia may still occur and cause partial or complete obstruction.
Whereas the clinical diagnosis of stent occlusion due to thrombosis is usually straightforward in patients with a recent stent implantation and with a subsequent onset of acute myocardial ischemia leading to acute myocardial infarction, the assessment of in-stent restenosis is more challenging. Restenosis occurs in approximately 10%–20% of patients with complex lesion characteristics (3).
Moreover, although several characteristics of high-risk populations have been described as clinical predictors, the likelihood of restenosis in a particular patient remains largely unpredictable (4–7). For these reasons, conventional coronary angiography is still the technique of choice for the diagnosis of in-stent restenosis, although cardiac catheterization may involve major complications and is associated with moderate to high costs. Magnetic resonance (MR) angiography also can depict the coronary anatomy and help detect stenoses in the proximal segments of coronary arteries (8). However, metallic stents cause magnetic susceptibility artifacts that may prevent visualization of the lumen (9). Electron-beam computed tomography (CT) has been used more successfully to visualize coronary stents (10,11). Electron-beam CT is the modality with the shortest image acquisition times, namely 100 msec for a 3-mm-thick section suitable for morphologic interpretation and 50 msec for an 8-mm-thick section used for flow analysis without direct visualization of the in-stent lumen. However, image noise is extremely high with the first technique, and only severe, flow-limiting stenoses can be detected by using the flow technique. Thus, the occurrence of nonobstructive neointimal hyperplasia remains unnoticed at electron-beam CT. In addition, patients who undergo percutaneous coronary intervention may experience a progression of atherosclerosis in native coronary vessels without a stent implant, but electron-beam CT is suboptimal for monitoring such progression, because it requires the sequential triggered acquisition of multiple 3-mm-thick sections.
Hence, the latest generation of multidetector (multisection) CT scanners, which offer a smaller voxel size, faster gantry rotation speed, and reconstruction of 64 sections per gantry rotation, provide an appealing alternative for noninvasive luminal assessment in patients with chest pain after coronary stent placement (12). The improved hardware configuration of 64-section CT scanners allows direct visualization of the stent struts and lumen for a more reliable assessment of in-stent patency than is allowed by the visualization of distal runoff. In symptomatic patients, multidetector CT may be used as a complement or a substitute for treadmill testing; because the latter lacks specificity, additional noninvasive investigations such as stress echocardiography and scintigraphy often are required before cardiac catheterization is undertaken. Multidetector CT also can be useful to assess the condition of the whole coronary tree, as it provides information about the number, severity, and location of coronary lesions. In the follow-up of asymptomatic patients after stent implantation, multidetector CT might help overcome the limited accuracy of treadmill testing to rule out restenosis and thus enable a reduction in the number of further examinations (ie, stress echocardiography, scintigraphy) needed because of a positive or inconclusive test result.
In this article, the capability of 64-section CT coronary angiography for the evaluation of in-stent patency is discussed. The authors provide explanatory illustrations of the major stent implantation techniques used to treat lesions at or near arterial branching points. They describe the CT features and artifacts that may be observed in or near stent implants on multiplanar reformatted images and volume-rendered images, the causes of those visual characteristics, and the postprocessing and image display methods that may be used to avoid or minimize artifacts.
|
Percutaneous Coronary Intervention and Stent Placement
|
|---|
Indications and Outcomes
The clinical indications for percutaneous coronary interventions cover the spectrum of ischemic heart disease, from unstable angina pectoris and acute myocardial infarction to silent ischemia, as summarized in the guidelines of the American College of Cardiologists and the American Heart Association (13). In patients with significant narrowing of a single coronary artery, the main benefit of revascularization is the relief of angina rather than an improvement of the already good prognosis with medical therapy. In contrast, in patients with significant left main coronary artery stenosis or multiple-vessel disease, revascularization may both relieve angina and improve long-term survival.
Epidemiology
An estimated 1,204,000 inpatients underwent percutaneous coronary interventions in the United States in 2002, and coronary stent placement accounted for 537,000 of such procedures (14–16). More than 80% of percutaneous interventions that were performed in 2004 involved the placement of a drug-eluting stent coated with sirolimus or paclitaxel (17). The procedural success rate of percutaneous revascularization is higher than 90%, and the risk of sudden arterial occlusion and subsequent myocardial infarction is low (13). Despite contradictory reports (18), there is evidence that the long-term survival of patients who have undergone percutaneous revascularization for two- or three-vessel disease is no worse than that achieved with bypass graft surgery (19). Thus, percutaneous coronary intervention has become the preferred coronary revascularization strategy in many countries.
Coronary Stents
Coronary stents are expandable devices that are delivered to the coronary artery via catheter and then expanded to preserve the luminal diameter. The currently available stents are premounted on dedicated delivery systems. Occurrences that may limit the success of stent implantation include in-stent restenosis and thrombosis, which may obstruct the flow through the stent.
Thrombosis
The frequency of in-stent thrombosis is low, with a cumulative incidence of 1.3%–1.7% at 9-month follow-up (20). However, even this incidence is clinically important because in-stent thrombosis is associated with high mortality and morbidity due to acute myocardial infarction. In the era of uncoated metallic stents, in-stent thrombosis typically occurred acutely (less than 48 hours after stent implantation) or subacutely (2–30 days after implantation). One of the current concerns about drug-eluting stents is the occurrence of delayed in-stent thrombosis that is manifested more than 30 days after stent implantation. This late manifestation may be related to delayed endothelialization of the stent and typically occurs when antiplatelet therapy is discontinued (21,22). In patients with additional risk factors (eg, renal failure, diabetes mellitus, or low ejection fraction), stents implanted at the level of coronary artery bifurcations are considered to present a higher risk of thrombosis (23). The same applies when very long stents are used, with a reported 1.03 relative risk of thrombosis for each 1-mm increase in length (20).
