Dual-Source CT Angiography of Peripheral Arterial Stents: In Vitro Evaluation of 22 Different Stent Types

Purpose. To test different peripheral arterial stents using four image reconstruction approaches with respect to lumen visualization, lumen attenuation and image noise in dual-source multidetector row CT (DSCT) in vitro. Methods and Materials. 22 stents (nitinol, steel, cobalt-alloy, tantalum, platinum alloy) were examined in a vessel phantom. All stents were imaged in axial orientation with standard parameters. Image reconstructions were obtained with four different convolution kernels. To evaluate visualization characteristics of the stent, the lumen diameter, intraluminal density and noise were measured. Results. The mean percentage of the visible stent lumen diameter from the nominal stent diameter was 74.5% ± 5.7 for the medium-sharp kernel, 72.8% ± 6.4 for the medium, 70.8% ± 6.4 for the medium-smooth and 67.6% ± 6.6 for the smooth kernel. Mean values of lumen attenuation were 299.7HU ± 127 (medium-sharp), 273.9HU ± 68 (medium), 270.7HU ± 53 (medium-smooth) and 265.8HU ± 43. Mean image noise was: 54.6 ± 6.3, 20.5 ± 1.7, 16.3 ± 1.7, 14.0 ± 2 respectively. Conclusion. Visible stent lumen diameter varies depending on stent type and scan parameters. Lumen diameter visibility increases with the sharpness of the reconstruction kernel. Smoother kernels provide more realistic density measurements inside the stent lumen and less image noise.


Introduction
Arterial stenoses and occlusions in different regions of the body are frequently treated with angioplasty and stent implantation. After stent implantation, there is a risk of in-stent restenosis which can be caused by neointimal proliferation, vessel wall inflammation, or stent thrombosis [1,2]. Therefore follow-up examinations are needed after successful revascularization for detecting restenosis.
At present, there are different imaging techniques available. Digital subtraction angiography (DSA) has been the standard modality for evaluating stent patency for a long time. However, its invasiveness is afflicted with possible complications and a less invasive alternative is eligible. Duplex sonography of arteries is noninvasive but highly operator dependant.
Three-dimensional contrast-enhanced MR angiography (MRA) is a frequently used noninvasive alternative to DSA for screening patients [3,4]. However, MRA of stented arteries is difficult because stent-related susceptibility artifacts may disturb stent lumen visibility [5][6][7]. Hamer et al. concluded in their studies in 2005 and 2006 that MRA is not yet a reliable technique to characterize in-stent stenoses [8,9].
Spiral computed tomography angiography (CTA) is another noninvasive method to evaluate peripheral arteries and has been shown to be an alternative to intra-arterial DSA in a postinterventional followup [10]. It was shown that CTA provides comparable findings to intra-arterial DSA for the detection of stenoses in renal arteries [11]. Studies with 16-row CT scanners showed promising results regarding assessability of in-stent stenosis with CTA [12][13][14].
Modern CT scanners are equipped with 64 or more detector rows and increased gantry rotation speed to allow the acquisition of more slices with a better spatial resolution in a shorter time. Clinical data regarding stent evaluation are available as single center data as well as a meta-analysis [14,15]. The purpose of this study was to investigate the 2 Radiology Research and Practice stent lumen visibility and artifacts of different peripheral stents for different locations of the body in a CT angiography examination. Furthermore, the study aimed at comparing different CT reconstruction kernels with regard to stent imaging.

Evaluated Stents and Experimental Setup.
Twenty-one different stents and one stentgraft of different material and design were studied. Manufacturer, material, design, length, and nominal diameter of the stents and stent graft are summarized in Table 1. Ten stents were made of Nitinol, seven of stainless surgical steel (316 L), two of a cobaltbased alloy, two of tantalum, and one of a platinum alloy. The Wallstent was made of a cobalt-based alloy covered by polyethylene (PET).
The stents and the stentgraft were inserted in plastic tubes with 5, 7, 8, 10, or 13 mm lumen diameter exactly matching their nominal diameter with one exception: the Palmaz Genesis Stent had a nominal diameter of 6 mm and was implanted into a 5 mm plastic tube. The wall of the small tubes (5, 7, and 8 mm) had a thickness of <0.3 mm, and the material of the bigger tubes (10 and 13 mm) was about 1 mm thick.
The tubes were filled with contrast material (Ultravist 300, Schering AG, Berlin, Germany) diluted to 250 HU, closed at both ends and positioned in a plastic container filled with vegetable oil. The density of the oil was adjusted to −70 HU by addition of Lipiodol Ultrafluid (Byk Gulden, Konstanz, Germany) to simulate perivascular fat. The tubes with implanted stents and stentgraft were then positioned in the gantry in an orientation parallel to the z-axis of the scanner.
Axial images with a slice thickness of 0.6 mm were used for the evaluation. Secondary multiplanar reformations (MPRs) were created for demonstration purposes only.

