Original Article
487
Quantification of Internal Carotid Artery Stenosis with 3D Ultrasound Angiography Quantifizierung von Stenosen der Arteria carotis interna mit 3D-Ultraschallangiografie Authors
J. O. Pelz1, A. Weinreich1, D. Fritzsch2, D. Saur1
Affiliations
1
Department of Neurology, University of Leipzig, Germany Department of Neuroradiology, University of Leipzig, Germany
Key words
Abstract
Zusammenfassung
"
!
!
Purpose: The aim of this study was to evaluate a new method of three-dimensional ultrasound (3D-US) angiography of carotid vessels including 3D-US quantification of internal carotid artery (ICA) stenosis (ICAS). Materials and Methods: Two neurologists performed native ultrasound scans of the brain-supplying carotid arteries in 73 probands (including 22 patients with 25 cases of ICAS) using a Toshiba Aplio 500 ultrasound machine equipped with the 3 D application of Curefab CS. Additionally, 25 probands underwent contrast-enhanced magnetic resonance angiography (CE-MRA) of the neck vessels. Results: The mean length of the proximal ICA was 32.1 mm ± 9.8 mm & 31.3 mm ± 9.0 mm (each n = 97). The interrater reliability (intraclass correlation, ICC) was 0.75 ± 0.23 for the common carotid artery (CCA, n = 90) and 0.78 ± 0.21 (n = 92) for the ICA. The intermethod agreement between 3D-US angiography and CE-MRA was 0.67 ± 0.19 (n = 45) & 0.66 ± 0.19 (n = 44) for the CCA and 0.79 ± 0.17 (n = 47) & 0.75 ± 0.19 (n = 46) for the ICA. Quantification of ICAS applying 3D-US in comparison to 2D color-coded duplexsonography (2D-CDS) showed a moderate to good intermethod agreement both by the Bland and Altman analysis and by ICC (0.8 & 0.72; with each p < 0.001). The interrater reliability for quantification of ICAS was 0.79; p < 0.001. Conclusion: Native 3D-US angiography of carotid vessels shows good interrater and intermethod agreement in comparison to CE-MRA. Together with a moderate to good intermethod and interrater agreement in the quantification of ICAS, when compared to 2D-CDS, 3D-US angiography is thus a promising complementary imaging technique for carotid artery disease.
Ziel: In dieser Studie wurde eine neue Methode der drei-dimensionalen Ultraschallangiografie (3D-US) der hirnversorgenden Halsgefäße einschließlich der Quantifizierung von Stenosen der Arteria carotis interna (ICAS) evaluiert. Material und Methoden: Zwei Neurologen untersuchten in 73 Probanden (davon 22 Patienten mit 25 ICAS) die extrakraniellen hirnversorgenden Gefäße mit einem Toshiba Aplio 500 Ultraschallgerät, das mit einer 3D-Anwendung von Curefab CS ausgerüstet war. Bei 25 Probanden wurde zusätzlich eine Kontrastmittel-Magnetresonanzangiografie der Halsgefäße (CE-MRA) durchgeführt. Ergebnisse: Die mittlere Länge der proximalen Arteria carotis interna (ICA; jeweils n = 97) betrug 32.1 mm ± 9.8 mm bzw. 31.3 mm ± 9.0 mm. Die Interrater Übereinstimmung (Intraklassen-Korrelationskoeffizient, ICC) betrug 0.75 ± 0.23 für die Arteria carotis communis (CCA; n = 90) und 0.78 ± 0.21 (n = 92) für die ICA. Die Intermethoden Übereinstimmung (ICC) zwischen der 3D-US Angiografie und der CE-MRA lag bei 0.67 ± 0.19 (n = 45) bzw. 0.66 ± 0.19 (n = 44) für die CCA und bei 0.79 ± 0.17 (n = 47) bzw. 0.75 ± 0.19 (n = 46) für die ICA. Im Vergleich des 3D-US mit der konventionellen 2D-Farbduplexsonografie fand sich bei der Quantifizierung von ICAS eine moderate bis gute Intermethoden Übereinstimmung sowohl in der Bland und Altman Analyse als auch mittels ICC (0.8 bzw. 0.72; p jeweils < 0.001). Die Interrater Übereinstimmung für die Stenosenquantifizierung lag bei 0.79; p < 0,001. Schlussfolgerung: Im Vergleich zur CE-MRA zeigt die native 3D-US Angiografie der extrakraniellen hirnversorgenden Gefäße eine gute Interrater und Intermethoden Übereinstimmung. Zusammen mit der moderaten bis guten Intermethoden und Interrater Übereinstimmung bei
● carotid arteries ● angiography ● 3 D ultrasound ● internal carotid " " "
artery stenosis
● ultrasound color Doppler "
received accepted
24.8.2014 24.11.2014
Bibliography DOI http://dx.doi.org/ 10.1055/s-0034-1398749 Published online: January 21, 2015 Ultraschall in Med 2015; 36: 487–493 © Georg Thieme Verlag KG Stuttgart · New York · ISSN 0172-4614 Correspondence Dr. Johann Otto Pelz Deparment of Neurology, University of Leipzig Liebigstraße 20 04103 Leipzig Germany Tel.: ++ 49/3 41/9 72 42 00 johann.pelz@medizin. uni-leipzig.de
Pelz JO et al. Quantification of Internal … Ultraschall in Med 2015; 36: 487–493
Downloaded by: Chinese University of Hong Kong. Copyrighted material.
2
Original Article
der Quantifizierung von ICAS, stellt die 3D-US Angiografie somit eine vielversprechende, komplementäre Bildgebungsmethode zur Untersuchung von Carotisstenosen dar.
Introduction
Materials and methods
!
!
Up to 20 % of ischemic strokes and transient ischemic attacks are caused by large-artery atherosclerosis, i. e., significant (more than 50 %) stenosis or even occlusion of mainly the internal carotid artery (ICA) [1]. Given that the risk of ipsilateral ischemic stroke directly depends on the stenotic value of internal carotid artery stenosis (ICAS) [2], quantification of ICAS as accurately as possible is essential. Applying the flow-based multi-parametric German “DEGUM (Deutsche Gesellschaft für Ultraschall in der Medizin) ultrasound criteria” [3], two-dimensional color-coded duplex sonography (2 D-CDS) has proven to be a noninvasive, cheap and bedside method for the detection and grading of ICAS with good sensitivity and specificity [4] and is recommended as the method of choice for the detection and quantification of ICAS [5]. However, there are still cases where a second imaging modality is desirable or even mandatory. Although digital subtraction angiography (DSA) is still considered the gold standard for the quantification of ICAS [4, 5], it is scarcely performed in the clinical routine mostly due to its invasive character and potential complications (6). Therefore, following the S3-guidelines, contrast-enhanced computed tomography angiography (CTA) or magnetic resonance angiography (CE-MRA) is recommended if there is reasonable doubt in the 2 D-CDS grading of ICAS (5). This might be the case for instance if there is chronic heart failure, a second stenosis either proximal or distal to ICAS or if there is a high-grade stenosis/occlusion of the contralateral ICA. In these conditions the flow-based ultrasound criteria might overestimate or underestimate the degree of ICAS [7]. Although routinely performed in everyday clinical practice, CTA and CE-MRA have some important limitations like exposure to contrast agent and radiation (CTA) or are absolutely contraindicated like CE-MRA in patients with a heart pacemaker or CTA in patients with chronic renal failure or manifest hyperthyreosis. Preliminary studies on 3 D ultrasound have demonstrated encouraging results (8 – 13). However, they suffered from some serious restrictions, e. g. the overall low quality of vessel reconstruction and the use of diameter-based procedures for the graduation of ICAS instead of the more reliable cross-sectional area (CSA)based measurements. Furthermore, 3 D vessel reconstruction turned out to be time-consuming. Therefore, 3 D ultrasound has so far failed to be able to be used in everyday practice. The aim of the present study is to perform and evaluate a new generation of native 3 D ultrasound angiography of the brainsupplying carotid arteries (common carotid artery [CCA] and ICA) in comparison to CE-MRA. Furthermore, we assess intermethod agreement between 3 D ultrasound and 2 D-CDS quantification according to the “DEGUM ultrasound criteria” in ICAS of different stenotic values. Since all ultrasound scans were done by two neurologists, we will provide interrater reliability for all analyses of the study.
Study population
Pelz JO et al. Quantification of Internal … Ultraschall in Med 2015; 36: 487–493
The study was approved by the local ethics committee and all 73 participants gave their written and informed consent. According to our study aims, we examined three different cohorts: Cohort 1 consisted of 25 healthy volunteers (12 male; mean age 29 years, age range 24 – 37), who all underwent 3 D ultrasound angiography of their right and left carotid vessels (CCA and ICA) in order to evaluate the interrater agreement of native 3 D ultrasound. Cohort 2 consisted of 25 vascular healthy patients from the department of neurology (6 male; mean age 42 years, age range 19 – 72) who underwent both 3 D ultrasound and CE-MRA of the neck vessels in order to calculate intermethod agreement. CE-MRA was done in addition to standard sequences of their diagnostic MRI. Cohort 3 consisted of 22 patients with 25 cases of ICAS (3 bilateral, 4 symptomatic, 21 asymptomatic; 14 male; mean age 71 years, age range 53 – 84) who were recruited from our stroke unit or neurovascular outpatient clinic. In these patients 2 D-CDS and 3 D ultrasound of the stenotic ICA were performed in order to assess intermethod agreement. Finally, a 30-year old female patient with chronic occlusion of the left ICA due to dissection was examined for illustration purposes. All 3 D ultrasound examinations were done by two neurologists (JP, examiner 1 and AW, examiner 2), except for 2 patients of cohort 2 and one patient with bilateral asymptomatic ICAS of cohort 3 who were examined only once.