Restenosis
Restenosis, the major limitation to the long-term outcome of percutaneous coronary intervention, is defined as vessel lumen narrowing of more than 50% after angioplasty, with the resultant recurrence of angina. Restenosis is an iatrogenic process caused by an excessive arterial healing response to vessel injury associated with dilation. It results from the combined effects of elastic recoil (24), vascular remodeling (25), and neointimal hyperplasia (26). Coronary stents represent a mechanical approach to the prevention of restenosis by virtually eliminating elastic recoil and negative remodeling of the vessel after balloon dilation (26). The occurrence of neointimal hyperplasia is mainly responsible for the observed rates of restenosis, which range from less than 10% with a drug-eluting stent to 40% with an uncoated metallic stent (1,2,27,28). For both stent types, excess stent length is associated with an increased risk of in-stent restenosis (29); thus, the arbitrary use of stents much longer than the actual lesion length is not advisable (30).
Drug-eluting Stents
Drug-eluting stents were developed to help prevent in-stent restenosis. Coating of a conventional metallic stent with an antiproliferative agent helps preserve the mechanical scaffolding properties of the stent; the drug is released locally, at the site of the vascular injury (31). The most extensive accumulated clinical experience to date is that with polymer-coated sirolimus- and paclitaxel-eluting stents. Favorable safety profiles and decreased restenosis rates have resulted in the widespread use of percutaneous coronary interventions and drug-eluting stents since their release in early 2003 (32–34). Ongoing developments in stent design include the creation of biodegradable, nonmetallic, and MR-compatible devices. Theoretically, a biodegradable drug-eluting stent may be the ideal solution to prevent in-stent resteno-sis. The response to vessel wall damage can be suppressed by the drug, while elastic recoil and negative remodeling are prevented by the stent. Eventually, the stent degrades, and chronic vessel injury related to the metal or polymer is thereby prevented (35).
|
Multidetector CT for Visualization of Coronary Stents
|
|---|
Four– and 16–Detector Row CT Scanners
The diagnostic accuracy of electron-beam CT and multidetector CT performed with different generations of scanners is summarized in the Table. Whereas in-stent lumen evaluation with CT was almost impossible with four–detector row scanners (36–38), the introduction of 16–detector row scanners (in combination with dedicated convolution filters) made CT a much more viable modality for the detection of significant in-stent restenosis, with reported sensitivity and specificity values in the ranges of 54%–100% and 88%–100%, respectively (12,39–44) (Table). It is worth noting that the observation of distal runoff cannot be considered an absolute indicator of patency, since the presence of vessel enhancement distal to a stent could also be secondary to retrograde filling. Indeed, whereas in conventional coronary angiography the contrast agent is selectively injected into the coronary artery, CT requires injection into a peripheral vein instead. This allows retrograde flow via collateral branches to the vessel segment distal to an occluded or diseased stent. Attempts also have been made to assess coronary artery stent patency with 16–detector row CT scanners on the basis of contrast enhancement measurements (43) or pixel count methods (44). However, the detection of more subtle degrees of in-stent neointimal hyperplasia was beyond the capabilities of that generation of CT scanners.
Multidetector 64-Section CT Systems
Although the calibers of coronary stents are no smaller than those of the major native coronary arteries and their branches, the depiction of the in-stent coronary lumen at CT is a greater challenge than is that of the native coronary artery lumen because of high-attenuation artifacts secondary to the metallic stent struts. High in-plane and through-plane spatial resolution, optimal contrast resolution, and minimization of high-attenuation artifacts are paramount in order to overcome the technical challenges. The superiority of 64-section CT systems over earlier generations of CT scanners with regard to the detection of coronary artery stenoses in native vessels has been demonstrated in recent investigations (45–47). Improvements in CT hardware technology, such as high x-ray output, isotropic voxel size of 0.4 x 0.4 x 0.4 mm (48), acquisition times of 6–14 seconds, and the capability of rendering 64 sections per rotation, can play a valuable role also in the evaluation of the intra–coronary artery stent lumen. Indeed, advances in multidetector CT technology mean that thinner sections can be obtained in a shorter time, with resultant increased spatial resolution along the z-axis and with almost motion-free data sets (49). Temporal resolution depends primarily on a gantry rotation speed faster than those available with earlier scanners. The reduced breath-holding time is better tolerated by patients and contributes to the minimization of motion-related artifacts. Coupled with these improvements in hardware design, electrocardiography (ECG)-based gating techniques and specialized methods of image reconstruction are used. Depending on the patient’s heart rate, data acquired during one cardiac cycle (monosegmental) or multiple cardiac cycles (multisegmental) can be used for section reconstruction. With these combined advantages, current 64-section CT systems provide better spatial and temporal resolution than do earlier generations of multidetector CT scanners, and higher spatial resolution and a better signal-to-noise ratio than do electron-beam CT scanners (50). In the clinical evaluation of coronary stents, better delineation of the graft struts and of the presence of in-stent restenosis is possible with 64-section CT technology.
|
Stent Imaging with Multidetector CT
|
|---|
General Issues
In addition to the specifications of scanner hardware, scanning technique, and dedicated postprocessing (ie, convolution filter), variables such as stent diameter, material, and design as well as patient characteristics may heavily affect the visibility of the in-stent lumen. The earliest experiments to assess the feasibility of coronary stent imaging with multidetector CT were performed in vitro with varying collimations, contrast material concentrations, stent calibers, and stent positions within the gantry (51–53). When imaging is performed in vivo, stent-related beam hardening artifacts are a constant phenomenon, and assessment is further complicated by vessel wall calcifications, poor contrast-to-noise ratios in obese patients, and motion.
