Author + information
- Received August 3, 2016
- Revision received February 21, 2017
- Accepted February 23, 2017
- Published online March 5, 2018.
- Roland Richard Macharzina, MDa,∗ (, )
- Sascha Kocher, MDa,
- Steven R. Messé, MDb,
- Thomas Rutkowski, MDa,
- Fabian Hoffmann, MDa,
- Matthias Vogt, MDa,
- Werner Vach, PhDc,
- Nian Fan, MSca,
- Aljoscha Rastan, MDa,
- Franz-Josef Neumann, MDa and
- Thomas Zeller, MDa
- aDepartment of Cardiology and Angiology II, University Heart Center of Freiburg–Bad Krozingen, Bad Krozingen, Germany
- bDepartment of Neurology, University of Pennsylvania, Philadelphia, Pennsylvania
- cCenter for Medical Biometry and Medical Informatics, University of Freiburg, Freiburg, Germany
- ↵∗Address for correspondence:
Dr. Roland Richard Macharzina, University Heart Center of Freiburg–Bad Krozingen, Suedring 15, D-79189 Bad Krozingen, Germany.
Objectives The purpose was to analyze the agreement and binary accuracy of the degree of internal carotid artery stenosis (ICAS) as determined by 4-dimensionally (4D) real-time gray-scale guided 3-dimensional (3D) color-Doppler ultrasonography (3DC-US) (4D/3DC-US) compared with catheter angiography (CA) and duplex ultrasonography (DUS). This study hypothesized that 4D/3DC-US is noninferior to CA and DUS in grading ICAS in selected patients.
Background Clinical stratification in patients with ICAS largely depends on a patient’s symptomatic status and the degree of stenosis.
Methods Screening with 4D/3DC-US was prospectively performed in 93 study patients (with 122 ICASs), thus yielding 80 patients for analysis (with 103 ICASs) after excluding patients with insufficient image quality, previous revascularization, and contraindications to CA. The ultrasound examination (10 MHz) consisted of consensus conform DUS examination and independent real-time 4D-guided gray-scale views for orientation followed by static 3DC-US NASCET (North American Symptomatic Carotid Endarterectomy Trial) percent stenosis quantification using off-line multiplanar rendering. Multiplanar selective CA of the same ICASs was quantified with dedicated software in a blinded fashion.
Results Quantitative CA of 103 stenoses with a mean degree of 65 ± 17% was compared with 4D/3DC-US, with a resulting concordance correlation coefficient of 0.89 and a standard deviation of differences (SDD) of 8.1% at a bias of +1.7%. Binary 50% and 70% stenosis detection with 4D/3DC-US revealed a sensitivity of 97% and 87%, respectively, and a specificity of 92% and 84%, respectively. Interobserver SDD for CA of 52 stenoses (7.2%) did not differ from SDD for 4D/3DC-US and CA (p = 0.274). Accuracy of 50% stenosis detection by 4D/3DC-US was tendentially higher compared with DUS (96% vs. 91%).
Conclusions The 4D/3DC-US method provides reliable and accurate stenosis quantification and binary classification with good diagnostic accuracy compared with CA and DUS.
- carotid artery stenosis
- catheter angiography
- 4-dimensional ultrasonography
- 3-dimensional ultrasonography
Atherosclerosis of the carotid arteries is a well-characterized cause of stroke and transient ischemic attacks. The risk of neurovascular events increases with the degree of internal carotid artery stenosis (ICAS), thus making accurate diagnostic assessment of ICAS, in addition to symptomatic status, essential for risk estimation in a patient (1–3). The accurate diagnosis of angiographically defined >50% ICAS in symptomatic patients and >70% ICAS in asymptomatic patients is important because these cut levels are established criteria for therapeutic stratification (1). Furthermore, diagnosis of ICAS stenosis progression by duplex ultrasonography (DUS) is a significant predictor of neurovascular events during follow-up in asymptomatic patients (4,5). Additionally, there is a clinical need for completely noninvasive ICAS imaging technologies that do not have the nephrotoxic side effects of angiographic contrast agents (1).