Three-dimensional B-mode ultrasound scanning 3 D ultrasound scans on the basis of B-mode images were performed using a Toshiba Aplio 500 (Toshiba Medical Systems GmbH, Neuss, Germany) equipped with a linear transducer (PLT-1204BT) set at 13 MHz. For three-dimensional sonography, this commercial ultrasound system was attached to the Curefab CS system (Curefab Technologies GmbH, Munich, Germany). Curefab CS comprised a freehand magnetic field tracking system and a workstation equipped with special software (Curefab CS, version 1.91). Briefly, two sensors which were mounted on the linear transducer traced its longitudinal movement and tilting relative to the 3 orthogonal room axes within a weak magnetic field. Using the spatial and temporal information, this setting enabled exact positioning of common two-dimensional native B-mode images, which could then be concatenated to a virtual 3 D stack. Within this 3 D stack, carotid arteries were reconstructed semiautomatically. First, the vessel’s lumen was marked with a centerline. Then, the contour of the vessel was detected automatically and, if necessary, manually corrected. Practically, all participants were lying in a supine position and were asked not to breathe or swallow during the scan. Subsequently, ultrasound scanning started cranial to the clavicle and medial to the sternocleidomastoid muscle, and the transducer was moved towards the mandible at a speed of about 1 cm per second in an axial direction. The whole scan took about 7 to 10 seconds while about 250 to 360 single B-mode images with a slice thickness of about 120 µm were recorded (36 images per second). In total, each examiner performed 3 ultrasound scans
Downloaded by: Chinese University of Hong Kong. Copyrighted material.
488
Original Article
of the respective carotid arteries, and the best scan was used for post-processing.
489
the quantification of ICAS was not done in real time but after the scanning offline, altogether taking no longer than 5 minutes for each case.
Analysis of 3D-reconstructed carotid arteries The 25 patients of cohort 2 underwent CE-MRA of the neck vessels on a 3 T MRI scanner (Trio, Siemens, Erlangen, Germany) with 0.1 mmol/kg body weight of Gadobutrol (Gadovist ®, Bayer Vital GmbH, Leverkusen, Germany) used as the contrast agent. Corresponding to 3 D ultrasound analyses, the CSA was measured every 2 mm in the vessel’s course. The ICA was measured in a distal direction beginning 2 mm distal to the carotid bifurcation, while the CCA was measured in a proximal direction from 10 mm to 40 mm proximal to the carotid bifurcation.
Quantification of internal carotid artery stenosis
Statistical analysis
First, the cases of ICAS of cohort 3 were graded by a neurologist (JP or AW) with 2 D-CDS applying the multi-parametric German “DEGUM ultrasound criteria” [3]. Then, for blinding both examiners, a combination of characters and numbers was assigned to every patient’s 3 D ultrasound dataset. Post-processing of 3 D ultrasound data started after all ultrasound examinations were completed. This way, it could be ensured that during post-processing of the 3 D ultrasound data, both examiners were unaware of the results of 2 D-CDS. Within the reconstructed ICA, the smallest luminal CSA within the stenosis as well as the normal luminal CSA in the distal course of the vessel were obtained. ICAS was then quantified by calculating the distal CSA reduction percentage, according to the North American Symptomatic Carotid Endarterectomy Trial (NASCET) [14]. Due to the blinding procedure,
Statistical analyses were performed with SPSS version 20.0 (IBM Corporation; New York, NY, USA). Intraclass correlation (ICC; consistency mode or absolute mode) was used to calculate the interrater reliability between examiner 1 and examiner 2, as well as between examiner 1 or examiner 2 and CE-MRA. Briefly, for each vessel, all CSAs measured in the vessel’s course were correlated with the corresponding CSAs of the second examiner or " Fig. 1). The mean ICC and the CSAs obtained from CE-MRA (● standard deviation across vessels were used for describing interrater reliability. Visualization and description of intermethod agreement between 2 D-CDS and 3 D ultrasound quantification of ICAS was achieved by a Bland and Altman analysis (MedCalc® Version 13.0; MedCalc Software) which, in contrast to correlational
Fig. 1 3 D ultrasound of the carotid vessels was performed by two examiners as shown by way of example on the left (examiner 1) and on the right (examiner 2). Within the reconstructed internal carotid artery (ICA), the cross-sectional area (CSA) was measured every 2 mm perpendicular to the vessel’s course with the bifurcation (asterisk) serving as a landmark. In this example the total length of the ICA was 55 mm (left) and 51 mm (right) resulting in 25 CSA pairs for calculating the intraclass correlation coefficient (here: ICC 0.98). CCA common carotid artery. Scale bar each 1 cm.
Abb. 1 Der 3D-Ultraschall der Halsgefäße erfolgte durch zwei Untersucher, wie es beispielhaft rechts (Untersucher 1) und links (Untersucher 2) gezeigt wird. Innerhalb der rekonstruierten Arteria carotis interna (ICA) wurde die Querschnittsfläche (CSA) alle 2 mm senkrecht zum Gefäßverlauf bestimmt, wobei die Bifurkation als Referenzpunkt diente (Stern). In diesem Beispiel beträgt die Länge der ICA 55 mm (links) beziehungsweise 51 mm (rechts). Anhand von insgesamt 25 CSA Paaren konnte so der Intraklassenkoeffizient (hier ICC 0.98) als Maß für die Übereinstimmung zwischen den Untersuchern berechnet werden. CCA Arteria carotis communis. Maßstab jeweils 1 cm.
Pelz JO et al. Quantification of Internal … Ultraschall in Med 2015; 36: 487–493
Downloaded by: Chinese University of Hong Kong. Copyrighted material.
Contrast-enhanced magnetic resonance angiography
Both ultrasound examiners reconstructed and analyzed his/her scans with the carotid bifurcation serving as a landmark for further measurements. Besides the total length of the reconstructable ICA, the CSA perpendicular to the vessel’s course was measured every 2 mm. The ICA was measured in a distal direction beginning 2 mm distal to the carotid bifurcation, while the CCA was measured in a proximal direction from 10 mm to 40 mm proximal to the carotid bifurcation excluding the carotid bulb from the measurements. All CSAs of one vessel were then used " Fig. 1). for calculating interrater and intermethod reliability (●
Original Article
methods, is considered to be more eligible for assessing intermethod agreement [15]. Generally, a p < 0.05 was considered as statistically significant.
mean interrater reliability was 0.75 ± 0.23 (SD) for the CCA (n = 90), while it was 0.78 ± 0.21 (SD) for the ICA (n = 92).
Results !
3 D reconstruction and interrater reliability for CCA and ICA Based on the high spatial and temporal resolution, it was possible to visualize the pulse wave as well as distinct structures of the vessel " Fig. 2). Subsequent semi-auwall like the intima-media complex (● tomatic reconstruction enabled a precise depiction not only of normal carotid vessels but also of different vascular phenomena, such as kinking or coiling of the ICA, a tapered ICA resulting from dissec" Fig. 3). Within the recontion or an ectasia of the proximal ICA (● structed vessels, parameters like average diameter, CSA or volume could be obtained at any point in the vessel’s course. The length of the reconstructable proximal ICA was measured in a total of 49 probands of cohorts 1 and 2 with 98 ICAs for each examiner. The mean length of the ICA was 32.1 mm ± 9.8 mm (standard deviation, SD) for examiner 1 (n = 97) and 31.3 mm ± 9.0 mm for examiner 2 (n = 97). In the one missing case the carotid bifurcation was located cranially, directly under the mandible, making it impossible to insonate even the proximal ICA. Generally, the length of the reconstructable ICA showed a great variance from a minimum of 5 mm to a maximum of 61.5 mm, which is explained by the variable distance from the carotid bifurcation to the mandible. Forty-eight probands of cohort 1 and 2 with 96 carotid vessels were examined by both neurologists. Since the use of ICC for calculating interrater reliability required a vessel length of at least 8 mm, i. e., at least 4 measuring points, a total of 4 ICAs and 6 CCAs had to be excluded. Applying ICC (consistency mode), the
Fig. 3 3 D ultrasound angiography of different vessel morphologies and pathologies: a Common (CCA), external (ECA) and internal (ICA) carotid artery of a healthy volunteer. b Elongated ICA with “kinking”. c Distal “coiling” of extracranial ICA – visualization by 3 D ultrasound (left) and by contrastenhanced MRA (right). d Occlusion of ICA due to dissection. e Ectasia of proximal ICA after thromboendarterectomy. Scale bar each 1 cm.
Pelz JO et al. Quantification of Internal … Ultraschall in Med 2015; 36: 487–493
Fig. 2 Reconstructed 2 D view of primarily axial ultrasound scans showing the pulse wave running through the common carotid artery and deflecting the vessel wall. Scale bar 1 cm. Abb. 2 Rekonstruierte 2D-Darstellung der Pulswelle, wie sie durch die Arteria carotis communis läuft und die Gefäßwand entsprechend auslenkt. Maßstab 1 cm.
Abb. 3 3D-Ultraschall Angiografie unterschiedlicher Gefäßverläufe und -pathologien: a Arteria carotis communis (CCA), externa (ECA) und interna (ICA) eines gesunden Probanden. b Elongierte ICA mit “kinking”. c Distales „coiling“ der extrakraniellen ICA im 3D-Ultraschall (links) und mit Kontrastmittel-MRA (rechts). d ICA-Verschluss aufgrund einer Dissektion. e Ektasie der proximalen ICA nach Thrombendarteriektomie. Maßstab jeweils 1 cm.
