Next generation coronary CT angiography: in vitro evaluation of 27 coronary stents

COMPUTED TOMOGRAPHY

Next generation coronary CT angiography: in vitroevaluation of 27 coronary stents

Tobias Gassenmaier & Nils Petri & Thomas Allmendinger &

Thomas Flohr & David Maintz & Wolfram Voelker &

Thorsten A. Bley

Received: 23 December 2013 /Revised: 23 May 2014 /Accepted: 7 July 2014 /Published online: 21 July 2014# European Society of Radiology 2014

AbstractObjectives To evaluate in-stent lumen visibility of 27 modernand commonly used coronary stents (16 individual stent types,two stents at six different sizes each) utilising a third-generation dual-source CT system.Methods Stents were implanted in a plastic tube filled withcontrast. Examinations were performed parallel to the sys-tem's z-axis for all stents (i.e. 0°) and in an orientation of90° for stents with a diameter of 3.0 mm. Two stents wereevaluated in different diameters (2.25 to 4.0 mm). Examina-tions were acquired with a collimation of 96×0.6 mm, tubevoltage of 120 kVp with 340 mAs tube current. Evaluationwas performed using a medium-soft (Bv40), a medium-sharp(Bv49) and a sharp (Bv59) convolution kernel optimised forvascular imaging.Results Mean visible stent lumen of stents with 3.0 mm di-ameter ranged from 53.3 % (IQR 48.9−56.7 %) to 73.9 %(66.7−76.7 %), depending on the kernel used at 0°, and washighest at an orientation of 90° with 80.0 % (75.6−82.8 %)using the Bv59 kernel, strength 4. Visible stent lumendeclined with decreasing stent size.

Conclusions Use of third-generation dual-source CT enablesstent lumen visibility of up to 80 % in metal stents and 100 %in bioresorbable stents.Key Points•Blooming artefacts impair in–stent lumen visibility of coronarystents in CT angiography.

• CT enables stent lumen visibility of up to 80 % in metalstents.

• Stent lumen visibility varies with stent orientation and size.•CTangiography may be a valid alternative for detectingin-stent restenosis.

Keywords Multislice computed tomography . Coronaryangiography . CTangiography . Stents . Artefacts

Introduction

A total of 303,832 coronary interventions were performed inGermany in 2008 and 271,329 stents have been implantedwith increasing numbers during the last couple of years [1].In-stent restenosis is a dreaded complication, which may leadto myocardial ischaemia and possibly to myocardial infarc-tion. The incidence of in-stent restenosis has decreased sincethe introduction of drug-eluting stents, but is still in the rangeof 3-20 % [2]. To detect or exclude an in-stent restenosis,several thousand coronary angiographies are performed eachyear. However, this invasive procedure exposes the patient toa small but definitive risk [3]. Thus, a non-invasive method todiagnose or exclude an in-stent restenosis is desirable.

Previous studies have shown that CT angiography (CTA)may be feasible for lumen assessment of coronary stents;however image quality is strongly dependent on the materialand architecture of the stents and depends on the technicalcapabilities of the CT system used [4, 5].

Tobias Gassenmaier and Nils Petri have equal contribution

T. Gassenmaier (*) : T. A. BleyDepartment of Diagnostic and Interventional Radiology, UniversityHospital Würzburg, Oberdürrbacher Str. 6, 97080 Würzburg,Germanye-mail: [email protected]

N. Petri :W. VoelkerDepartment of Internal Medicine I, University Hospital Würzburg,Würzburg, Germany

T. Allmendinger : T. FlohrSiemens Healthcare, Forchheim, Germany

D. MaintzDepartment of Radiology, University of Cologne, Cologne, Germany

Eur Radiol (2014) 24:2953–2961DOI 10.1007/s00330-014-3323-6

Various stent materials lead to different levels of x-rayattenuation and subsequently blooming artefacts that limitthe stent lumen assessment. With increasing numbers of de-tector rows in modern CT systems (from 4 to 320 detectorrows and dual-source scanners), significant improvements inthe lumen assessment of coronary stents have been found[4–10]. Novel reconstruction techniques have contributedsubstantially to these improvements in image quality [11].

This in vitro study evaluated the in-stent lumen visibility of27 modern and commonly used coronary stents made ofdifferent materials, different strut architectures and strut thick-nesses utilising a third-generation dual-source CT system(SOMATOM Force, Siemens Healthcare, Forchheim,Germany).