DUS has numerous pitfalls when quantification of stenosis is attempted by spectral blood velocity parameters (1,6,7). In particular, for stenoses within the low (<50%) to moderate (50% to 70%) range, intrastenotic velocity increases only slowly relative to the luminal loss, thereby resulting in a lack of diagnostic accuracy that often requires additional morphological estimation of plaque stenosis in B-mode or color-Doppler ultrasonography (US) images according to the NASCET (North American Symptomatic Carotid Endarterectomy Trial) method (6–10). To improve detection of moderate stenoses, US societies suggested combining spectral DUS with visual stenosis estimation within 2-dimensional (2D) US images (6,7). However, direct measurement of stenotic lumen reduction within 2D color-Doppler US images delivered inconsistent results and may not improve accuracy when this method is used alone (8,9). It was hypothesized that 3-dimensional (3D) sonomorphometry could be an important innovation and enhance validity of direct stenosis measurement.
Previously, 3D color-Doppler US (3DC-US) of ICAS was evaluated in reconstructed images using resource-intensive approaches such as freehand 2D color-Doppler scanning and post hoc 3D reconstruction from serial slices using interpolation algorithms (11–13). Stenosis grading with 3DC-US using a dedicated volume probe generating static 3D images for immediate online and post hoc off-line multiplanar rendering (MPR) of relevant vessel segments is an innovative approach. The static 3DC-US volumetric images can be acquired after a quick topoanatomic orientation with real-time 4-dimensionally (4D) guided gray-scale imaging using the same probe. The possibility of remote interpretation of 3D raw data provided by a sonographer implies advantages similar to those of teleradiology and may be clinically applicable to 3DC-US, while bearing in mind observer dependency and other limitations of US (1). Our intention was to assess the diagnostic features of 4D real-time guided 3DC-US (4D/3DC-US) by measuring the NASCET percentage of lumen reduction in comparison with catheter angiography (CA) and DUS (10,14). We hypothesized that the percentage of stenosis grades measured within 3D multiplanar-rendered 3DC-US volume datasets would allow grading of ICAS at an equivalent accuracy compared with these modalities.
We conducted this multimodality imaging study in a prospective, controlled, and blinded fashion by providing patients with a clinically relevant indication for exact stenosis quantification before possible revascularization (1).
Material and Methods
Enrollment of patients and data collection
Consecutive ambulatory patients with suspected carotid artery stenoses were offered participation in our prospective bicentric multimodality imaging study at the University of Freiburg and the University Heart Center of Freiburg–Bad Krozingen in Germany, independent of other studies on patients with carotid artery stenosis (15). US imaging and all analyses were performed in individual patients with a symptomatic indication or relevant carotid plaque stenosis with suspected progression on prescreening. CA was clinically justified and offered to patients with an indication to undergo carotid artery revascularization eventually according to a multidisciplinary clinical board decision.
All therapeutic procedures and decisions were performed independent of this research study protocol. All patients received a neurological examination at baseline. Pre-screening of 216 eligible patients was performed with 2D DUS, followed by clinical and US screening with 4D/3DC-US; the final study examinations were performed using 4D/3DC-US, DUS, and angiography was performed after anonymization (Figures 1 and 2). The study recordings were independently analyzed (Figures 1 and 2). A total of 101 study patients had given written informed consent and were informed about the potential angiographic risk, including irreversible cerebral ischemic events (1,15). Inclusion criteria were informed consent and good US image quality (Figure 1). Study exclusions were made on the basis of clinical (n = 8) and imaging (n = 13) criteria (Figure 1). Other pre-specified exclusion criteria that did not apply were pregnancy, asymptomatic status in patients younger than 50 years of age, mental illness other than dementia, suspected nonatherosclerotic stenosis, and other severe comorbidities (American Society of Anesthesiologists grade ≥III). Finally, 80 patients with good image quality in 4D/3DC-US and CA examinations were included and were blinded for all observers (Figures 1 and 2). The following screening US examination was performed by an experienced examiner who was not involved in the post hoc rendering or reading of study examinations (R.R.M.). Study examinations were scheduled within 2 weeks after inclusion. The off-line readers of 3D Digital Imaging and Communications in Medicine (DICOM) (National Electronic Distributors Association, Rosslyn, Virginia) images were corelab approved (S.K. and F.H.). Data were assessed by corelab. The study fulfilled ethical standards and rules of good clinical practice. All patients in the study had given informed consent. The study was approved by the authorized Ethics Committee. The Standards for the Reporting of Diagnostic Accuracy Studies (STARD) criteria were applied.