Downloaded by: Chinese University of Hong Kong. Copyrighted material.
490
Original Article
Comparison of 3 D ultrasound with CE-MRA Each neurologist examined 24 patients (with 48 carotid vessels) who also underwent CE-MRA (cohort 2). Again, calculation of inter-
491
method reliability required a minimum vessel length (ultrasound and CE-MRA) of at least 8 mm, leading to the exclusion of 3 CCAs and 1 ICA for examiner 1 and 4 CCAs and 2 ICAs for examiner 2. Overall, comparison of 3 D ultrasound angiography of carotid vessels with CE-MRA showed a moderate to good agreement. The mean ICC (consistency mode) for the CCA was 0.67 ± 0.19 (SD, n = 45) for examiner 1, and 0.66 ± 0.19 (n = 44) for examiner 2. Concerning the ICA, the mean ICC (consistency mode) was 0.79 ± 0.17 (n = 47) for examiner 1 and 0.75 ± 0.19 (n = 46) for examiner 2.
Fig. 4 A very high-grade (80 % according to DEGUM NASCET) eccentric stenosis of the internal carotid artery. a Coronal view and b sagittal view obtained by a 90° clockwise rotation. Inlets show corresponding native ultrasound scans exhibiting an irregular plaque texture which ranges from echorich to echolucent. Origin of external carotid artery (not shown) is marked by an asterisk. Scale bar 1 cm. Abb. 4 Sehr hochgradige (80 % nach DEGUM NASCET) exzentrische Stenose der Arteria carotis interna. a koronare und b sagittale Darstellung nach 90° Drehung im Uhrzeigersinn. Die kleinen Bilder zeigen entsprechende native Ultraschallaufnahmen in denen gut die irreguläre Plaquetextur von echoreich bis echoarm zu erkennen ist. Der Ursprung der Arteria carotis externa (nicht dargestellt) ist durch ein Sternchen markiert. Maßstab 1 cm.
In cohort 3, 22 patients with 25 cases of ICAS were examined with 2 D and 3 D ultrasound. According to 2 D-CDS and the German “DEGUM ultrasound criteria”, 11 of 25 cases of ICAS were graded as ≥ 70 % or high-grade. Three-dimensional ultrasound scanning with subsequent reconstruction and evaluation of ICAS was possible in 21 of 25 cases of ICAS (84 %, examiner 1) and in 19 of 23 cases of ICAS (83 %, examiner 2). The 4 missing cases of ICAS could not be reconstructed due to calcified plaques resulting in extensive acoustic shadowing. The reconstructed ICAS could be rotated in any direction, allowing also a qualitative analysis of (eccentric) ste" Fig. 4). Intermethod agreement between 2 D-CDS and 3 D nosis (● ultrasound was 0.8 with p < 0.001 (ICC, consistency mode) for examiner 1 and 0.72 with p < 0.001 for examiner 2. Bland and Altman analysis showed moderate intermethod agreement for both examiners (examiner 1: bias – 0.08 %; 1.96 SD limits of agreement + 11.6 % and – 27.6 % – examiner 2: bias – 0.06 %; 1.96 SD limits of " Fig. 5). Altogether, 18 cases of agreement + 19.2 % and – 32.3 %; ● ICAS were graded by both examiners, yielding an interrater reliability for 3 D ultrasound quantification of 0.79 with p < 0.001 (ICC, absolute mode; n = 18). In 9 of 18 (50 %) cases of ICAS, both examiners assessed the same stenotic value while in 6 of 18 (33 %) cases of ICAS, a difference of ≤ 10 % was found. The remaining 3 cases of ICAS (17 %) showed a difference in stenotic value of ≤ 20 %.
Fig. 5 Intermethod agreement between 3 D ultrasound and 2 D color-coded duplexsonography for the quantification of internal carotid artery stenosis. Bland and Altman analyses for a examiner 1 and b examiner 2 show moderate to good agreement. Abb. 5 Intermethoden-Vergleich der Quantifizierung von Stenosen der Arteria carotis interna zwischen 3D-Ultraschall und 2D-Farbduplexsonografie. Die Bland und Altman Analysen für a Untersucher 1 und b Untersucher 2 zeigen jeweils eine moderate bis gute Übereinstimmung.
Pelz JO et al. Quantification of Internal … Ultraschall in Med 2015; 36: 487–493
Downloaded by: Chinese University of Hong Kong. Copyrighted material.
Quantification of ICA stenosis by 3 D ultrasound and comparison with 2 D-CDS
Original Article
Discussion !
In contrast to 2 D-CDS of extracranial carotid vessels, which is routinely performed in everyday clinical practice, so far, 3 D ultrasound angiography has only played a minor preclinical role as a diagnostic tool. However, our results demonstrate that, owing to major recent technical developments, 3 D ultrasound now allows reconstruction of neck vessels with high quality and reliability. In our study 3 D ultrasound angiography of extracranial carotid vessels could visualize the ICA and CCA with good interrater reliability and moderate to good intermethod agreement when compared to CE-MRA. Most importantly, the quantification of ICAS by 3 D ultrasound also yielded good results in comparison to common 2 D-CDS. Moreover, full 3 D reconstruction of carotid vessels with free rotation allows surgeons and interventionalists to study course and morphology of ICAS in detail which is further facilitated by the high spatial resolution of 3 D ultrasound " Fig. 2, 4). Starting at the carotid bifurcation, the proximal ICA (● could be depicted – on average – over a distance of more than 30 mm. This appears to be sufficient for a profound inspection of its origin, the preferential site for stenosis and dissection [16]. Additionally, 3 D ultrasound angiography of neck vessels did not only show a good interrater reliability but also a moderate to good intermethod agreement when compared to CE-MRA. It is noteworthy that a better interrater and intermethod agreement for 3 D ultrasound was probably limited by arterial pulsation arti" Fig. 2). However, APAs are present facts (APA) in the 3 D data (● to a lesser degree also in CE-MRA. The high temporal resolution of 36 images per second in B-mode ultrasound compared to 3 images per second in CE-MRA thereby explains the higher vulnerability for APAs in 3 D ultrasound. In general, APAs could be resolved by electrocardiogram gating [17], which was not implemented in our study but is also not used in routinely performed CE-MRA. Swallowing and deep breathing are other conditions that might induce considerable motion artifacts in both 3 D ultrasound and CE-MRA [17]. However, unlike CE-MRA, 3 D ultrasound can easily be repeated given the short scan time of approximately 7 – 10 seconds and the possibility to check the raw data online. The high quality of 3 D ultrasound angiography of carotid vessels was further emphasized by the reliable depiction of complex vascular phenomena, such as a tapered occlusion of " Fig. 3d), which is considered to be pathognomonic for the ICA (● ICA dissection known to be a difficult ultrasound diagnosis [18]. In our patient cohort, 84 % of cases of ICAS could be examined by 3 D ultrasound. The remaining cases of ICAS could not be analyzed due to calcified plaques resulting in extensive acoustic shadowing. Yet, this is a common problem for native B-mode ultrasound of ICAS, and our results are in line with larger trials like the Asymptomatic Carotid Emboli Study (ACES), where 15.4 % of stenotic carotid plaques could not be classified as either echolucent or echogenic due to heavy calcification [19]. A promising solution for overcoming insufficient B-mode image quality in cases of extensive acoustic shadowing or echolucent stenosis could be the use of second-generation ultrasound contrast agents. Owing to a safety profile which is comparable to MRI contrast agents [20, 21], second-generation ultrasound contrast agents are nowadays routinely used in neurosonology research [e. g. 22] – even in acute ischemic stroke [23]. Quantification of ICAS by 2 D-CDS applying the “DEGUM ultrasound criteria” compared with 3 D graduation using the distal CSA reduction percentage showed moderate to good intermethod agreement in both the Bland and Altman analysis and intraclass correlation. Notably, 3 D ultra-
Pelz JO et al. Quantification of Internal … Ultraschall in Med 2015; 36: 487–493
" Fig. 5). sound overestimated ICAS by an average of about 8 % (● This overestimation of stenotic value when using CSA reduction percentage is a well-known fact owing to the second power of diameter going into the circle’s formula [24]. Taking into account that the “DEGUM ultrasound criteria” are conformed to DSA measurements which used the distal diameter reduction approach [14], our results indeed fit fairly well. Furthermore, the morphology of a stenosis has to be considered. Dodds [25] could demonstrate in vitro that flow disturbances created by an eccentric stenosis would be different from that created by a concentric one when expressing stenotic value only as diameter reduction percentage, while hemodynamic effects of eccentric as well as concentric stenoses would be the same if stenotic value was assessed as CSA reduction percentage [25]. As exemplarily shown " Fig. 4, adequate assessment of diameter reduction percenin ● tage also highly depends on the perspective (i. e., in the case of DSA on the direction of X-rays), especially in eccentric ICAS, which might easily result in underestimation of stenotic value. Restrictively, it should be noted that a high-grade ICAS with a distal diameter reduction percentage of about 70 % would be a stenotic value of 80 – 90 %, when using the CSA as a reference, which might make it difficult then to assess a further progression of stenotic value over time [7, 24]. Although this progression of asymptomatic ICAS was shown to increase the risk of ipsilateral stroke or transient ischemic attack [26], there are no recommendations in the current S3-guidelines as to what should be the best treatment in this case, probably due to the lack of clinical trials [5]. Additionally, parameters like plaque echolucency or microembolic signals alone or in combination are more suited for estimating the stroke risk of asymptomatic ICAS [19]. Furthermore, there are only 2 other studies analyzing 3 D ultrasound for the quantification of ICAS and offering a Bland and Altman analysis, which is considered to be the statistical “gold standard” for describing intermethod agreement [15]. Wessels et al. obtained a similar intermethod agreement using 3 D ultrasound (power mode) with – unlike our study – the distal diameter reduction percentage approach in comparison to 2 D-CDS [12]. But only Yao et al. examined ICAS with native 3 D ultrasound by measuring both the local diameter and local CSA reduction percentage and comparing it with DSA. They found that intermethod agreement was better when using local CSA reduction percentage [7]. However, the use of local stenotic value and a small sample size of 14 cases of ICAS hinder profound interpretation of their data. This is also the case in most of the previous 3 D ultrasound studies, in which due to poor image quality and a lack of real three-dimensional vessel reconstruction only the diameter and not the CSA reduction percentage had been measured. Additionally, astonishingly high intermethod agreement and interrater agreement are attributable to inadequate statistical methods [9 – 11, 13]. This way, by assessing morphology of ICAS and distal CSA reduction percentage in one examination step, 3 D ultrasound angiography as performed in our study seems to be most eligible for a profound examination of ICAS and therefore has the potential to overcome important methodological and clinical limitations of currently used angiographic techniques like CTA, CE-MRA or DSA. Contrary to conventional 2 D-CDS which is still regarded as an imaging modality considerably depending on the examiner’s experience, 3 D ultrasound is more objective since both the raw data and the reconstructed vessel/stenosis allow complete reevaluation and repeated measurements independent from the initial examination. This way, vascular surgeons and interventional neuroradiologists could study the anatomy of ICAS in de-
Downloaded by: Chinese University of Hong Kong. Copyrighted material.