Methods

Stent selection and experimental setup

Based on 16 individual stent types, 27 balloon-expandablecoronary stents of different size, design and material wereexamined. Twenty stents were made of cobalt-chromium al-loy, four of stainless steel, one of platinum-chromium alloy,one of stainless steel and PTFE, and one of bioresorbablePoly(L-lactide) (PLLA). Stent characteristics are provided inTable 1. To determine the influence of stent diameter on lumenvisibility, two stents (Orsiro® and Integrity®) were analysed insix different diameters ranging from 2.25 mm to 4.0 mm.

All stents were implanted into a coronary vessel phantommade of a heat shrink tube consisting of polyolefin with a wallthickness ranging from 0.1 and 0.5 mm, depending on thelumen of the tube. For each stent, a plastic tube with an innerdiameter closest to the intended stent diameter was selected.All stents were expanded with the pressure defined by themanufacturer’s in vitro stent compliance table for the innerdiameter intended, which was their nominal diameter in moststents. The Taxus Liberté®, the Cypher® and one of theOrsiro® stents were expanded to an inner diameter of3.0 mm instead of their nominal diameter of 2.75 mm. Tubesnot exactly fitting the stent were carefully heated to shrink thetube’s inner diameter until it matched the outer diameter of thestent during stent’s balloon inflation. The tube was filled withcontrast medium (Imeron® 300, Bracco Imaging, Konstanz,Germany) diluted to a density of 350 HU at 120 kV. The tubewas placed in a plastic box filled with an emulsion ofsunflower oil and Lipiodol® Ultra-Fluid (Guerbet, RoissyCdG Cedex, France) adjusted to a density of -70 HU tosimulate the attenuation of epicardial fat prior to scan-ning. Examinations were performed in an orientationparallel to the scanner’s z-axis for all stents (i.e. 0°)and additionally in a stent orientation of 90° to the z-axis for stents expanded to a diameter of 3.0 mm.

CT imaging protocol

The study was performed on a third-generation dual-sourceCT (SOMATOM Force, Siemens Healthcare, Forchheim,Germany) equipped with two integrated circuit detectors(Stellar Infinity, Siemens), one providing 920 detector chan-nels to cover a 50-cm scan field of view (SFOV), the otherproviding 620 detector channels to cover a 35.4-cm SFOV.The images were acquired in a retrospectively ECG-gatedcardiac spiral dual-source mode with a collimation of 96×0.6 mm. By means of a diagonal flying focal spot, both thenumber of detector channels and the number of acquired slicesare virtually doubled during data acquisition at a gantry rota-tion time of 250 ms (1,840 respectively 1,240 virtual detectorchannels, 192×0.6-mm slice acquisition). The tube voltagewas set to 120 kVp with 340 mAs/rot quality reference tubecurrent and default ECG-based current modulation at a 70 %RR interval. Automatic exposure control was active during theacquisitions enabling the adjustment of the tube current to thesize of the phantom based on the topogram information, thusallowing clinically more realistic noise amplitude of the ex-aminations. The resulting dose index (CTDIvol) in each scanwas 3.5 mGy with a corresponding current of 12 mAs/rot anddose length product of 40mGy cm for 11.5 cm imaging range.

Data reconstruction

All CT images were reconstructed with a specified strengthlevel (possible strength settings 1 to 5) of a third-generationiterative reconstruction technique (ADMIRE, Siemens) usingan image slice thickness of 0.6 mm, an increment of 0.6 mmand field of view of 170 mm with an image matrix of 512×512 pixels. Each data set was reconstructed with a medium-soft (Bv40, strength 3), a medium-sharp (Bv49, strength 3)and a sharp (Bv59, strength 3 and strength 4) convolutionkernel optimised for vascular imaging.