Beam Hardening and Blooming Effect
Metallic struts cause a severe CT artifact known as blooming effect. Blooming effect results from beam hardening and causes the stent struts to appear thicker than they are (51) and, often, to overlap the vessel lumen. The result is an underestimation of the in-stent luminal diameter (Fig 1a). The energy spectrum of the x-ray beam as it passes through a hyperattenuating structure increases because lower-energy photons are absorbed more rapidly than are higher-energy photons, with the result that the beam is more intense when it reaches the detectors. Calcified spots of the vessel wall near or at the outer surface of an implanted stent also contribute to beam hardening, which further erodes the assessability of the stent lumen (Fig 1b, 1c). Depending on the type of metal and the design of the stent, the magnitude of the artifact varies (Fig 2) (53). As a rule, the depiction of stents with the slimmest profile is least affected by blooming artifacts.
View larger version (125K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1a. Blooming effect on follow-up images obtained with 64-section CT in a patient who underwent stent implantation in the left circumflex coronary artery. (a) Longitudinal multiplanar reformatted image shows, at the outer edge of the stent, a calcified spot that contributes to beam hardening and hampers visualization of the in-stent lumen. Note the insufficient dilation of the stent proximal to the bulky calcification. (b, c) Sharp-kernel-filtered cross-sectional image (b), obtained at the level indicated in a (line), is less affected by blooming than is the smooth-kernel-filtered cross-sectional image (c).
|
|
View larger version (131K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1b. Blooming effect on follow-up images obtained with 64-section CT in a patient who underwent stent implantation in the left circumflex coronary artery. (a) Longitudinal multiplanar reformatted image shows, at the outer edge of the stent, a calcified spot that contributes to beam hardening and hampers visualization of the in-stent lumen. Note the insufficient dilation of the stent proximal to the bulky calcification. (b, c) Sharp-kernel-filtered cross-sectional image (b), obtained at the level indicated in a (line), is less affected by blooming than is the smooth-kernel-filtered cross-sectional image (c).
|
|
View larger version (125K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1c. Blooming effect on follow-up images obtained with 64-section CT in a patient who underwent stent implantation in the left circumflex coronary artery. (a) Longitudinal multiplanar reformatted image shows, at the outer edge of the stent, a calcified spot that contributes to beam hardening and hampers visualization of the in-stent lumen. Note the insufficient dilation of the stent proximal to the bulky calcification. (b, c) Sharp-kernel-filtered cross-sectional image (b), obtained at the level indicated in a (line), is less affected by blooming than is the smooth-kernel-filtered cross-sectional image (c).
|
|
View larger version (99K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2a. Variation in the severity of metal-related artifacts at 64-section CT with variations in metallic content, design, and luminal diameter of the stent. (a) Curved multiplanar reformatted image obtained in a patient with a 4-mm-caliber stent in the proximal right coronary artery (arrow) and 2.50-mm-caliber (arrowhead) and 2.25-mm-caliber stents in the posterolateral artery. Although all three stents consist of the same material, the in-stent lumen in the two stents in the posterolateral artery is not visible because of the small stent caliber. Note the gap between the stents implanted in the posterolateral artery. (b) Image obtained in another patient, who underwent stent implantation (different stent type, 5-mm caliber) in the proximal circumflex artery, shows a more pronounced metal-related artifact than is visible in a.
|
|
View larger version (93K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2b. Variation in the severity of metal-related artifacts at 64-section CT with variations in metallic content, design, and luminal diameter of the stent. (a) Curved multiplanar reformatted image obtained in a patient with a 4-mm-caliber stent in the proximal right coronary artery (arrow) and 2.50-mm-caliber (arrowhead) and 2.25-mm-caliber stents in the posterolateral artery. Although all three stents consist of the same material, the in-stent lumen in the two stents in the posterolateral artery is not visible because of the small stent caliber. Note the gap between the stents implanted in the posterolateral artery. (b) Image obtained in another patient, who underwent stent implantation (different stent type, 5-mm caliber) in the proximal circumflex artery, shows a more pronounced metal-related artifact than is visible in a.
|
|
Beam hardening is counterbalanced mainly by increasing the spatial resolution (decreasing the voxel size) and performing dedicated data filtering.
Tube voltage is usually a constant parameter in cardiac protocols, but the contrast between structures on images also depends on the amount of x-ray energy. At lower-voltage settings, which correspond to lower energy of the x-ray beam, the CT number of metal increases substantially, and, hence, beam hardening artifacts are more likely to occur. Beam hardening artifacts also may be exacerbated by motion or by inappropriate selection of the reconstruction window (54). Conversely, they may be minimized by reducing the amount of motion inherent in the data set and optimizing the reconstruction window.
Partial Volume Averaging and Interpolation
Another obstacle to coronary stent imaging is related to partial volume averaging and interpolation. Inherent in all digital tomographic imaging techniques, partial volume averaging yields a CT number that represents the average attenuation of the materials within a voxel. At stent imaging in vessels with a large diameter, such as the aorta or iliac arteries, beam hardening and partial volume averaging effects are present but are limited to the proximity of the vessel wall. In coronary arteries with smaller diameters, the artifacts are of the same magnitude, but a reliable assessment of the lumen is much more problematic. The smaller the stent, the more detrimental the effect of partial volume averaging on the assessability of the in-stent lumen (Fig 2a).
The thinner detector width on 64-section CT scanners partly solves this problem by reducing the voxel size and thereby the general assessability of the stent lumen (48,55,56).