492
tail offline facilitating the choice of the appropriate revascularization procedure (operation or stenting) as well as the preparation of the respective procedure itself. Moreover, given the amended certification criteria of regional and national stroke units in Germany, which demand that 85 – 90 % of patients with ischemic stroke should routinely be examined by 2 D-CDS [27], both methods could be applied in a complementary manner in one bedside examination, thus minimizing time effort. Such an approach of starting the examination of carotid vessels in ischemic stroke or transient ischemic attack by the combination of 2 D-CDS and confirming 3 D ultrasound would also be cost-saving since an expensive and potentially harmful imaging modality like DSA and – with respect to costs – to a lesser extent also CE-MRA and CTA would be reserved for cases where 2 D and 3 D ultrasound were not conclusive [28, 29]. In our study, the gold standard was 2 D-CDS instead of DSA. However, nowadays, 2 D-CDS could be considered the clinical standard [5] and an invasive diagnostic tool like DSA would have exposed patients to an inappropriate risk [6]. Another potential limitation of our study is that only 4 of the 22 patients with cases of ICAS presented with an acute stroke as a result of a symptomatic high-grade ICAS. Because patients with symptomatic ICAS are in particular need of a second imaging modality before undergoing intervention, future 3 D ultrasound studies should focus on this study population. However, we are optimistic that the easy acquisition of 3 D ultrasound data will be accomplished even in acute stroke patients. In conclusion, we propose that complementary to 2 D-CDS, 3 D ultrasound is a promising bed-side diagnostic tool for the examination of the carotid vessels, particularly for the quantification of ICAS. Together with its time and cost-saving aspects, 3 D ultrasound has the potential to replace X-ray-based angiographic techniques in this indication.
Acknowledgment !
The authors wish to thank Prof. Volker Keim for giving organisational and technical support.
References 01 Grau AJ, Weimar C, Buggle F et al. Risk factors, outcome and treatment in subtypes of ischemic stroke: The German Stroke Data Bank. Stroke 2001; 32: 2559 – 2566 02 Nicolaides AN, Kakkos SK, Griffin M et al. Severity of asymptomatic carotid stenosis and risk of ipsilateral hemispheric ischaemic events: results from the ACSRS study. Eur J Vasc Endovasc Surg 2005; 30: 275 – 284 03 Arning C, Widder B, von Reutern G et al. Ultraschallkriterien zur Graduierung von Stenosen der A. carotis interna – Revision der DEGUMKriterien und Transfer in NASCET-Stenosierungsgrade. Ultraschall in Med 2010; 31: 251 – 257 04 Wardlaw JM, Chappell FM, Best JJ et al. Non-invasive imaging compared with intra-arterial angiography in the diagnosis of symptomatic carotid stenosis: a meta-analysis. Lancet 2006; 367: 1503 – 1512 05 S3-Leitlinie Diagnostik, Therapie und Nachsorge der extracraniellen Carotisstenose AWMF-Registernummer 004–028. www.awmf.org Last accessed on 25 May 2013 06 Kaufmann TJ, Huston J, Mandrekar JN et al. Complications of diagnostic cerebral angiography: Evaluation of 19826 consecutive patients. Radiology 2007; 243: 812 – 819
07 Widder B, Görtler M. Stenosen und Verschlüsse der extrakraniellen A. carotis. In: Widder B, Görtler M (eds.): Doppler- und Duplexsonographie der hirnversorgenden Arterien. Berlin, Heidelberg, New York: Springer; 2004: 167 – 190 08 Yao J, van Sambeek MR, Dall'Agata A et al. Three-dimensional ultrasound study of carotid arteries before and after endarterectomy. Stroke 1998; 29: 2026 – 2031 09 Bendick PJ, Brown OW, Hernandez D et al. Three-dimensional vascular imaging using Doppler ultrasound. Am J Surg 1998; 176: 183 – 187 10 Keberle M, Jenett M, Beissert M et al. Three-dimensional power Doppler sonography in screening for carotid artery disease. J Clin Ultrasound 2000; 28: 441 – 451 11 Keberle M, Jenett M, Wittenberg G et al. Comparison of 3D power-doppler-ultrasound, color doppler ultrasound and digital subtraction angiography in carotid artery disease. Rofo 2001; 173: 133 – 138 12 Wessels T, Harrer JU, Stetter S et al. Three-dimensional assessment of extracranial Doppler sonography in carotid artery Stenosis compared with digital subtraction angiography. Stroke 2004; 35: 1847 – 1851 13 Vicenzini E, Galloni L, Ricciardi M et al. Advantages and pitfalls of threedimensional ultrasound imaging of carotid bifurcation. Eur Neurol 2011; 65: 309 – 316 14 North American Symptomatic Carotid Endarterectomy Trial collaborators. Beneficial effect of carotid endarterectomy in symptomatic patients with high-grade carotid stenosis. N Engl J Med 1991; 325: 445 – 453 15 Grouven U, Bender R, Ziegler A et al. Comparing methods of measurement. Dtsch med Wochenschr 2007; 132: e69 – e73 16 Widder B, Görtler M. Extrakranielle Duplexsonographie. In: Widder B, Görtler M (eds.) Doppler- und Duplexsonographie der hirnversorgenden Arterien. Berlin, Heidelberg, New York: Springer: 2004: 167 – 190 17 Boussel L, Herigault G, de La Vega A et al. Swallowing, arterial pulsation and breathing induce motion artefacts in carotid artery MRI. J Magn Reson Imaging 2006; 23: 413 – 415 18 Arning C. Ultrasonographic criteria for diagnosing a dissection of the internal carotid artery. Ultraschall in Med 2005; 26: 24 – 28 19 Topakian R, King A, Kwon SU et al. Ultrasonic plaque echolucency and emboli signals predict stroke in asymptomatic carotid stenosis. Neurology 2011; 77: 751 – 758 20 Piscaglia F, Nolsoe C, Dietrich CF et al. The EFSUMB Guidelines and Recommendations on the Clinical Practice of Contrast Enhanced Ultrasound (CEUS): update 2011 on non-hepatic applications. Ultraschall in Med 2012; 33: 33 – 59 21 ter Haar G. Safety and bio-effects of ultrasound contrast agents. Med Biol Eng Comput 2009; 47: 893 – 900 22 Staub D, Patel MB, Tibrewala A et al. Vasa Vasorum and Plaque Neovascularization on Contrast-Enhanced Carotid Ultrasound Imaging Correlates With Cardiovascular Disease and Past Cardiovascular Events. Stroke 2009; 41: 41 – 47 23 Herzberg M, Boy S, Hölscher T et al. Prehospital stroke diagnostics based on neurological examination and transcranial ultrasound. Crit Ultrasound J 2014; 6: 3 24 Ota H, Takase K, Rikimaru H et al. Quantitative vascular measurements in arterial occlusive disease. RadioGraphics 2005; 25: 1141 – 1158 25 Dodds S. The haemodynamics of asymmetric stenoses. Eur J Vasc Endovasc Surg 2002; 24: 332 – 337 26 Lewis RF, Abrahamowicz M, Cote R et al. Predictive power of duplex ultrasonography in asymptomatic carotid disease. Ann Intern Med 1997; 127: 13 – 20 27 Nabavi DG, Ringelstein EB, Faiss J et al. Regional and national stroke units in Germany. Amended certification criteria. Nervenarzt 2012; 83: 1039 – 1052 28 U-King-Im JM, Hollingworth W, Trivedi RA et al. Cost-effectiveness of diagnostic strategies prior to carotid endarterectomy. Ann Neurol 2005; 58: 506 – 515 29 Buskens E, Nederkoorn PJ, Buijs-Van Der Woude T et al. Imaging of carotid arteries in symptomatic patients: cost-effectiveness of diagnostic strategies. Radiology 2004; 233: 101 – 112
Pelz JO et al. Quantification of Internal … Ultraschall in Med 2015; 36: 487–493
493
Downloaded by: Chinese University of Hong Kong. Copyrighted material.