The Bv40 kernel is 10 % sharper than the establishedstandard coronary vessel evaluation kernel B26f on thesecond-generation dual-source CT system SOMATOM Defi-nition Flash (Siemens Healthcare, Forchheim, Germany) withModulation Transfer Function (MTF) parameter rho50=3.95 lp/cm and rho10=6.61 lp/cm. The Bv49 kernel is equiv-alent to a B46f coronary stent evaluation kernel with a 16 %sharpness increase and MTF parameter rho50=5.62 lp/cm andrho10=7.55 lp/cm. The very sharp Bv59 kernel does not havean equivalent in the former systems. It reflects the increasedclinical capabilities, in combination with the iterative recon-struction, of the new detector system with MTF parameterrho50=8.32 lp/cm and rho10=11.73 lp/cm. It has to be notedthat the selected kernels operate below the theoretical resolu-tion limits of the system; however as they are applied in acardiac image acquisition, which is optimised for highesttemporal resolution based on a partial image type

2954 Eur Radiol (2014) 24:2953–2961

reconstruction, these kernels reflect the best trade-off betweenmaximum applicable sharpness and clinically acceptable im-age noise.

Image analysis

Axial reformations of all stents were used for evaluation usinga window width of 1,500 HU and a centre of 300 HU. Sec-ondary longitudinal multiplanar reformations (MPRs) weregenerated for demonstration purposes only (Figs. 1 and 2).The visible lumen diameter was measured on three represen-tative axial slices using the electronic calliper tool provided

with the manufacturer’s standard PACS software, including adarker rim at the vicinity of the stent struts into the measuredstent lumen diameter. Attenuation inside the visible stent lu-men was evaluated on the same three axial slices using a regionof interest (ROI) measurement as large as possible not includ-ing the stent struts artefacts (Fig. 3). Attenuation within thestent was compared to the attenuation in the tube outside of thestent. The difference was considered as stent-related attenua-tion. All measurements from the slices above were averaged toreduce measurement-related errors. Image noise was definedas the standard deviation of density in HU determined within aROI placed inside the surrounding oil emulsion.

Table 1 Stent characteristics

Number Company Name Material Coating Nominaldiameter (mm)

Length(mm)

Strut thickness(mm)

1 Abbott Absorb BVS® Poly-L-lactic acid (PLLA) Poly-D,L-lactic acid(PDLLA), Everolimus

3.0 18 0.150

2 Abbott JOSTENT Graftmaster® Stainless steel + PTFE None 3.0 16 0.300 (wallthickness)

3 Abbott Multilink Vision® Cobalt-chromium alloy None 3.0 23 0.081

4 Abbott Xience Pro® Cobalt-chromium alloy Everolimus 3.0 15 0.081

5 Biotronik Orsiro® Cobalt-chromium alloy Poly-L-lactic acid (PLLA)polymer, Sirolimus

2.25 18 0.060


2.5 13 0.060


2.75 13 0.060


3.0 18 0.060


3.5 26 0.060


4.0 15 0.060

11 Biotronik PRO-Kinetic Energy® Cobalt-chromium alloy PROBIO Amorphous SiliconCarbide Coating

3.0 20 0.060

12 Boston Scientific Omega® Platinum-chromium alloy None 3.0 16 0.081

13 Boston Scientific Taxus Liberté® Stainless steel None 3.0 38 0.100

14 Braun Coroflex Blue® Cobalt-chromium alloy None 3.0 19 0.065

15 Cordis Cypher® Stainless steel Sirolimus 3.0 28 0.140

16 Medtronic Resolute RX® Cobalt-chromiumalloy

Zotarolimus 3.0 18 0.091

17 Medtronic Resolute Integrity RX® Cobalt-chromium alloy Zotarolimus 3.5 26 0.091

18 Medtronic Resolute Integrity RX® Cobalt-chromium alloy Zotarolimus 4.0 34 0.091

19 Medtronic Integrity RX® Cobalt-chromium alloy Zotarolimus 2.25 30 0.091

20 Medtronic Integrity BMS® Cobalt-chromium alloy None 2.5 26 0.091

21 Medtronic Integrity RX® Cobalt-chromium alloy None 2.75 30 0.091




25 Orbus Neich Genous® Stainless steel Anti-hCD34 antibody 3.0 18 0.081

26 Terumo Kaname® Cobalt- chromium alloy None 3.0 28 0.080

27 Terumo Nobori® Stainless steel Poly-L-lactic acid (PLLA)polymer, Biolimus A9

3.0 28 0.120

Eur Radiol (2014) 24:2953–2961 2955

Statistical analysis

Normal distribution was evaluated by the Kolmogorov-Smirnov test and the Shapiro-Wilk test. As variables wereshown not to be entirely normally distributed, all values aredisplayed as medians, interquartile range and range. For test-ing for overall differences among the four reconstructionkernels and different stent orientations, the nonparametricFriedman test was used. A paired comparison was performedbetween all kernels using the Wilcoxon test in a post hocanalysis with the significance level adjusted according to theBonferroni correction. Correlation among stent diameter, vis-ible stent lumen and differences in attenuation values wasvalidated using Spearman's rank correlation coefficient.