Optimization of Contrast Enhancement
Prominent contrast enhancement in the lumen is a prerequisite for robust coronary CT angiography (57). It is achieved not only by optimizing the contrast material injection parameters (ie, using a high-concentration contrast agent and a fast injection rate) but also by accurately synchronizing the CT data acquisition with the passage of the contrast agent by means of bolus tracking or a test bolus. Edge-enhancing convolution filters, which may be used for better delineation of stents, have the drawback of producing noisier data sets. If such a convolution filter is used, the assessability of the in-stent lumen particularly benefits from the presence of a high degree of intraluminal contrast enhancement, which somewhat compensates for the kernel-related noise. A high degree of intraluminal enhancement is recommended especially for the investigation of stent patency in vessels that have a small diameter and thus contain less blood.
Residual Cardiac Motion
Residual cardiac motion is of the utmost importance as a cause of vessel nonassessability at multidetector CT coronary angiography. Residual cardiac motion also plays a role in exacerbating metal-related artifacts such as beam hardening and partial volume averaging effects (Fig 3). The use of high gantry rotation speeds, multisegmental reconstruction techniques, and ß-blockers to lower the heart rate consistently improves the interpretability of multidetector CT coronary angiograms. ECG-based editing techniques allow an improvement of image quality for patients with mild irregularities in sinus rhythm, such as premature beats, and for those with bundle-branch block.
View larger version (97K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3a. Residual cardiac motion exacerbates metal-related artifacts at 64-section CT in a patient with a stent in the midportion of the right coronary artery and with a premature heartbeat recorded at ECG during scanning. (a, b) Images obtained from data acquired during gating with the original ECG tracing. On the volume-rendered image (a), a stepladder artifact (arrowheads) is visible at the level of the midportion of the right coronary artery. On the multiplanar reformatted image (b), a blurring of contours is visible. (c, d) Images obtained with cardiac gating after editing of the ECG tracing. To avoid a gap in the image data, the reconstruction window during the premature heartbeat was deleted and another was added to the subsequent cardiac cycle. This step eliminated the abrupt heart rate change related to the premature beat and allowed a more coherent reconstruction of the data set. On the volume-rendered image (c), the appearance of the stent (arrow) is unaffected by motion artifacts. Likewise, the in-stent lumen is well depicted on the multiplanar reformatted image (d).
|
|
View larger version (143K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3b. Residual cardiac motion exacerbates metal-related artifacts at 64-section CT in a patient with a stent in the midportion of the right coronary artery and with a premature heartbeat recorded at ECG during scanning. (a, b) Images obtained from data acquired during gating with the original ECG tracing. On the volume-rendered image (a), a stepladder artifact (arrowheads) is visible at the level of the midportion of the right coronary artery. On the multiplanar reformatted image (b), a blurring of contours is visible. (c, d) Images obtained with cardiac gating after editing of the ECG tracing. To avoid a gap in the image data, the reconstruction window during the premature heartbeat was deleted and another was added to the subsequent cardiac cycle. This step eliminated the abrupt heart rate change related to the premature beat and allowed a more coherent reconstruction of the data set. On the volume-rendered image (c), the appearance of the stent (arrow) is unaffected by motion artifacts. Likewise, the in-stent lumen is well depicted on the multiplanar reformatted image (d).
|
|
View larger version (100K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3c. Residual cardiac motion exacerbates metal-related artifacts at 64-section CT in a patient with a stent in the midportion of the right coronary artery and with a premature heartbeat recorded at ECG during scanning. (a, b) Images obtained from data acquired during gating with the original ECG tracing. On the volume-rendered image (a), a stepladder artifact (arrowheads) is visible at the level of the midportion of the right coronary artery. On the multiplanar reformatted image (b), a blurring of contours is visible. (c, d) Images obtained with cardiac gating after editing of the ECG tracing. To avoid a gap in the image data, the reconstruction window during the premature heartbeat was deleted and another was added to the subsequent cardiac cycle. This step eliminated the abrupt heart rate change related to the premature beat and allowed a more coherent reconstruction of the data set. On the volume-rendered image (c), the appearance of the stent (arrow) is unaffected by motion artifacts. Likewise, the in-stent lumen is well depicted on the multiplanar reformatted image (d).
|
|
View larger version (152K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3d. Residual cardiac motion exacerbates metal-related artifacts at 64-section CT in a patient with a stent in the midportion of the right coronary artery and with a premature heartbeat recorded at ECG during scanning. (a, b) Images obtained from data acquired during gating with the original ECG tracing. On the volume-rendered image (a), a stepladder artifact (arrowheads) is visible at the level of the midportion of the right coronary artery. On the multiplanar reformatted image (b), a blurring of contours is visible. (c, d) Images obtained with cardiac gating after editing of the ECG tracing. To avoid a gap in the image data, the reconstruction window during the premature heartbeat was deleted and another was added to the subsequent cardiac cycle. This step eliminated the abrupt heart rate change related to the premature beat and allowed a more coherent reconstruction of the data set. On the volume-rendered image (c), the appearance of the stent (arrow) is unaffected by motion artifacts. Likewise, the in-stent lumen is well depicted on the multiplanar reformatted image (d).