Original Article
487
Quantification of Internal Carotid Artery Stenosis with 3D Ultrasound Angiography Quantifizierung von Stenosen der Arteria carotis interna mit 3D-Ultraschallangiografie Authors
J. O. Pelz1, A. Weinreich1, D. Fritzsch2, D. Saur1
Affiliations
1
Department of Neurology, University of Leipzig, Germany Department of Neuroradiology, University of Leipzig, Germany
Key words
Abstract
Zusammenfassung
"
!
!
Purpose: The aim of this study was to evaluate a new method of three-dimensional ultrasound (3D-US) angiography of carotid vessels including 3D-US quantification of internal carotid artery (ICA) stenosis (ICAS). Materials and Methods: Two neurologists performed native ultrasound scans of the brain-supplying carotid arteries in 73 probands (including 22 patients with 25 cases of ICAS) using a Toshiba Aplio 500 ultrasound machine equipped with the 3 D application of Curefab CS. Additionally, 25 probands underwent contrast-enhanced magnetic resonance angiography (CE-MRA) of the neck vessels. Results: The mean length of the proximal ICA was 32.1 mm ± 9.8 mm & 31.3 mm ± 9.0 mm (each n = 97). The interrater reliability (intraclass correlation, ICC) was 0.75 ± 0.23 for the common carotid artery (CCA, n = 90) and 0.78 ± 0.21 (n = 92) for the ICA. The intermethod agreement between 3D-US angiography and CE-MRA was 0.67 ± 0.19 (n = 45) & 0.66 ± 0.19 (n = 44) for the CCA and 0.79 ± 0.17 (n = 47) & 0.75 ± 0.19 (n = 46) for the ICA. Quantification of ICAS applying 3D-US in comparison to 2D color-coded duplexsonography (2D-CDS) showed a moderate to good intermethod agreement both by the Bland and Altman analysis and by ICC (0.8 & 0.72; with each p < 0.001). The interrater reliability for quantification of ICAS was 0.79; p < 0.001. Conclusion: Native 3D-US angiography of carotid vessels shows good interrater and intermethod agreement in comparison to CE-MRA. Together with a moderate to good intermethod and interrater agreement in the quantification of ICAS, when compared to 2D-CDS, 3D-US angiography is thus a promising complementary imaging technique for carotid artery disease.
Ziel: In dieser Studie wurde eine neue Methode der drei-dimensionalen Ultraschallangiografie (3D-US) der hirnversorgenden Halsgefäße einschließlich der Quantifizierung von Stenosen der Arteria carotis interna (ICAS) evaluiert. Material und Methoden: Zwei Neurologen untersuchten in 73 Probanden (davon 22 Patienten mit 25 ICAS) die extrakraniellen hirnversorgenden Gefäße mit einem Toshiba Aplio 500 Ultraschallgerät, das mit einer 3D-Anwendung von Curefab CS ausgerüstet war. Bei 25 Probanden wurde zusätzlich eine Kontrastmittel-Magnetresonanzangiografie der Halsgefäße (CE-MRA) durchgeführt. Ergebnisse: Die mittlere Länge der proximalen Arteria carotis interna (ICA; jeweils n = 97) betrug 32.1 mm ± 9.8 mm bzw. 31.3 mm ± 9.0 mm. Die Interrater Übereinstimmung (Intraklassen-Korrelationskoeffizient, ICC) betrug 0.75 ± 0.23 für die Arteria carotis communis (CCA; n = 90) und 0.78 ± 0.21 (n = 92) für die ICA. Die Intermethoden Übereinstimmung (ICC) zwischen der 3D-US Angiografie und der CE-MRA lag bei 0.67 ± 0.19 (n = 45) bzw. 0.66 ± 0.19 (n = 44) für die CCA und bei 0.79 ± 0.17 (n = 47) bzw. 0.75 ± 0.19 (n = 46) für die ICA. Im Vergleich des 3D-US mit der konventionellen 2D-Farbduplexsonografie fand sich bei der Quantifizierung von ICAS eine moderate bis gute Intermethoden Übereinstimmung sowohl in der Bland und Altman Analyse als auch mittels ICC (0.8 bzw. 0.72; p jeweils < 0.001). Die Interrater Übereinstimmung für die Stenosenquantifizierung lag bei 0.79; p < 0,001. Schlussfolgerung: Im Vergleich zur CE-MRA zeigt die native 3D-US Angiografie der extrakraniellen hirnversorgenden Gefäße eine gute Interrater und Intermethoden Übereinstimmung. Zusammen mit der moderaten bis guten Intermethoden und Interrater Übereinstimmung bei
● carotid arteries ● angiography ● 3 D ultrasound ● internal carotid " " "
artery stenosis
● ultrasound color Doppler "
received accepted
24.8.2014 24.11.2014
Bibliography DOI http://dx.doi.org/ 10.1055/s-0034-1398749 Published online: January 21, 2015 Ultraschall in Med 2015; 36: 487–493 © Georg Thieme Verlag KG Stuttgart · New York · ISSN 0172-4614 Correspondence Dr. Johann Otto Pelz Deparment of Neurology, University of Leipzig Liebigstraße 20 04103 Leipzig Germany Tel.: ++ 49/3 41/9 72 42 00 johann.pelz@medizin. uni-leipzig.de
Pelz JO et al. Quantification of Internal … Ultraschall in Med 2015; 36: 487–493
Downloaded by: Chinese University of Hong Kong. Copyrighted material.
2
Original Article
der Quantifizierung von ICAS, stellt die 3D-US Angiografie somit eine vielversprechende, komplementäre Bildgebungsmethode zur Untersuchung von Carotisstenosen dar.
Introduction
Materials and methods
!
!
Up to 20 % of ischemic strokes and transient ischemic attacks are caused by large-artery atherosclerosis, i. e., significant (more than 50 %) stenosis or even occlusion of mainly the internal carotid artery (ICA) [1]. Given that the risk of ipsilateral ischemic stroke directly depends on the stenotic value of internal carotid artery stenosis (ICAS) [2], quantification of ICAS as accurately as possible is essential. Applying the flow-based multi-parametric German “DEGUM (Deutsche Gesellschaft für Ultraschall in der Medizin) ultrasound criteria” [3], two-dimensional color-coded duplex sonography (2 D-CDS) has proven to be a noninvasive, cheap and bedside method for the detection and grading of ICAS with good sensitivity and specificity [4] and is recommended as the method of choice for the detection and quantification of ICAS [5]. However, there are still cases where a second imaging modality is desirable or even mandatory. Although digital subtraction angiography (DSA) is still considered the gold standard for the quantification of ICAS [4, 5], it is scarcely performed in the clinical routine mostly due to its invasive character and potential complications (6). Therefore, following the S3-guidelines, contrast-enhanced computed tomography angiography (CTA) or magnetic resonance angiography (CE-MRA) is recommended if there is reasonable doubt in the 2 D-CDS grading of ICAS (5). This might be the case for instance if there is chronic heart failure, a second stenosis either proximal or distal to ICAS or if there is a high-grade stenosis/occlusion of the contralateral ICA. In these conditions the flow-based ultrasound criteria might overestimate or underestimate the degree of ICAS [7]. Although routinely performed in everyday clinical practice, CTA and CE-MRA have some important limitations like exposure to contrast agent and radiation (CTA) or are absolutely contraindicated like CE-MRA in patients with a heart pacemaker or CTA in patients with chronic renal failure or manifest hyperthyreosis. Preliminary studies on 3 D ultrasound have demonstrated encouraging results (8 – 13). However, they suffered from some serious restrictions, e. g. the overall low quality of vessel reconstruction and the use of diameter-based procedures for the graduation of ICAS instead of the more reliable cross-sectional area (CSA)based measurements. Furthermore, 3 D vessel reconstruction turned out to be time-consuming. Therefore, 3 D ultrasound has so far failed to be able to be used in everyday practice. The aim of the present study is to perform and evaluate a new generation of native 3 D ultrasound angiography of the brainsupplying carotid arteries (common carotid artery [CCA] and ICA) in comparison to CE-MRA. Furthermore, we assess intermethod agreement between 3 D ultrasound and 2 D-CDS quantification according to the “DEGUM ultrasound criteria” in ICAS of different stenotic values. Since all ultrasound scans were done by two neurologists, we will provide interrater reliability for all analyses of the study.
Study population
Pelz JO et al. Quantification of Internal … Ultraschall in Med 2015; 36: 487–493
The study was approved by the local ethics committee and all 73 participants gave their written and informed consent. According to our study aims, we examined three different cohorts: Cohort 1 consisted of 25 healthy volunteers (12 male; mean age 29 years, age range 24 – 37), who all underwent 3 D ultrasound angiography of their right and left carotid vessels (CCA and ICA) in order to evaluate the interrater agreement of native 3 D ultrasound. Cohort 2 consisted of 25 vascular healthy patients from the department of neurology (6 male; mean age 42 years, age range 19 – 72) who underwent both 3 D ultrasound and CE-MRA of the neck vessels in order to calculate intermethod agreement. CE-MRA was done in addition to standard sequences of their diagnostic MRI. Cohort 3 consisted of 22 patients with 25 cases of ICAS (3 bilateral, 4 symptomatic, 21 asymptomatic; 14 male; mean age 71 years, age range 53 – 84) who were recruited from our stroke unit or neurovascular outpatient clinic. In these patients 2 D-CDS and 3 D ultrasound of the stenotic ICA were performed in order to assess intermethod agreement. Finally, a 30-year old female patient with chronic occlusion of the left ICA due to dissection was examined for illustration purposes. All 3 D ultrasound examinations were done by two neurologists (JP, examiner 1 and AW, examiner 2), except for 2 patients of cohort 2 and one patient with bilateral asymptomatic ICAS of cohort 3 who were examined only once.