Results

Mean visible stent diameters and in-stent attenuation values ofthe four different reconstruction kernels are provided inTable 2. Visible stent lumen diameters and in-stent attenua-tion values of the individual stents, including attenuationdifferences to the un-stented vessel, are provided in Table 3.

Stent lumen of 3.0-mm stents at 0° to the z-axis

At an orientation parallel to the scanner’s z-axis, mean visiblestent lumen in the Bv40 kernel was 53.3 % (48.9–56.7 %),ranging from 40 % in the Kaname® to 100 % in the AbsorbBVS®.When evaluated in the Bv49 kernel, mean visible stentlumen was 58.3 % (55.6–66.7 %), ranging from 51 % in theKaname® to 100 % in the Absorb BVS®. Using the Bv59kernel, strength 3, the mean visible stent lumen diameter was69.4% (65.6–75.6%) and 73.9% (66.7–76.7%) at strength 4,each ranging from 62 % in the Graftmaster® to 100 % in theAbsorb BVS®. Differences between all kernels used weresignificant with p≤0.002. The effect of the reconstructionkernel on stent lumen visibility is illustrated in Fig. 4.

Mean lumen density was 560 HU (543–615 HU) in theBv40 kernel, 442 HU (431–474 HU) in the Bv49 kernel, 473HU (441–514 HU) in the Bv59 kernel, strength 3, and 483 HU(448–521 HU) at strength 4. Mean lumen density in the Bv40kernel was significantly higher than in the other kernels used(p≤0.001).

Stent lumen of 3.0-mm stents at 90° to the z-axis

The Bv59 kernel, strength 4, which proved to provide the bestresults of all kernels at an orientation parallel to the scanner’s

Fig. 1 Longitudinal multiplanar reformations of all stents with a diameter of 3.0 mm at 0° and 90° to the z-axis using the Bv59 kernel, strength 4

2956 Eur Radiol (2014) 24:2953–2961

z-axis, was also evaluated at an orientation of 90° to the z-axis.Mean visible stent lumen using this setup was 80.0 %

(75.6–82.8 %), ranging from 72 % in the Omega® to100% in the Absorb BVS®. Compared to a mean visible stentlumen of 73.9% (66.7–76.7%) at an orientation parallel to thez-axis, this difference was highly significant with a p≤0.001.

Mean lumen density at an orientation of 90° to the z-axiswas 416 HU (407–436 HU) using the Bv59 kernel, strength 4.This value proved to be significantly different from the meanlumen density at an orientation of 0° [483 HU (448–521 HU);p=0.004].

Influence of stent diameter on lumen visibility

Lumen visibility in the Orsiro® and Integrity®, which wasanalysed at six different sizes ranging from 2.25 mm to4.0 mm, declined with decreasing stent size (p<0.001,Fig. 5). Comparing the smallest diameter of 2.25 mm andthe largest diameter of 4.0 mm, in the Bv40 kernel, lumenvisibility ranged from 31 % to 68 % in the Orsiro® and 31 %to 64 % in the Integrity®. When evaluated in the Bv49 kernel,the visible stent lumen ranged from 41 % to 68 % in theOrsiro® and 44 % to 63 % in the Integrity®. Using the Bv59kernel, the visible stent lumen diameter ranged from 52% and50 % at strength 3 and 4, respectively, to 76 % at bothstrengths in the Orsiro®. In the Integrity®, lumen visibilityusing this kernel ranged from 53 % to 76 % at both strengths.

In-stent attenuation for the smallest diameter of 2.25 mmwas 900 HU using the Bv40 kernel and 308 HU using theBv59 kernel, strength 4, in the Orsiro®, and 1,094 HU and305 HU, respectively, in the Integrity®. Differences in atten-uation between the stented and un-stented tube increasedsignificantly with decreasing stent size (p<0.001).