Spectral duplex ultrasonography
Pulsed-wave Doppler peak systolic and end-diastolic blood flow velocity measurements across the stenosis and in common carotid arteries were combined with plaque estimation within gray-scale or color-Doppler images and analyzed according to established guidelines by a separate observer (7).
4D and 3D ultrasonography
A 4D 10L linear volume probe and a LOGIQ 9 Ultrasound System (GE Healthcare Technologies, Milwaukee, Wisconsin) allowed 4D-guided gray-scale imaging and 3DC-US. The transducer had 192 elements and a bandwidth from 3.5 to 11 MHz. The field of view of the 4D 10L transducer measured 37.4 mm, and the volume angle was 29°. All continuously acquired 2D gray-scale frames were automatically arranged in a volume instantaneously displaying proper temporal and spatial location in the parasagittal, transverse, and coronal planes without delay (Figure 3A). Static 3DC-US used the same principle for spatial allocation by assigning Doppler shift–related color voxel overlays superimposed on gray-scale signals within an individually minimized frame size at a defined priority.
US study examination was a sequential combination of continuous real-time 4D gray-scale US preview imaging at 5.5 Hz volume rate immediately followed by static 3DC-US image acquisition at the same probe angulation. A superimposed multiplanar color-Doppler view on the gray-scale image resulted, for later off-line rendering and selection of an optimal 3D projection displaying the minimal lumen of the stenosis or the distal vessel section (Figure 3, Online Figures for steps 1 to 3, Online Video 1, and Online Figures for distal vessel segment analogous imaging and measurement). Each of the 4D views was acquired within a few seconds at various angulations around the neck (approximately 10 s/view) to obtain an online anatomic overview of stenosis morphology, the location of measurement points, and the amount of calcification. The static 3DC-US images were made at 2 to 5 angulations during a short breath hold and acquired multiple frames within a single cardiac cycle to allow off-line selection of an optimal view on the target lesion up to the individual reader’s choice.
One volumetric target vessel acquisition was typically <150 ms. Frame rates of 3DC-US full-volume loops depended on the color-Doppler sample size, usually resulting in up to 15 to 25 Hz.
The angulation to the sagittal plane was recorded with a mechanical angulometer fixed to the transducer. The transducer was held in a longitudinal orientation to the vessel center line to acquire oblique sagittal images for instant reconstruction. Color-Doppler and B-mode settings were rigorously adjusted for optimal lumen to plaque border and vessel wall delineation by using lowest possible B-gain at a just sufficient color gain for luminal filling; a reduced color signal priority to prohibit overwriting of gray-scale plaque structures; a pulse repetition frequency <3 kHz at low-pass filter settings; no gray-scale speckle reduction; a small, focused color window; lowest color frame averaging; and high line density. Anonymized images were saved for post hoc analysis in a raw DICOM format indicating a study number, the side, and the angulation of the transducer.
Off-line 3DC-US measurements
The 3D MPR was done in 1 of 2 to 3 angulations after selection of the best plane to resolve off-line still frames of the minimal lumen diameter (MLD) according to the NASCET method by using the LOGIQ 9 Ultrasound System (10,14). These off-line still frames showed separate views of the 3DC-US volume datasets in 3 perpendicular planes with luminal color filling (Figure 3, Online Figures for steps 1 to 3, Online Video 1, and Online Figures for distal vessel segment analogous imaging and measurement). Two of the planes of the ICAS were rendered to the center of the MLD, zoomed and controlled in the third plane in a 3D multiplanar fashion to measure the minimal distance in the coaxial cross section strictly perpendicular to the vessel center line of the residual lumen as verified in the sagittal plane (Figure 3B, Online Figures for steps 1 to 3, Online Video 1, and Online Figures for distal vessel segment analogous imaging and measurement). Lumen measurement distal to the stenosis was made equivalently in a healthy segment without post-stenotic tapering or dilatation, approximately 2 cm distal to the end of a stenosis (Figure 3C, Online Figures for steps 1 to 3, Online Video 1, and Online Figures for distal vessel segment analogous imaging and measurement) (14). All off-line measurements were made within 3DC-US views. Fulfilling all these imaging criteria, each observer chose the point of measurements individually with the best target visualization. Single-diameter measurements at the MLD and distal location were performed in separate single planes after MPR of 3D volumes and were stored digitally. The whole process including 3D MPR and measurements of NASCET diameters in 2 appropriate views took less than an average of 5 min.