|
|
|
Convolution Filters
|
|---|
Once the technical requirements and acquisition parameters for adequate coronary CT angiography have been fulfilled, additional dedicated postprocessing can be performed to optimize the visualization of the stent. The use of a dedicated edge-enhancing convolution kernel allows a significant decrease in the severity of blooming artifacts at the edges of high-attenuation structures. Indeed, the in-stent contrast-enhanced attenuation measured on sharp-kernel images is closer to that measured in the proximal or distal lumen than is the in-stent attenuation measured on smooth-kernel images (43). According to the authors of a recently published article about the effects of reconstruction kernels on the delineation of coronary stents (58), when a medium-smooth filter is applied, as a consequence of the blooming effect an average in-stent lumen narrowing of 37% is observed in comparison with the diameter of the untreated vessel segment, and the measured in-stent attenuation exceeds the aortic attenuation by more than 100 HU because of the partial volume averaging effect. In contrast, the in-stent lumen narrowing is 29% of the diameter of the normal vessel segment on sharp-filtered images, and the in-stent attenuation exceeds the attenuation in the aorta by only 60 HU. The results of such studies support the superior depiction of the stent lumen with dedicated edge-enhancing convolution kernels in comparison with conventional medium-smooth kernels. While spatial resolution is increased and blooming artifact is reduced by the application of edge-enhancing filters, an increase in image noise has to be accepted as a trade-off. Thus, the most appropriate filter must be chosen so that an advantageous balance is achieved between the visualization of low-contrast structures and image noise (Fig 4). In these instances, high intraluminal contrast enhancement is also very helpful to counterbalance the increased image noise.
View larger version (107K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4a. Visibility of low-contrast structures with different convolution filters. The most appropriate filter must be chosen so that an advantageous balance is achieved between the visibility of low-contrast structures and the quantity of image noise. (a) Conventional coronary angiogram shows nonsignificant neointimal hyperplasia in the distal portion of the right coronary artery (arrowheads). (b–f) Multidetector 64-section CT angiograms obtained in a patient with multiple stents in the right coronary artery. On the image reconstructed with a smooth convolution filter (B20f) (b), the luminal defect (*) is hardly visible. On the image reconstructed with a medium-smooth convolution filter (B30f) (c), the defect (*) is visible but quite blurred. The image reconstructed with a dedicated edge-enhancing kernel (B46f) (d) allows visualization of in-stent neointimal hyperplasia (*), with good contrast between the defect and the surrounding structures (stent scaffold, enhanced lumen). On the images reconstructed with sharp (B60f) (e) and very sharp (B70f) (f) convolution filters, the edge enhancement does not provide clearer depiction of the defect (*) but, instead, greater amounts of image noise.
|
|
View larger version (130K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4b. Visibility of low-contrast structures with different convolution filters. The most appropriate filter must be chosen so that an advantageous balance is achieved between the visibility of low-contrast structures and the quantity of image noise. (a) Conventional coronary angiogram shows nonsignificant neointimal hyperplasia in the distal portion of the right coronary artery (arrowheads). (b–f) Multidetector 64-section CT angiograms obtained in a patient with multiple stents in the right coronary artery. On the image reconstructed with a smooth convolution filter (B20f) (b), the luminal defect (*) is hardly visible. On the image reconstructed with a medium-smooth convolution filter (B30f) (c), the defect (*) is visible but quite blurred. The image reconstructed with a dedicated edge-enhancing kernel (B46f) (d) allows visualization of in-stent neointimal hyperplasia (*), with good contrast between the defect and the surrounding structures (stent scaffold, enhanced lumen). On the images reconstructed with sharp (B60f) (e) and very sharp (B70f) (f) convolution filters, the edge enhancement does not provide clearer depiction of the defect (*) but, instead, greater amounts of image noise.
|
|
View larger version (138K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4c. Visibility of low-contrast structures with different convolution filters. The most appropriate filter must be chosen so that an advantageous balance is achieved between the visibility of low-contrast structures and the quantity of image noise. (a) Conventional coronary angiogram shows nonsignificant neointimal hyperplasia in the distal portion of the right coronary artery (arrowheads). (b–f) Multidetector 64-section CT angiograms obtained in a patient with multiple stents in the right coronary artery. On the image reconstructed with a smooth convolution filter (B20f) (b), the luminal defect (*) is hardly visible. On the image reconstructed with a medium-smooth convolution filter (B30f) (c), the defect (*) is visible but quite blurred. The image reconstructed with a dedicated edge-enhancing kernel (B46f) (d) allows visualization of in-stent neointimal hyperplasia (*), with good contrast between the defect and the surrounding structures (stent scaffold, enhanced lumen). On the images reconstructed with sharp (B60f) (e) and very sharp (B70f) (f) convolution filters, the edge enhancement does not provide clearer depiction of the defect (*) but, instead, greater amounts of image noise.
|
|
View larger version (159K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4d. Visibility of low-contrast structures with different convolution filters. The most appropriate filter must be chosen so that an advantageous balance is achieved between the visibility of low-contrast structures and the quantity of image noise. (a) Conventional coronary angiogram shows nonsignificant neointimal hyperplasia in the distal portion of the right coronary artery (arrowheads). (b–f) Multidetector 64-section CT angiograms obtained in a patient with multiple stents in the right coronary artery. On the image reconstructed with a smooth convolution filter (B20f) (b), the luminal defect (*) is hardly visible. On the image reconstructed with a medium-smooth convolution filter (B30f) (c), the defect (*) is visible but quite blurred. The image reconstructed with a dedicated edge-enhancing kernel (B46f) (d) allows visualization of in-stent neointimal hyperplasia (*), with good contrast between the defect and the surrounding structures (stent scaffold, enhanced lumen). On the images reconstructed with sharp (B60f) (e) and very sharp (B70f) (f) convolution filters, the edge enhancement does not provide clearer depiction of the defect (*) but, instead, greater amounts of image noise.