Three-dimensional B-mode ultrasound scanning 3 D ultrasound scans on the basis of B-mode images were performed using a Toshiba Aplio 500 (Toshiba Medical Systems GmbH, Neuss, Germany) equipped with a linear transducer (PLT-1204BT) set at 13 MHz. For three-dimensional sonography, this commercial ultrasound system was attached to the Curefab CS system (Curefab Technologies GmbH, Munich, Germany). Curefab CS comprised a freehand magnetic field tracking system and a workstation equipped with special software (Curefab CS, version 1.91). Briefly, two sensors which were mounted on the linear transducer traced its longitudinal movement and tilting relative to the 3 orthogonal room axes within a weak magnetic field. Using the spatial and temporal information, this setting enabled exact positioning of common two-dimensional native B-mode images, which could then be concatenated to a virtual 3 D stack. Within this 3 D stack, carotid arteries were reconstructed semiautomatically. First, the vessel’s lumen was marked with a centerline. Then, the contour of the vessel was detected automatically and, if necessary, manually corrected. Practically, all participants were lying in a supine position and were asked not to breathe or swallow during the scan. Subsequently, ultrasound scanning started cranial to the clavicle and medial to the sternocleidomastoid muscle, and the transducer was moved towards the mandible at a speed of about 1 cm per second in an axial direction. The whole scan took about 7 to 10 seconds while about 250 to 360 single B-mode images with a slice thickness of about 120 µm were recorded (36 images per second). In total, each examiner performed 3 ultrasound scans
Downloaded by: Chinese University of Hong Kong. Copyrighted material.
488
Original Article
of the respective carotid arteries, and the best scan was used for post-processing.
489
the quantification of ICAS was not done in real time but after the scanning offline, altogether taking no longer than 5 minutes for each case.
Analysis of 3D-reconstructed carotid arteries The 25 patients of cohort 2 underwent CE-MRA of the neck vessels on a 3 T MRI scanner (Trio, Siemens, Erlangen, Germany) with 0.1 mmol/kg body weight of Gadobutrol (Gadovist ®, Bayer Vital GmbH, Leverkusen, Germany) used as the contrast agent. Corresponding to 3 D ultrasound analyses, the CSA was measured every 2 mm in the vessel’s course. The ICA was measured in a distal direction beginning 2 mm distal to the carotid bifurcation, while the CCA was measured in a proximal direction from 10 mm to 40 mm proximal to the carotid bifurcation.
Quantification of internal carotid artery stenosis
Statistical analysis
First, the cases of ICAS of cohort 3 were graded by a neurologist (JP or AW) with 2 D-CDS applying the multi-parametric German “DEGUM ultrasound criteria” [3]. Then, for blinding both examiners, a combination of characters and numbers was assigned to every patient’s 3 D ultrasound dataset. Post-processing of 3 D ultrasound data started after all ultrasound examinations were completed. This way, it could be ensured that during post-processing of the 3 D ultrasound data, both examiners were unaware of the results of 2 D-CDS. Within the reconstructed ICA, the smallest luminal CSA within the stenosis as well as the normal luminal CSA in the distal course of the vessel were obtained. ICAS was then quantified by calculating the distal CSA reduction percentage, according to the North American Symptomatic Carotid Endarterectomy Trial (NASCET) [14]. Due to the blinding procedure,
Statistical analyses were performed with SPSS version 20.0 (IBM Corporation; New York, NY, USA). Intraclass correlation (ICC; consistency mode or absolute mode) was used to calculate the interrater reliability between examiner 1 and examiner 2, as well as between examiner 1 or examiner 2 and CE-MRA. Briefly, for each vessel, all CSAs measured in the vessel’s course were correlated with the corresponding CSAs of the second examiner or " Fig. 1). The mean ICC and the CSAs obtained from CE-MRA (● standard deviation across vessels were used for describing interrater reliability. Visualization and description of intermethod agreement between 2 D-CDS and 3 D ultrasound quantification of ICAS was achieved by a Bland and Altman analysis (MedCalc® Version 13.0; MedCalc Software) which, in contrast to correlational
Fig. 1 3 D ultrasound of the carotid vessels was performed by two examiners as shown by way of example on the left (examiner 1) and on the right (examiner 2). Within the reconstructed internal carotid artery (ICA), the cross-sectional area (CSA) was measured every 2 mm perpendicular to the vessel’s course with the bifurcation (asterisk) serving as a landmark. In this example the total length of the ICA was 55 mm (left) and 51 mm (right) resulting in 25 CSA pairs for calculating the intraclass correlation coefficient (here: ICC 0.98). CCA common carotid artery. Scale bar each 1 cm.
Abb. 1 Der 3D-Ultraschall der Halsgefäße erfolgte durch zwei Untersucher, wie es beispielhaft rechts (Untersucher 1) und links (Untersucher 2) gezeigt wird. Innerhalb der rekonstruierten Arteria carotis interna (ICA) wurde die Querschnittsfläche (CSA) alle 2 mm senkrecht zum Gefäßverlauf bestimmt, wobei die Bifurkation als Referenzpunkt diente (Stern). In diesem Beispiel beträgt die Länge der ICA 55 mm (links) beziehungsweise 51 mm (rechts). Anhand von insgesamt 25 CSA Paaren konnte so der Intraklassenkoeffizient (hier ICC 0.98) als Maß für die Übereinstimmung zwischen den Untersuchern berechnet werden. CCA Arteria carotis communis. Maßstab jeweils 1 cm.
Pelz JO et al. Quantification of Internal … Ultraschall in Med 2015; 36: 487–493
Downloaded by: Chinese University of Hong Kong. Copyrighted material.
Contrast-enhanced magnetic resonance angiography
Both ultrasound examiners reconstructed and analyzed his/her scans with the carotid bifurcation serving as a landmark for further measurements. Besides the total length of the reconstructable ICA, the CSA perpendicular to the vessel’s course was measured every 2 mm. The ICA was measured in a distal direction beginning 2 mm distal to the carotid bifurcation, while the CCA was measured in a proximal direction from 10 mm to 40 mm proximal to the carotid bifurcation excluding the carotid bulb from the measurements. All CSAs of one vessel were then used " Fig. 1). for calculating interrater and intermethod reliability (●
Original Article
methods, is considered to be more eligible for assessing intermethod agreement [15]. Generally, a p < 0.05 was considered as statistically significant.
mean interrater reliability was 0.75 ± 0.23 (SD) for the CCA (n = 90), while it was 0.78 ± 0.21 (SD) for the ICA (n = 92).
Results !
3 D reconstruction and interrater reliability for CCA and ICA Based on the high spatial and temporal resolution, it was possible to visualize the pulse wave as well as distinct structures of the vessel " Fig. 2). Subsequent semi-auwall like the intima-media complex (● tomatic reconstruction enabled a precise depiction not only of normal carotid vessels but also of different vascular phenomena, such as kinking or coiling of the ICA, a tapered ICA resulting from dissec" Fig. 3). Within the recontion or an ectasia of the proximal ICA (● structed vessels, parameters like average diameter, CSA or volume could be obtained at any point in the vessel’s course. The length of the reconstructable proximal ICA was measured in a total of 49 probands of cohorts 1 and 2 with 98 ICAs for each examiner. The mean length of the ICA was 32.1 mm ± 9.8 mm (standard deviation, SD) for examiner 1 (n = 97) and 31.3 mm ± 9.0 mm for examiner 2 (n = 97). In the one missing case the carotid bifurcation was located cranially, directly under the mandible, making it impossible to insonate even the proximal ICA. Generally, the length of the reconstructable ICA showed a great variance from a minimum of 5 mm to a maximum of 61.5 mm, which is explained by the variable distance from the carotid bifurcation to the mandible. Forty-eight probands of cohort 1 and 2 with 96 carotid vessels were examined by both neurologists. Since the use of ICC for calculating interrater reliability required a vessel length of at least 8 mm, i. e., at least 4 measuring points, a total of 4 ICAs and 6 CCAs had to be excluded. Applying ICC (consistency mode), the
Fig. 3 3 D ultrasound angiography of different vessel morphologies and pathologies: a Common (CCA), external (ECA) and internal (ICA) carotid artery of a healthy volunteer. b Elongated ICA with “kinking”. c Distal “coiling” of extracranial ICA – visualization by 3 D ultrasound (left) and by contrastenhanced MRA (right). d Occlusion of ICA due to dissection. e Ectasia of proximal ICA after thromboendarterectomy. Scale bar each 1 cm.
Pelz JO et al. Quantification of Internal … Ultraschall in Med 2015; 36: 487–493
Fig. 2 Reconstructed 2 D view of primarily axial ultrasound scans showing the pulse wave running through the common carotid artery and deflecting the vessel wall. Scale bar 1 cm. Abb. 2 Rekonstruierte 2D-Darstellung der Pulswelle, wie sie durch die Arteria carotis communis läuft und die Gefäßwand entsprechend auslenkt. Maßstab 1 cm.
Abb. 3 3D-Ultraschall Angiografie unterschiedlicher Gefäßverläufe und -pathologien: a Arteria carotis communis (CCA), externa (ECA) und interna (ICA) eines gesunden Probanden. b Elongierte ICA mit “kinking”. c Distales „coiling“ der extrakraniellen ICA im 3D-Ultraschall (links) und mit Kontrastmittel-MRA (rechts). d ICA-Verschluss aufgrund einer Dissektion. e Ektasie der proximalen ICA nach Thrombendarteriektomie. Maßstab jeweils 1 cm.