Noise

Mean noise values for the various kernels are provided inTable 2. Lowest noise levels were achieved using the Bv40

Fig. 2 Longitudinal multiplanar reformations of the Biotronik Orsiro®and Medtronic Integrity® in all available sizes using the Bv59 kernel,strength 4

Fig. 3 Measurement example ofin-stent lumen visibility andattenuation

Eur Radiol (2014) 24:2953–2961 2957

kernel with a median of 12.0 HU (11.5–12.4 HU). The highestnoise levels were observed using the Bv59 kernel, strength 4,in an orientation of 90° with a median of 46.0 HU (45.0–48.0HU). Differences between the Bv59 kernel, strength 3, and theBv59 kernel, strength 4, were not significant (p>0.05), whileall other kernels were significantly different (p≤0.001).

Discussion

Blooming artefacts caused by x-ray attenuation impair the in–stent lumen visibility in CTA. The severity of blooming arte-facts depends on the stent architecture, stent strut thicknessand material composition as well as the spatial resolu-tion and reconstruction methods of the CT. This in vitrostudy analysed the in-stent lumen visibility of variouscommonly used stents examined with a third-generationdual-source CT (SOMATOM Force, Siemens Healthcare,Forchheim, Germany).

Previous studies have shown improvements in lumen vis-ibility with advancing CT technology and reconstructionmethods [4, 5, 9]. Maintz et al. evaluated 68 stents in anin vitro setting using a 64-MSCT and later on provided anupdate with 29 stents examined by a DSCT [4, 5]. This studyshows that the evaluated third-generation DSCT providesfurther improvements regarding in-stent lumen assessment.Two previously evaluated stents, the PRO-Kinetic Energy®and the Cypher®, presented an in-stent lumen visibility of upto 76 % and 83% respectively. Compared to visible diametersof 52 % respectively 56.7 % reported previously, this is aconsiderable increase when it comes to reliably excluding in-stent restenosis [4, 5]. This is achieved by a combination of anincreased number of in-plane detector channels, diagonalflying focal spot and a newly introduced iterative reconstruc-tion technique (ADMIRE, Siemens Healthcare, Forchheim,Germany). Depending on the kernel used, an average of80.0 % of the lumen of various current stents with a diameterof 3.0 mm that are commonly used was visible. Giventhe fact that blooming artefacts obscure the stent lumencircularly, it can be assumed that CTA is a valid toolfor detection of in-stent restenosis, particularly when ofsignificant (>50 %) extent.

Stent lumen visibility varied with the individual stent type.The ISAR-STEREO-2 trial has shown that stents with athinner strut design elicit less angiographic and clinical reste-nosis than stents with thicker struts [12]. This finding hascaused the trend to thinner stent strut designs. Through thedecrease in thickness, decreased blooming artefacts are to beexpected. However, in order to increase the structural integrityand the visibility of the stent during coronary stent implanta-tion, Boston Scientific has employed a platinum-chromiumalloy for their Omega® stent [13]. The platinum-chromiumT

able2

Meanvisiblelumen

diam

eters,in-stent

attenuationandnoiseof

thedifferentk

ernelsandorientations.V

aluesprovided

asmedian,interquartile

rangeandrange

Bv40

Bv49

Bv59(3)

Bv59(4)

0°90°

Visiblediam

eter

[%]

53.3

(48.9-56.7;3

1-100)

56.7(52.2-65.6;4

1-100)

67.9(64.4-75.6;5

2-100)

70.0(66.7-

75.8;5

0-100)

80.0(75.6-82.8;7

2-100)

Attenuation[H

U]

582(533

-636;

456-1094)

468(432

-506;

370-1065)

462(438

-501;

279-544)

463(415

-491;

305-550)

416(407

-436;

342-476)

Noise

12.0

(11.5-12.4;8

.3-13.6)

21.5(20.5-22.0;1

4.7-25.5)

44.3(43.4-46.0;3

1.7-57.1)

35.9(28.3-37.0;2

5.6-46.2)

46.0(45.0-48.0;4

1.0-80.0)

2958 Eur Radiol (2014) 24:2953–2961

alloy is a denser material and thus negates the effects of thereduced stent thickness regarding the stent lumen visibility.