|
|
View larger version (201K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4e. Visibility of low-contrast structures with different convolution filters. The most appropriate filter must be chosen so that an advantageous balance is achieved between the visibility of low-contrast structures and the quantity of image noise. (a) Conventional coronary angiogram shows nonsignificant neointimal hyperplasia in the distal portion of the right coronary artery (arrowheads). (b–f) Multidetector 64-section CT angiograms obtained in a patient with multiple stents in the right coronary artery. On the image reconstructed with a smooth convolution filter (B20f) (b), the luminal defect (*) is hardly visible. On the image reconstructed with a medium-smooth convolution filter (B30f) (c), the defect (*) is visible but quite blurred. The image reconstructed with a dedicated edge-enhancing kernel (B46f) (d) allows visualization of in-stent neointimal hyperplasia (*), with good contrast between the defect and the surrounding structures (stent scaffold, enhanced lumen). On the images reconstructed with sharp (B60f) (e) and very sharp (B70f) (f) convolution filters, the edge enhancement does not provide clearer depiction of the defect (*) but, instead, greater amounts of image noise.
|
|
View larger version (203K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4f. Visibility of low-contrast structures with different convolution filters. The most appropriate filter must be chosen so that an advantageous balance is achieved between the visibility of low-contrast structures and the quantity of image noise. (a) Conventional coronary angiogram shows nonsignificant neointimal hyperplasia in the distal portion of the right coronary artery (arrowheads). (b–f) Multidetector 64-section CT angiograms obtained in a patient with multiple stents in the right coronary artery. On the image reconstructed with a smooth convolution filter (B20f) (b), the luminal defect (*) is hardly visible. On the image reconstructed with a medium-smooth convolution filter (B30f) (c), the defect (*) is visible but quite blurred. The image reconstructed with a dedicated edge-enhancing kernel (B46f) (d) allows visualization of in-stent neointimal hyperplasia (*), with good contrast between the defect and the surrounding structures (stent scaffold, enhanced lumen). On the images reconstructed with sharp (B60f) (e) and very sharp (B70f) (f) convolution filters, the edge enhancement does not provide clearer depiction of the defect (*) but, instead, greater amounts of image noise.
|
|
Additional "intelligent" noise-reduction filters may prove beneficial for the depiction of low-contrast structures within the in-stent lumen and therefore may help detect in-stent restenosis (58).
|
Interpretation of Multidetector CT Data
|
|---|
Display Techniques and Windowing
The clinical evaluation of coronary arteries on CT angiograms is routinely performed by using multiplanar reformation of the data volume. Curved multiplanar reformation, maximum intensity projection, and volume rendering techniques also are often used. The same techniques are applied for the evaluation of coronary stents on CT angiograms. Multiplanar reformatted images and cross-sectional images of the stent are the most useful views on which to assess patency, restenosis, or a minor degree of neointimal hyperplasia. Advances in CT technology have provided 64-section CT scanners with submillimeter spatial resolution of 0.4 x 0.4 x 0.4 mm (43,48). With the very small isotropic voxel size, the assessability of the stent lumen on multiplanar reformatted images remains unaffected by angulation in relation to the z-axis.
CT window settings affect image contrast and noise. If the window is set too narrow, image noise may be significantly increased, and gray-scale differentiation of fine structural details may be lost. Wide window settings are necessary for accurate evaluation of the in-stent lumen at CT angiography (window width, 1500 HU; window center, 300 HU) (Fig 5).
View larger version (165K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5a. Combined effects of the selected filter and window settings on image contrast and noise at 64-section CT coronary angiography. (a–c) Images obtained with a medium-smooth convolution kernel (B30f). (d–f) Images obtained with a dedicated sharp convolution kernel (B46f). Note that d–f more clearly depict the in-stent lumen than do a–c. The standard soft-tissue window width (W) (a, d) is too narrow and accentuates blooming artifacts. On the image filtered with a medium-smooth convolution kernel (a), the blooming effect totally obscures the in-stent lumen. With exaggerated widening of the window (b, e), the blooming effect is decreased, but this occurs at the expense of overall image contrast. Setting the window center (C) at approximately 300 HU and choosing a width (W) of approximately 1500 HU allows a more favorable balance between image contrast and noise (c, f). However, the window settings alone are not sufficient to ensure optimal depiction of the inner lumen (c). The combined use of a dedicated edge-enhancing convolution kernel, which increases image spatial resolution, and appropriate window settings to compensate for filter-related noise allows the most favorable in-stent lumen visualization (f).
|
|
View larger version (98K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5b. Combined effects of the selected filter and window settings on image contrast and noise at 64-section CT coronary angiography. (a–c) Images obtained with a medium-smooth convolution kernel (B30f). (d–f) Images obtained with a dedicated sharp convolution kernel (B46f). Note that d–f more clearly depict the in-stent lumen than do a–c. The standard soft-tissue window width (W) (a, d) is too narrow and accentuates blooming artifacts. On the image filtered with a medium-smooth convolution kernel (a), the blooming effect totally obscures the in-stent lumen. With exaggerated widening of the window (b, e), the blooming effect is decreased, but this occurs at the expense of overall image contrast. Setting the window center (C) at approximately 300 HU and choosing a width (W) of approximately 1500 HU allows a more favorable balance between image contrast and noise (c, f). However, the window settings alone are not sufficient to ensure optimal depiction of the inner lumen (c). The combined use of a dedicated edge-enhancing convolution kernel, which increases image spatial resolution, and appropriate window settings to compensate for filter-related noise allows the most favorable in-stent lumen visualization (f).