Downloaded by: Chinese University of Hong Kong. Copyrighted material.
490
Original Article
Comparison of 3 D ultrasound with CE-MRA Each neurologist examined 24 patients (with 48 carotid vessels) who also underwent CE-MRA (cohort 2). Again, calculation of inter-
491
method reliability required a minimum vessel length (ultrasound and CE-MRA) of at least 8 mm, leading to the exclusion of 3 CCAs and 1 ICA for examiner 1 and 4 CCAs and 2 ICAs for examiner 2. Overall, comparison of 3 D ultrasound angiography of carotid vessels with CE-MRA showed a moderate to good agreement. The mean ICC (consistency mode) for the CCA was 0.67 ± 0.19 (SD, n = 45) for examiner 1, and 0.66 ± 0.19 (n = 44) for examiner 2. Concerning the ICA, the mean ICC (consistency mode) was 0.79 ± 0.17 (n = 47) for examiner 1 and 0.75 ± 0.19 (n = 46) for examiner 2.
Fig. 4 A very high-grade (80 % according to DEGUM NASCET) eccentric stenosis of the internal carotid artery. a Coronal view and b sagittal view obtained by a 90° clockwise rotation. Inlets show corresponding native ultrasound scans exhibiting an irregular plaque texture which ranges from echorich to echolucent. Origin of external carotid artery (not shown) is marked by an asterisk. Scale bar 1 cm. Abb. 4 Sehr hochgradige (80 % nach DEGUM NASCET) exzentrische Stenose der Arteria carotis interna. a koronare und b sagittale Darstellung nach 90° Drehung im Uhrzeigersinn. Die kleinen Bilder zeigen entsprechende native Ultraschallaufnahmen in denen gut die irreguläre Plaquetextur von echoreich bis echoarm zu erkennen ist. Der Ursprung der Arteria carotis externa (nicht dargestellt) ist durch ein Sternchen markiert. Maßstab 1 cm.
In cohort 3, 22 patients with 25 cases of ICAS were examined with 2 D and 3 D ultrasound. According to 2 D-CDS and the German “DEGUM ultrasound criteria”, 11 of 25 cases of ICAS were graded as ≥ 70 % or high-grade. Three-dimensional ultrasound scanning with subsequent reconstruction and evaluation of ICAS was possible in 21 of 25 cases of ICAS (84 %, examiner 1) and in 19 of 23 cases of ICAS (83 %, examiner 2). The 4 missing cases of ICAS could not be reconstructed due to calcified plaques resulting in extensive acoustic shadowing. The reconstructed ICAS could be rotated in any direction, allowing also a qualitative analysis of (eccentric) ste" Fig. 4). Intermethod agreement between 2 D-CDS and 3 D nosis (● ultrasound was 0.8 with p < 0.001 (ICC, consistency mode) for examiner 1 and 0.72 with p < 0.001 for examiner 2. Bland and Altman analysis showed moderate intermethod agreement for both examiners (examiner 1: bias – 0.08 %; 1.96 SD limits of agreement + 11.6 % and – 27.6 % – examiner 2: bias – 0.06 %; 1.96 SD limits of " Fig. 5). Altogether, 18 cases of agreement + 19.2 % and – 32.3 %; ● ICAS were graded by both examiners, yielding an interrater reliability for 3 D ultrasound quantification of 0.79 with p < 0.001 (ICC, absolute mode; n = 18). In 9 of 18 (50 %) cases of ICAS, both examiners assessed the same stenotic value while in 6 of 18 (33 %) cases of ICAS, a difference of ≤ 10 % was found. The remaining 3 cases of ICAS (17 %) showed a difference in stenotic value of ≤ 20 %.
Fig. 5 Intermethod agreement between 3 D ultrasound and 2 D color-coded duplexsonography for the quantification of internal carotid artery stenosis. Bland and Altman analyses for a examiner 1 and b examiner 2 show moderate to good agreement. Abb. 5 Intermethoden-Vergleich der Quantifizierung von Stenosen der Arteria carotis interna zwischen 3D-Ultraschall und 2D-Farbduplexsonografie. Die Bland und Altman Analysen für a Untersucher 1 und b Untersucher 2 zeigen jeweils eine moderate bis gute Übereinstimmung.
Pelz JO et al. Quantification of Internal … Ultraschall in Med 2015; 36: 487–493
Downloaded by: Chinese University of Hong Kong. Copyrighted material.
Quantification of ICA stenosis by 3 D ultrasound and comparison with 2 D-CDS
Original Article
Discussion !
In contrast to 2 D-CDS of extracranial carotid vessels, which is routinely performed in everyday clinical practice, so far, 3 D ultrasound angiography has only played a minor preclinical role as a diagnostic tool. However, our results demonstrate that, owing to major recent technical developments, 3 D ultrasound now allows reconstruction of neck vessels with high quality and reliability. In our study 3 D ultrasound angiography of extracranial carotid vessels could visualize the ICA and CCA with good interrater reliability and moderate to good intermethod agreement when compared to CE-MRA. Most importantly, the quantification of ICAS by 3 D ultrasound also yielded good results in comparison to common 2 D-CDS. Moreover, full 3 D reconstruction of carotid vessels with free rotation allows surgeons and interventionalists to study course and morphology of ICAS in detail which is further facilitated by the high spatial resolution of 3 D ultrasound " Fig. 2, 4). Starting at the carotid bifurcation, the proximal ICA (● could be depicted – on average – over a distance of more than 30 mm. This appears to be sufficient for a profound inspection of its origin, the preferential site for stenosis and dissection [16]. Additionally, 3 D ultrasound angiography of neck vessels did not only show a good interrater reliability but also a moderate to good intermethod agreement when compared to CE-MRA. It is noteworthy that a better interrater and intermethod agreement for 3 D ultrasound was probably limited by arterial pulsation arti" Fig. 2). However, APAs are present facts (APA) in the 3 D data (● to a lesser degree also in CE-MRA. The high temporal resolution of 36 images per second in B-mode ultrasound compared to 3 images per second in CE-MRA thereby explains the higher vulnerability for APAs in 3 D ultrasound. In general, APAs could be resolved by electrocardiogram gating [17], which was not implemented in our study but is also not used in routinely performed CE-MRA. Swallowing and deep breathing are other conditions that might induce considerable motion artifacts in both 3 D ultrasound and CE-MRA [17]. However, unlike CE-MRA, 3 D ultrasound can easily be repeated given the short scan time of approximately 7 – 10 seconds and the possibility to check the raw data online. The high quality of 3 D ultrasound angiography of carotid vessels was further emphasized by the reliable depiction of complex vascular phenomena, such as a tapered occlusion of " Fig. 3d), which is considered to be pathognomonic for the ICA (● ICA dissection known to be a difficult ultrasound diagnosis [18]. In our patient cohort, 84 % of cases of ICAS could be examined by 3 D ultrasound. The remaining cases of ICAS could not be analyzed due to calcified plaques resulting in extensive acoustic shadowing. Yet, this is a common problem for native B-mode ultrasound of ICAS, and our results are in line with larger trials like the Asymptomatic Carotid Emboli Study (ACES), where 15.4 % of stenotic carotid plaques could not be classified as either echolucent or echogenic due to heavy calcification [19]. A promising solution for overcoming insufficient B-mode image quality in cases of extensive acoustic shadowing or echolucent stenosis could be the use of second-generation ultrasound contrast agents. Owing to a safety profile which is comparable to MRI contrast agents [20, 21], second-generation ultrasound contrast agents are nowadays routinely used in neurosonology research [e. g. 22] – even in acute ischemic stroke [23]. Quantification of ICAS by 2 D-CDS applying the “DEGUM ultrasound criteria” compared with 3 D graduation using the distal CSA reduction percentage showed moderate to good intermethod agreement in both the Bland and Altman analysis and intraclass correlation. Notably, 3 D ultra-
Pelz JO et al. Quantification of Internal … Ultraschall in Med 2015; 36: 487–493
" Fig. 5). sound overestimated ICAS by an average of about 8 % (● This overestimation of stenotic value when using CSA reduction percentage is a well-known fact owing to the second power of diameter going into the circle’s formula [24]. Taking into account that the “DEGUM ultrasound criteria” are conformed to DSA measurements which used the distal diameter reduction approach [14], our results indeed fit fairly well. Furthermore, the morphology of a stenosis has to be considered. Dodds [25] could demonstrate in vitro that flow disturbances created by an eccentric stenosis would be different from that created by a concentric one when expressing stenotic value only as diameter reduction percentage, while hemodynamic effects of eccentric as well as concentric stenoses would be the same if stenotic value was assessed as CSA reduction percentage [25]. As exemplarily shown " Fig. 4, adequate assessment of diameter reduction percenin ● tage also highly depends on the perspective (i. e., in the case of DSA on the direction of X-rays), especially in eccentric ICAS, which might easily result in underestimation of stenotic value. Restrictively, it should be noted that a high-grade ICAS with a distal diameter reduction percentage of about 70 % would be a stenotic value of 80 – 90 %, when using the CSA as a reference, which might make it difficult then to assess a further progression of stenotic value over time [7, 24]. Although this progression of asymptomatic ICAS was shown to increase the risk of ipsilateral stroke or transient ischemic attack [26], there are no recommendations in the current S3-guidelines as to what should be the best treatment in this case, probably due to the lack of clinical trials [5]. Additionally, parameters like plaque echolucency or microembolic signals alone or in combination are more suited for estimating the stroke risk of asymptomatic ICAS [19]. Furthermore, there are only 2 other studies analyzing 3 D ultrasound for the quantification of ICAS and offering a Bland and Altman analysis, which is considered to be the statistical “gold standard” for describing intermethod agreement [15]. Wessels et al. obtained a similar intermethod agreement using 3 D ultrasound (power mode) with – unlike our study – the distal diameter reduction percentage approach in comparison to 2 D-CDS [12]. But only Yao et al. examined ICAS with native 3 D ultrasound by measuring both the local diameter and local CSA reduction percentage and comparing it with DSA. They found that intermethod agreement was better when using local CSA reduction percentage [7]. However, the use of local stenotic value and a small sample size of 14 cases of ICAS hinder profound interpretation of their data. This is also the case in most of the previous 3 D ultrasound studies, in which due to poor image quality and a lack of real three-dimensional vessel reconstruction only the diameter and not the CSA reduction percentage had been measured. Additionally, astonishingly high intermethod agreement and interrater agreement are attributable to inadequate statistical methods [9 – 11, 13]. This way, by assessing morphology of ICAS and distal CSA reduction percentage in one examination step, 3 D ultrasound angiography as performed in our study seems to be most eligible for a profound examination of ICAS and therefore has the potential to overcome important methodological and clinical limitations of currently used angiographic techniques like CTA, CE-MRA or DSA. Contrary to conventional 2 D-CDS which is still regarded as an imaging modality considerably depending on the examiner’s experience, 3 D ultrasound is more objective since both the raw data and the reconstructed vessel/stenosis allow complete reevaluation and repeated measurements independent from the initial examination. This way, vascular surgeons and interventional neuroradiologists could study the anatomy of ICAS in de-
Downloaded by: Chinese University of Hong Kong. Copyrighted material.