Our findings have shown a 67–72% lumen visibility using theBv59 kernel for this stent. New types of stents are the

Fig. 4 Comparison of the four reconstruction kernels. Stent: MedtronicResolute RX® (3.0 mm)

Fig. 5 Influence of stent diameter on lumen visibility [Bv59(4) kernel, 0°orientation to the z-axis]

Table 3 Mean visible stent lu-men diameters and attenuationusing the Bv59 kernel, strengthlevel 4

Stent name Diameter Visible lumen [%] Attenuation [HU] Δ Attenuation [HU]

0° 90° 0° 90° 0° 90°

Absorb BVS 3.00 100 100 477 342 -8 -6

JOSTENT Graftmaster 3.00 62 77 483 424 -21 92

Multilink Vision 3.00 79 83 491 411 -17 58

Xience Pro 3.00 76 80 521 407 25 54

Orsiro 2.25 50 308 -177

Orsiro 2.50 64 417 -92

Orsiro 2.75 68 435 -81

Orsiro 3.00 74 73 451 389 -48 36

Orsiro 3.50 68 500 -2

Orsiro 4.00 76 469 -30

PRO-Kinetic Energy 3.00 70 76 524 407 18 68

Omega 3.00 67 72 541 445 39 61

Taxus Liberté 3.00 77 82 483 436 -10 53

Coroflex Blue 3.00 72 81 485 416 -10 73

Cypher 3.00 81 83 365 385 -144 46

Resolute RX 3.00 74 82 447 476 -55 127

Resolute Integrity RX 3.50 69 405 -82

Resolute Integrity RX 4.00 81 474 -9

Integrity RX 2.25 53 305 -204

Integrity BMS 2.50 63 448 -165

Integrity RX 2.75 62 401 -108

Integrity RX 3.00 67 77 550 430 -49 58

Integrity RX 3.50 69 415 -90

Integrity RX 4.00 76 497 -11

Genous 3.00 72 78 410 414 -98 105

Kaname 3.00 67 73 463 447 -36 124

Nobori 3.00 74 83 448 420 -39 73

Eur Radiol (2014) 24:2953–2961 2959

bioresorbable scaffolds. As a representative, we included theAbsorb BVS® by Abbott in this study. This stent is construct-ed of the poly-L-lactic acid polymer (PLLA) and resorbs withtime [14, 15]. Apart from the platinum markers, this stentcauses no blooming artefacts; in fact, it is not even visible inthe CT angiography, providing a 100 % lumen visibility.

The Graftmaster® stent is constructed of two stainless-steelstents with a PTFE graft material wrapped between the twostents [16]. This construction causes high x-ray attenuationand thus a visible diameter of 62 % at an orientation of 0° tothe z-axis, the lowest lumen visibility from all stents includedwith a diameter of 3.0 mm.

The stent lumen visibility also varies with the diameter ofthe stent. Carrabba et al. performed a meta-analysis of trialsthat employed 64-MSCT in detecting significant in-stent re-stenosis. They suggested stent lumen evaluation with 64-MSCT should be restricted to stents with a diameter largerthan 3 mm [17]. A meta-analysis by Sun et al., reviewing 64-MSCT for detecting in-stent restenosis, came to the sameconclusion [18]. A very recently published study by Andréet al. employed a modern 256-MSCTand a special reconstruc-tion kernel for coronary stent lumen visualisation. They con-cluded that this setting provides a good lumen visualisation ofcoronary stents with a diameter >3 mm. They deemed assess-ment of stents with a diameter of 3 mm feasible but did notrecommend the clinical evaluation of smaller stents [19]. As inthe former study, the darker rim at the vicinity of the stentstruts (Fig. 3) was added to the measured visible lumendiameter. This artefact might be due to the high-density dif-ference of 2,000 HU (metal) and 350 HU (contrast) or -70 HU(oil emulsion) within the closest proximity. Despite the localsignal dropout it did not hamper the overall image qualitysubstantially in our in vitro study setting because of its rathersmall calibre. Yet, it may impair the diagnosis of in-stentrestenosis particularly in small stents with concentric stenosisand should be further addressed in future in vivo studies. Weincluded the Medtronic Integrity® and the Biotronik Orsiro®stents in all available diameters (2.25, 2.5, 2.75, 3.0, 3.5 and4.0 mm) in this study. Stent lumen visibility is higher in stentswith a larger diameter, but even in the smallest diameter stentsa lumen visibility of 50-53% has been achieved in this in vitrostudy. This finding is important, because it suggests that evenstents with a smaller diameter can be evaluated regardingrelevant in-stent restenosis. However, increasing attenuationin smaller stents might impair the detection of in-stent reste-nosis as differentiation between thrombosis and contrast-enhanced blood may be substantially affected. Future in vivostudies should therefore point to this particular issue.