|
|
View larger version (131K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5c. Combined effects of the selected filter and window settings on image contrast and noise at 64-section CT coronary angiography. (a–c) Images obtained with a medium-smooth convolution kernel (B30f). (d–f) Images obtained with a dedicated sharp convolution kernel (B46f). Note that d–f more clearly depict the in-stent lumen than do a–c. The standard soft-tissue window width (W) (a, d) is too narrow and accentuates blooming artifacts. On the image filtered with a medium-smooth convolution kernel (a), the blooming effect totally obscures the in-stent lumen. With exaggerated widening of the window (b, e), the blooming effect is decreased, but this occurs at the expense of overall image contrast. Setting the window center (C) at approximately 300 HU and choosing a width (W) of approximately 1500 HU allows a more favorable balance between image contrast and noise (c, f). However, the window settings alone are not sufficient to ensure optimal depiction of the inner lumen (c). The combined use of a dedicated edge-enhancing convolution kernel, which increases image spatial resolution, and appropriate window settings to compensate for filter-related noise allows the most favorable in-stent lumen visualization (f).
|
|
View larger version (176K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5d. Combined effects of the selected filter and window settings on image contrast and noise at 64-section CT coronary angiography. (a–c) Images obtained with a medium-smooth convolution kernel (B30f). (d–f) Images obtained with a dedicated sharp convolution kernel (B46f). Note that d–f more clearly depict the in-stent lumen than do a–c. The standard soft-tissue window width (W) (a, d) is too narrow and accentuates blooming artifacts. On the image filtered with a medium-smooth convolution kernel (a), the blooming effect totally obscures the in-stent lumen. With exaggerated widening of the window (b, e), the blooming effect is decreased, but this occurs at the expense of overall image contrast. Setting the window center (C) at approximately 300 HU and choosing a width (W) of approximately 1500 HU allows a more favorable balance between image contrast and noise (c, f). However, the window settings alone are not sufficient to ensure optimal depiction of the inner lumen (c). The combined use of a dedicated edge-enhancing convolution kernel, which increases image spatial resolution, and appropriate window settings to compensate for filter-related noise allows the most favorable in-stent lumen visualization (f).
|
|
View larger version (116K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5e. Combined effects of the selected filter and window settings on image contrast and noise at 64-section CT coronary angiography. (a–c) Images obtained with a medium-smooth convolution kernel (B30f). (d–f) Images obtained with a dedicated sharp convolution kernel (B46f). Note that d–f more clearly depict the in-stent lumen than do a–c. The standard soft-tissue window width (W) (a, d) is too narrow and accentuates blooming artifacts. On the image filtered with a medium-smooth convolution kernel (a), the blooming effect totally obscures the in-stent lumen. With exaggerated widening of the window (b, e), the blooming effect is decreased, but this occurs at the expense of overall image contrast. Setting the window center (C) at approximately 300 HU and choosing a width (W) of approximately 1500 HU allows a more favorable balance between image contrast and noise (c, f). However, the window settings alone are not sufficient to ensure optimal depiction of the inner lumen (c). The combined use of a dedicated edge-enhancing convolution kernel, which increases image spatial resolution, and appropriate window settings to compensate for filter-related noise allows the most favorable in-stent lumen visualization (f).
|
|
View larger version (155K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5f. Combined effects of the selected filter and window settings on image contrast and noise at 64-section CT coronary angiography. (a–c) Images obtained with a medium-smooth convolution kernel (B30f). (d–f) Images obtained with a dedicated sharp convolution kernel (B46f). Note that d–f more clearly depict the in-stent lumen than do a–c. The standard soft-tissue window width (W) (a, d) is too narrow and accentuates blooming artifacts. On the image filtered with a medium-smooth convolution kernel (a), the blooming effect totally obscures the in-stent lumen. With exaggerated widening of the window (b, e), the blooming effect is decreased, but this occurs at the expense of overall image contrast. Setting the window center (C) at approximately 300 HU and choosing a width (W) of approximately 1500 HU allows a more favorable balance between image contrast and noise (c, f). However, the window settings alone are not sufficient to ensure optimal depiction of the inner lumen (c). The combined use of a dedicated edge-enhancing convolution kernel, which increases image spatial resolution, and appropriate window settings to compensate for filter-related noise allows the most favorable in-stent lumen visualization (f).
|
|
Although they are not routinely used to assess CT scans in the clinical setting, other three-dimensional volume rendering techniques are available with current CT scanners. After suitable threshold ranges or transparency settings are selected, an endoscopic view of the internal surface of the vessel can be simulated (Fig 6). This technique may allow visualization of different stent designs (eg, slotted tube and corrugated ring stents) (59). In stents with small diameters, though, the amount of image noise may be too great for such techniques to be reproducibly applied.
View larger version (212K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6a. Selection of suitable threshold ranges and opacity settings allows simulation of an endoscopic view of the inner vessel surface and makes it possible to recognize different designs of the metal scaffold in stents. A slotted tubular stent (a) and a corrugated ring stent (b) are currently used for the treatment of most coronary lesions. Both stents have a diameter of 3 mm. The amount of image noise may prevent successful application of this technique for depiction of the lumen in stents with very small diameters.
|
|
View larger version (222K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6b. Selection of suitable threshold ranges and opacity settings allows simulation of an endoscopic view of the inner vessel surface and makes it possible to recognize different designs of the metal scaffold in stents. A slotted tubular stent (a) and a corrugated ring stent (b) are currently used for the treatment of most coronary lesions. Both stents have a diameter of 3 mm. The amount of image noise may prevent successful application of this technique for depiction of the lumen in stents with very small diameters.
|
|
In-Stent Lumen Evaluation
As mentioned earlier (39), the direct visualization of the in-stent lumen is important for assessing patency, because collateral vessels may be feeding the vessel segment distal to the occluded stent in a retrograde direction. An accurate intraluminal evaluation can best be performed by means of multiplanar reformation of the CT data volume. The stent may be considered to be occluded if the lumen inside the device appears darker than the contrast-enhanced vessel lumen proximal to the stent.