Evaluation of Visualization Characteristics of the Stent:
Lumen Diameter, Intraluminal Density, and Noise. Axial reformations of all stents were evaluated in a window width of 1500 HU and a center of 300 HU as shown for the SAXX Large and the Wallstent Uni in Figure 1. This setting has proven to be useful for the evaluation of coronary stents in previous studies [16,17]. The diameter of the visible stent lumen in the center of the stent and on two adjacent images was measured as shown in Figure 2 using the electronic caliper for distance measurements provided with the CT system's standard software. From these three measurements a mean value for each stent was calculated. Attenuation values inside the visible stent lumen were measured by a region of interest technique (ROI) in the same three images to calculate a mean stent lumen attenuation. The ROI with a size of 12 pixels was placed in the center of the visible stent lumen without inclusion of the stent struts or streak artifacts.
Image noise was defined as the standard deviation of a ROI density measurement outside the vessels in the surrounding oily fluid.
All measurement results are displayed as mean, standard deviation, and range. As the measurement results for visible lumen diameter, lumen attenuation, and image noise proved not to be normally distributed within the groups of the four different kernels the nonparametric Friedman test was used to check for overall differences among the reconstruction methods.
A posthoc analysis was carried out with a paired comparison between all kernels using the Wilcoxon test. At P < .05 statistical significance was assumed.

Diameters of the Visible Stent
Lumen. The visible stent diameters using the four reconstruction protocols are summarized in Table 2.
Using the B20f kernel reconstruction, the visible lumen diameter ranged from 49.0% in the Renal 137 stent to 77.3% in the Evo Target stent, and the mean visible lumen diameter was 67.6 ± 6.6%.
Using the B30f kernel for image reconstruction, the visible lumen diameter ranged from 52.4% in the Renal 137 stent to 81.7% in the Evo Target stent (mean 70.8 ± 6.4%).
Using the B40f kernel, the visible lumen diameter ranged from 53.3% in the Renal 137 stent to 83.7% in the Evo Target stent (mean 72.8 ± 6.4%).
With the B50f kernel the visible lumen diameter ranged from 57.1% in the Renal 137 stent to 83.3% in the Evo Target stent (mean 74.5 ± 5.7%). In this reconstruction severe artifacts were found especially within the stents Renal 109, Renal 137, and CP Stent, so that the stent lumen could not be evaluated.
Differences between all the kernels were highly significant with P < .01 in the Wilcoxon test. The improvement of the stent lumen visibility by using sharper kernel reconstructions can be estimated in Figure 1.

Noise.
Mean noise values of the four reconstruction kernels are summarized in Table 2. The noise increased from the smooth to the sharper kernels: 14.0 ± 2.0 HU in B20f, 16.3 ± 1.7 HU in B30f, 20.5 ± 1.7 HU in B40f, and 54.6 ± 6.3 HU in B50f. The differences between the examined reconstruction protocols were highly significant with P < .01 in the Friedman test and all the paired Wilcoxon tests.

Comparison of Lumen Visibility and Lumen Attenuation between the Different Materials.
We compared the results of the stents depending on their material. The results are shown in Table 3

Discussion
In the present study a considerable number of different peripheral stents were examined in a state-of-the-art CT system regarding their lumen visibility using CT angiography (CTA). The investigated stents were made from different  materials (stainless steel (316 L), Nitinol, Cobalt alloy, Tantal, and Platinum-Iridium alloy) and for different areas of applications (arterial and peripheral arteries, in general, biliary duct, carotid arteries, renal arteries, iliac arteries, femoral arteries and aorta). Our study shows that CTA can be used for follow-up examinations in the majority of the evaluated stents. However, the lumen visibility differs extensively between different stent types.
The phantom used was designed to simulate conditions comparable to an in vivo CTA examination. Nevertheless, some limitations have to be considered. In all scans the stents were positioned parallel to the z-axis of the scanner. This would resemble the in vivo position of an aortic stent but not that of renal or iliac stents. Several groups have shown that stent artifacts of coronary stents depend on the angle between stent and scanner [18,19]; therefore the visible stent lumen in an in vivo study may differ from our results.
We used a static fluid model, but the effect of flow on the artifact expression should be negligible because CTA works with differences in radiopacity and not with flow parameters.
In one stent the nominal diameter did not exactly match the diameter of the "vessel": the 6 mm Palmaz stent was implanted in a 5 mm tube. This was necessary because no appropriate 6 mm tube was available. The possible effect of stent strut compaction when implanting a large diameter stent in a smaller diameter lumen should be minor in this case but must be considered.
It was stated before that the artifacts depend on the stent design and material. Our study demonstrates that stents made from stainless steel (316 L) or Nitinol show the best results in CTA compared to other stents we examined. The Cobalt alloy stents also showed good results in our study. The Platinum-Iridium alloy stent and the tantalum stents showed poorer results, because "blooming" artifacts obscured parts of the stent lumen. The higher magnitude of artifacts in these stents may be mainly due to the higher atomic number of platinum (78) and tantalum (73) when compared to steel (26), cobalt (27), chromium (24), or nickel (28). Therefore, it is useful to know what kind of stent was implanted before the CT examination.
Our study showed better results for lumen visibility with a modern state-of-the-art CT scanner with 64 detectors than previous studies with a four-slice scanner [20,21]. Eichhorn et al. stated that image quality rises with number of detectors and the diameter of the stent [22]. Our study underlines that sharper kernels show better lumen visibility than smoother ones as Heuschmid et al. [23] stated. Furthermore in our study the B40f reconstruction showed the best compromise  between lumen visibility and noise. The increase of lumen visibility of B50f compared to B40f is only 2.3%. The effect of high pitch protocols on coronary artery stent imaging has been investigated [24]; results for iliac artery stents remain to be published. A recent study shows the feasibility of a low-dose protocol for detecting in-stent restenoses of iliac artery stents [25].
In summary, all investigated stents seem to be suitable for the evaluation of high-grade stenoses (lumen visibility >50%). Except for the tantalum stents it should even be possible to detect smaller stenoses (lumen visibility >66%). After all CTA with a modern CT scanner seems to be a helpful noninvasive method for the follow-up examination of stented arterial stenoses.