492
tail offline facilitating the choice of the appropriate revascularization procedure (operation or stenting) as well as the preparation of the respective procedure itself. Moreover, given the amended certification criteria of regional and national stroke units in Germany, which demand that 85 – 90 % of patients with ischemic stroke should routinely be examined by 2 D-CDS [27], both methods could be applied in a complementary manner in one bedside examination, thus minimizing time effort. Such an approach of starting the examination of carotid vessels in ischemic stroke or transient ischemic attack by the combination of 2 D-CDS and confirming 3 D ultrasound would also be cost-saving since an expensive and potentially harmful imaging modality like DSA and – with respect to costs – to a lesser extent also CE-MRA and CTA would be reserved for cases where 2 D and 3 D ultrasound were not conclusive [28, 29]. In our study, the gold standard was 2 D-CDS instead of DSA. However, nowadays, 2 D-CDS could be considered the clinical standard [5] and an invasive diagnostic tool like DSA would have exposed patients to an inappropriate risk [6]. Another potential limitation of our study is that only 4 of the 22 patients with cases of ICAS presented with an acute stroke as a result of a symptomatic high-grade ICAS. Because patients with symptomatic ICAS are in particular need of a second imaging modality before undergoing intervention, future 3 D ultrasound studies should focus on this study population. However, we are optimistic that the easy acquisition of 3 D ultrasound data will be accomplished even in acute stroke patients. In conclusion, we propose that complementary to 2 D-CDS, 3 D ultrasound is a promising bed-side diagnostic tool for the examination of the carotid vessels, particularly for the quantification of ICAS. Together with its time and cost-saving aspects, 3 D ultrasound has the potential to replace X-ray-based angiographic techniques in this indication.
Acknowledgment !
The authors wish to thank Prof. Volker Keim for giving organisational and technical support.
References 01 Grau AJ, Weimar C, Buggle F et al. Risk factors, outcome and treatment in subtypes of ischemic stroke: The German Stroke Data Bank. Stroke 2001; 32: 2559 – 2566 02 Nicolaides AN, Kakkos SK, Griffin M et al. Severity of asymptomatic carotid stenosis and risk of ipsilateral hemispheric ischaemic events: results from the ACSRS study. Eur J Vasc Endovasc Surg 2005; 30: 275 – 284 03 Arning C, Widder B, von Reutern G et al. Ultraschallkriterien zur Graduierung von Stenosen der A. carotis interna – Revision der DEGUMKriterien und Transfer in NASCET-Stenosierungsgrade. Ultraschall in Med 2010; 31: 251 – 257 04 Wardlaw JM, Chappell FM, Best JJ et al. Non-invasive imaging compared with intra-arterial angiography in the diagnosis of symptomatic carotid stenosis: a meta-analysis. Lancet 2006; 367: 1503 – 1512 05 S3-Leitlinie Diagnostik, Therapie und Nachsorge der extracraniellen Carotisstenose AWMF-Registernummer 004–028. www.awmf.org Last accessed on 25 May 2013 06 Kaufmann TJ, Huston J, Mandrekar JN et al. Complications of diagnostic cerebral angiography: Evaluation of 19826 consecutive patients. Radiology 2007; 243: 812 – 819
07 Widder B, Görtler M. Stenosen und Verschlüsse der extrakraniellen A. carotis. In: Widder B, Görtler M (eds.): Doppler- und Duplexsonographie der hirnversorgenden Arterien. Berlin, Heidelberg, New York: Springer; 2004: 167 – 190 08 Yao J, van Sambeek MR, Dall'Agata A et al. Three-dimensional ultrasound study of carotid arteries before and after endarterectomy. Stroke 1998; 29: 2026 – 2031 09 Bendick PJ, Brown OW, Hernandez D et al. Three-dimensional vascular imaging using Doppler ultrasound. Am J Surg 1998; 176: 183 – 187 10 Keberle M, Jenett M, Beissert M et al. Three-dimensional power Doppler sonography in screening for carotid artery disease. J Clin Ultrasound 2000; 28: 441 – 451 11 Keberle M, Jenett M, Wittenberg G et al. Comparison of 3D power-doppler-ultrasound, color doppler ultrasound and digital subtraction angiography in carotid artery disease. Rofo 2001; 173: 133 – 138 12 Wessels T, Harrer JU, Stetter S et al. Three-dimensional assessment of extracranial Doppler sonography in carotid artery Stenosis compared with digital subtraction angiography. Stroke 2004; 35: 1847 – 1851 13 Vicenzini E, Galloni L, Ricciardi M et al. Advantages and pitfalls of threedimensional ultrasound imaging of carotid bifurcation. Eur Neurol 2011; 65: 309 – 316 14 North American Symptomatic Carotid Endarterectomy Trial collaborators. Beneficial effect of carotid endarterectomy in symptomatic patients with high-grade carotid stenosis. N Engl J Med 1991; 325: 445 – 453 15 Grouven U, Bender R, Ziegler A et al. Comparing methods of measurement. Dtsch med Wochenschr 2007; 132: e69 – e73 16 Widder B, Görtler M. Extrakranielle Duplexsonographie. In: Widder B, Görtler M (eds.) Doppler- und Duplexsonographie der hirnversorgenden Arterien. Berlin, Heidelberg, New York: Springer: 2004: 167 – 190 17 Boussel L, Herigault G, de La Vega A et al. Swallowing, arterial pulsation and breathing induce motion artefacts in carotid artery MRI. J Magn Reson Imaging 2006; 23: 413 – 415 18 Arning C. Ultrasonographic criteria for diagnosing a dissection of the internal carotid artery. Ultraschall in Med 2005; 26: 24 – 28 19 Topakian R, King A, Kwon SU et al. Ultrasonic plaque echolucency and emboli signals predict stroke in asymptomatic carotid stenosis. Neurology 2011; 77: 751 – 758 20 Piscaglia F, Nolsoe C, Dietrich CF et al. The EFSUMB Guidelines and Recommendations on the Clinical Practice of Contrast Enhanced Ultrasound (CEUS): update 2011 on non-hepatic applications. Ultraschall in Med 2012; 33: 33 – 59 21 ter Haar G. Safety and bio-effects of ultrasound contrast agents. Med Biol Eng Comput 2009; 47: 893 – 900 22 Staub D, Patel MB, Tibrewala A et al. Vasa Vasorum and Plaque Neovascularization on Contrast-Enhanced Carotid Ultrasound Imaging Correlates With Cardiovascular Disease and Past Cardiovascular Events. Stroke 2009; 41: 41 – 47 23 Herzberg M, Boy S, Hölscher T et al. Prehospital stroke diagnostics based on neurological examination and transcranial ultrasound. Crit Ultrasound J 2014; 6: 3 24 Ota H, Takase K, Rikimaru H et al. Quantitative vascular measurements in arterial occlusive disease. RadioGraphics 2005; 25: 1141 – 1158 25 Dodds S. The haemodynamics of asymmetric stenoses. Eur J Vasc Endovasc Surg 2002; 24: 332 – 337 26 Lewis RF, Abrahamowicz M, Cote R et al. Predictive power of duplex ultrasonography in asymptomatic carotid disease. Ann Intern Med 1997; 127: 13 – 20 27 Nabavi DG, Ringelstein EB, Faiss J et al. Regional and national stroke units in Germany. Amended certification criteria. Nervenarzt 2012; 83: 1039 – 1052 28 U-King-Im JM, Hollingworth W, Trivedi RA et al. Cost-effectiveness of diagnostic strategies prior to carotid endarterectomy. Ann Neurol 2005; 58: 506 – 515 29 Buskens E, Nederkoorn PJ, Buijs-Van Der Woude T et al. Imaging of carotid arteries in symptomatic patients: cost-effectiveness of diagnostic strategies. Radiology 2004; 233: 101 – 112
Pelz JO et al. Quantification of Internal … Ultraschall in Med 2015; 36: 487–493
493
Downloaded by: Chinese University of Hong Kong. Copyrighted material.
Original Article