Unless severe artifacts affect the CT data set, stent evaluation may proceed beyond a judgment of patency or occlusion. Nonocclusive in-stent neointimal hyperplasia is characterized by the presence of a darker rim between the stent and the contrast-enhanced vessel lumen (Fig 4) and is secondary to the healing response to procedure-related vessel injury. If neointimal hyperplasia exceeds a luminal diameter reduction of 50%, the process is consistent with hemodynamically significant in-stent restenosis (Fig 7) (39). In-stent restenosis typically occurs as a localized nonenhancing lesion, often (but not invariably) associated with complex lesion anatomy (ie, ostial lesions) and discontinuity in lesion coverage. It occurs with higher frequency in patients with diabetes mellitus (26). Restenosis may occur either within or adjacent to the stent (within 5 mm of the stent extremities). Edge restenosis might occur because of a decrease in local drug availability, incomplete lesion coverage due to a gap between two stents, procedure-related trauma, or damage to the polymer coating of a stent from calcifications or an overlapping stent (33).
View larger version (173K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7a. In-stent occlusion in a patient with recurrent angina pectoris 18 months after implantation of two stents in the right coronary artery. CT was performed after conventional angiography failed to depict the right coronary artery. (a) Multiplanar reformatted image shows lower attenuation inside the stent lumina than in the proximal untreated tract of the right coronary artery, a gap between the occluded stents, and collateral filling (*). (b–d) Cross-sectional images obtained at the proximal end of the stent (b) (1 in a), in the middle portion (c) (2 in a), and at the distal end (d) (3 in a) show the appearances of patency, occlusion, and patency, respectively. (e, f) Conventional angiograms provide information about the presence of collateral filling (* in e) and the length of the occlusion. This information enabled planning for percutaneous revascularization, which was successful, as evident from a comparison of the pretreatment image (e) and the posttreatment image (f).
|
|
View larger version (102K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7b. In-stent occlusion in a patient with recurrent angina pectoris 18 months after implantation of two stents in the right coronary artery. CT was performed after conventional angiography failed to depict the right coronary artery. (a) Multiplanar reformatted image shows lower attenuation inside the stent lumina than in the proximal untreated tract of the right coronary artery, a gap between the occluded stents, and collateral filling (*). (b–d) Cross-sectional images obtained at the proximal end of the stent (b) (1 in a), in the middle portion (c) (2 in a), and at the distal end (d) (3 in a) show the appearances of patency, occlusion, and patency, respectively. (e, f) Conventional angiograms provide information about the presence of collateral filling (* in e) and the length of the occlusion. This information enabled planning for percutaneous revascularization, which was successful, as evident from a comparison of the pretreatment image (e) and the posttreatment image (f).
|
|
View larger version (94K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7c. In-stent occlusion in a patient with recurrent angina pectoris 18 months after implantation of two stents in the right coronary artery. CT was performed after conventional angiography failed to depict the right coronary artery. (a) Multiplanar reformatted image shows lower attenuation inside the stent lumina than in the proximal untreated tract of the right coronary artery, a gap between the occluded stents, and collateral filling (*). (b–d) Cross-sectional images obtained at the proximal end of the stent (b) (1 in a), in the middle portion (c) (2 in a), and at the distal end (d) (3 in a) show the appearances of patency, occlusion, and patency, respectively. (e, f) Conventional angiograms provide information about the presence of collateral filling (* in e) and the length of the occlusion. This information enabled planning for percutaneous revascularization, which was successful, as evident from a comparison of the pretreatment image (e) and the posttreatment image (f).
|
|
View larger version (89K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7d. In-stent occlusion in a patient with recurrent angina pectoris 18 months after implantation of two stents in the right coronary artery. CT was performed after conventional angiography failed to depict the right coronary artery. (a) Multiplanar reformatted image shows lower attenuation inside the stent lumina than in the proximal untreated tract of the right coronary artery, a gap between the occluded stents, and collateral filling (*). (b–d) Cross-sectional images obtained at the proximal end of the stent (b) (1 in a), in the middle portion (c) (2 in a), and at the distal end (d) (3 in a) show the appearances of patency, occlusion, and patency, respectively. (e, f) Conventional angiograms provide information about the presence of collateral filling (* in e) and the length of the occlusion. This information enabled planning for percutaneous revascularization, which was successful, as evident from a comparison of the pretreatment image (e) and the posttreatment image (f).
|
|
View larger version (98K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7e. In-stent occlusion in a patient with recurrent angina pectoris 18 months after implantation of two stents in the right coronary artery. CT was performed after conventional angiography failed to depict the right coronary artery. (a) Multiplanar reformatted image shows lower attenuation inside the stent lumina than in the proximal untreated tract of the right coronary artery, a gap between the occluded stents, and collateral filling (*). (b–d) Cross-sectional images obtained at the proximal end of the stent (b) (1 in a), in the middle portion (c) (2 in a), and at the distal end (d) (3 in a) show the appearances of patency, occlusion, and patency, respectively. (e, f) Conventional angiograms provide information about the presence of collateral filling (* in e) and the length of the occlusion. This information enabled planning for percutaneous revascularization, which was successful, as evident from a comparison of the pretreatment image (e) and the posttreatment image (f).
|
|
View larger version (106K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7f. In-stent occlusion in a patient with recurrent angina pectoris 18 months after implantation of two stents in the right coronary artery. CT was performed after conventional angiography failed to depict the right coronary artery. (a) Multiplanar reformatted image shows lower attenuation inside the stent lumina than in the proximal untreated tract of the right coronary artery, a gap between the occluded stents, and collateral filling (*). (b–d) Cross-sectional images obtained at the proximal end of t | |
Top
Abstract
Introduction