Interestingly, orientation of the stent towards the gantry hada significant effect regarding both visible stent lumen diameterand in-stent attenuation. The mechanism leading to the dis-crepant performance between different imaging angles is pre-sumably based on the processing of the original image data by

the 3D application software used for image reconstruction.When the stent is orientated at 0°, also the axial reformation isorientated parallel to the scanner’s z-axis. When the stentorientation is 90° to the scanner’s z-axis, so is the angle ofthe axial reformation in these stents. The 3D applicationsoftware now presumably performs interpolation to thereformatted images depending on their orientation. This sub-sequently reduces isotropy of the images and eventually leadsto both reduced sharpness and image noise at an orientation of0° compared to an orientation of 90°.

Due to the novelty of both the CT system and reconstruc-tion technique used, the individual benefit of each of theseaspects cannot be definitely discriminated.With an increase indetector channels as well as a new iterative reconstructiontechnique and a new reconstruction kernel both hardwareand software may be considered responsible for the benefitin stent lumen visibility. At an orientation of 0°, the Bv59 (4)kernel provided a 15.6 percentage point better lumen visibilityin 3.0 mm stents compared to the Bv49 kernel, which isequivalent to the previously used B46f kernel. This improve-ment was accompanied by a 67 % increase in image noise,which was rated as diagnostically acceptable. To eventuallydiscriminate between hardware- and software-related benefits,further studies evaluating the same stents using the samekernels on both second- and third-generation CT will berequired.

This study has several limitations. As tubes were not al-ways available in sizes exactly matching the intended stentdiameter, some of them were carefully heated. To minimise apossible effect of the shrinking process on stent lumen diam-eter, this procedure was performed during the stent’s ballooninflation. Although no visible differences in stent lumen di-ameter were observed, a possible effect on the stent lumendiameter due to heating of the tubes cannot be totally exclud-ed. Although increasing temporal resolution of recent dual-source CT has substantially reduced motion artefacts, theycannot be totally neglected in patients with higher heart ratesor arrhythmia. Cardiac movement, however, was not simulat-ed in our current setup. Second, measurement of visible lumendiameter was performed at a fixed windowwidth of 1,500 HUat a centre of 300 HU by a single observer. Therefore, mea-surement errors and possibly different results for visible lumendiameter using different window settings have to be consid-ered. Furthermore, as the current study is an in vitro study, thepossible influence of the surrounding torso on stent lumenvisualisation in vivo has to be taken into account.

Conclusion

With an average lumen visibility of up to 80 % in metal stentsand 100 % in the bioresorbable stents, CTA performed with athird-generation CT scanner may be a valid alternative for

2960 Eur Radiol (2014) 24:2953–2961

detecting in-stent restenosis. Patient studies are required tovalidate this in vitro finding for translation into clinical practice.

Acknowledgments This work was supported by the ComprehensiveHeart Failure Center Würzburg. The scientific guarantor of this publica-tion is Tobias Gassenmaier. The authors of this manuscript declarerelationships with the following companies: Thomas Allmendinger andThomas Flohr: employees of Siemens Healthcare. The authors state thatthis work has not received any funding. No complex statistical methodswere necessary for this article. Institutional Review Board approval wasnot required because no human subjects were involved in this in vitrostudy. Methodology: experimental.

References

1. Buuren F (2010) 25. Bericht über die Leistungszahlen derHerzkatheterlabore in der Bundesrepublik Deutschland. DerKardiol 4:502–508

2. Dangas GD, Claessen BE, Caixeta A et al (2010) In-stent restenosisin the drug-eluting stent era. J Am Coll Cardiol 56:1897–1907

3. Peberdy MA, Donnino MW, Callaway CW et al (2013) Impact ofpercutaneous coronary intervention performance reporting on cardiacresuscitation centers: a scientific statement from the American HeartAssociation. Circulation 128:762–773

4. Maintz D, Seifarth H, Raupach R et al (2006) 64-slice multidetectorcoronary CT angiography: in vitro evaluation of 68 different stents.Eur Radiol 16:818–826

5. Maintz D, Burg MC, Seifarth H et al (2009) Update on mul-tidetector coronary CT angiography of coronary stents: in vitroevaluation of 29 different stent types with dual-source CT. EurRadiol 19:42–49

6. Seifarth H, OzgünM, Raupach R et al (2006) 64- Versus 16-slice CTangiography for coronary artery stent assessment: in vitro experience.Invest Radiol 41:22–27

7. Pugliese F, Weustink AC, Van Mieghem C et al (2008) Dual sourcecoronary computed tomography angiography for detecting in-stentrestenosis. Heart 94:848–854

8. Krüger S, Mahnken AH, Sinha AM et al (2003) Multislice spiralcomputed tomography for the detection of coronary stent restenosisand patency. Int J Cardiol 89:167–172

9. Donati OF, Burg MC, Desbiolles L et al (2010) High-pitch 128-slicedual-source CT for the assessment of coronary stents in a phantommodel. Acad Radiol 17:1366–1374

10. Rief M, Zimmermann E, Stenzel F et al (2013) Computed tomogra-phy angiography and myocardial computed tomography perfusion inpatients with coronary stents: prospective intraindividual comparisonwith conventional coronary angiography. J Am Coll Cardiol 62:1476–1485

11. Seifarth H, Raupach R, Schaller S et al (2005) Assessment of coro-nary artery stents using 16-slice MDCT angiography: evaluation of adedicated reconstruction kernel and a noise reduction filter. EurRadiol 15:721–726

12. Pache J, Kastrati A, Mehilli J et al (2003) Intracoronary stenting andangiographic results: strut thickness effect on restenosis outcome(ISAR-STEREO-2) trial. J Am Coll Cardiol 41:1283–1288

13. Boston Scientific: OMEGATM Platinum Chromium Coronary StentSystem. http://www.bostonscientific-international.com/Device.bsci?page=HCP_Overview&navRelId=1000.1003&method=DevDetailHCP&id=10144661&pageDisclaimer=Disclaimer.ProductPage,Disclaimer.ReservedForMedProfs

14. Ormiston JA, Serruys PW, Regar E et al (2008) A bioabsorbableeverolimus-eluting coronary stent system for patients with single de-novo coronary artery lesions (ABSORB): a prospective open-labeltrial. Lancet 371:899–907

15. Nieman K, Serruys PW, Onuma Y et al (2013) Multislice computedtomography angiography for noninvasive assessment of the 18-Month performance of a novel radiolucent bioresorbable vascularscaffolding device: the ABSORB Trial (a clinical evaluation of thebioabsorbable everolimus eluting coronary stent). J Am Coll Cardiol62:1813–1814

16. Abbott Vascular: JOSTENT GRAFTMASTER with HYDREXCoating System INSTRUCTIONS FOR USE. http://www.abbottvascular.com/static/cms_workspace/pdf/ifu/humanitarian_use/IFU_Jostent_Graftmaster_OTW.pdf

17. Carrabba N, Schuijf JD, deGraaf FR et al (2010) Diagnostic accuracyof 64-slice computed tomography coronary angiography for thedetection of in-stent restenosis: a meta-analysis. J Nucl Cardiol 17:470–478

18. Sun Z, Almutairi AMD (2010) Diagnostic accuracy of 64 multisliceCT angiography in the assessment of coronary in-stent restenosis: ameta-analysis. Eur J Radiol 73:266–273

19. André F, Müller D, Korosoglou G et al (2014) In-vitro assessment ofcoronary artery stents in 256-multislice computed tomographyangiography. BMC Res Notes 7:38

Eur Radiol (2014) 24:2953–2961 2961

http://www.bostonscientific-international.com/Device.bsci?page=HCP_Overview&navRelId=1000.1003&method=DevDetailHCP&id=10144661&pageDisclaimer=Disclaimer.ProductPage,Disclaimer.ReservedForMedProfs




http://www.abbottvascular.com/static/cms_workspace/pdf/ifu/humanitarian_use/IFU_Jostent_Graftmaster_OTW.pdf



Documents

Next generation coronary CT angiography: in vitro evaluation of 27 coronary stents