CA was performed in study patients selectively to image suspected stenosis with a digital system (Axiom, Siemens, Germany) according to protocol shortly after the clinical screening visit (Figures 1 and 2). A 6-F diagnostic calibration catheter (Cordis, Milpitas, California) was placed in the common carotid artery. Cine runs (16 bit) were acquired after injection of 10 ml iodixanol (Visipaque 270) (GE Healthcare) at a rate of 5 ml/s. Multiple oblique projections were made to capture the highest stenosis grade. The digital subtraction mode was used. The best angulation of the x-ray beam was independent and determined without knowledge of 4D/3DC-US probe angulation. The plane with the maximum degree of stenosis was chosen to measure the MLD and the distal diameter in the ICAS according to the NASCET method by using semiautomatic lumen detection with quantitative vascular analysis software (MEDIS, Leiden, the Netherlands) (10).
For angiographic interobserver analysis, the raw data of 52 consecutive stenoses was rated by 2 blinded observers who were approved by corelab (T.R. and F.N.). Out of several projections the optimal static image was selected and rendered independently to quantify the NASCET percentage of stenosis. Similarly, the interobserver reliability of 4D/3DC-US NASCET percentage of stenosis measurements was assessed by 2 independent readers within the subset of 52 consecutive stenoses (S.K. and F.H.). A comparative metric and binary analysis was performed for both modalities.
Results were computed with SPSS software version 20.0 (SPSS, Chicago, Illinois) and MedCalc version 126.96.36.199 (MedCalc Software, Ostend, Belgium). Normal distribution was checked using the Kolmogorov-Smirnov test. The accuracy of continuous 4D/3DC-US measurements compared with CA as the gold standard was assessed by the mean and standard deviation of differences (SDD) supplemented by Lin’s concordance correlation coefficient (CCC) and by binary analysis of the area under the receiver-operating characteristic (ROC) curve (AUC) when predicting a CA stenosis grade >50% and 70% and Cohen’s linear weighted kappa. Binary accuracy was considered for the cutpoints and was assessed by sensitivity, specificity, and predictive values. McNemar test was used to compare intermethod sensitivities and specificities. Levene test was applied to detect inhomogeneity of variances. Differences of the AUC were tested for significance. Significance was assumed at a 2-sided p < 0.05. Interobserver variation was assessed by Bland-Altman and kappa analysis using the following categories for interpretation of kappa: <0.2, poor; 0.21 to 0.40, fair; 0.41 to 0.60, moderate; 0.61 to 0.80, good; 0.81 to 1.00, very good (16). The Pearson correlation coefficient was calculated to show intermethod associations for comparison with published reports.
We prospectively examined a referral cohort of 93 patients with 4D/3DC-US and CA after patient selection according to our exclusion and inclusion criteria (Figure 1). Thirteen patients were excluded because of limited image quality. Finally, 80 patients (103 stenoses) were analyzed (Figure 1). Measurements were taken in 55 right and 48 left carotid arteries. In 21 patients both sides were analyzed. The cohort consisted of 60 men (20 women) 40 to 87 years of age. The mean age was 69.0 ± 9.8 years. Ten patients were older than 80 years of age. Within 6 months before inclusion, 36 patients had a neurological symptom (stroke or transitory ischemic event) ipsilateral to the stenoses examined. Cardiovascular risk factors of all patients included are listed in Table 1.
The mean percentage of stenosis was 65 ± 17%, range 9% to 100% for CA; and 67 ± 18%), range 11% to 100% for 4D/3DC-US. Percentages of stenosis were classified by 4D/3DC-US as follows: 0% to 49%, 15 (15%); 50% to 69%, 35 (34%); 70% to 99%, 52 (50%); 100%, 1 (1%). For categories of CA percentages of stenosis, see Table 1.
Metric percentage of stenosis comparison of 4D/3DC-US with CA
The mean of the differences between 4D/3DC-US and CA indicates a systematic difference of +1.7% (p < 0.001) (Table 2, Figure 4). The SDD was ±8.1%, a finding suggesting that 95% of all differences will be between −14.5% and 17.9% (Table 2, Figure 4). The CCC between 4D/3DC-US and CA was 0.89 (Table 2). The Pearson correlation coefficient was 0.89. Comparison of metric measures between the 2 cohorts (subgroup of interobserver analysis, n = 52; with all stenoses, n = 103) showed no significant differences (p for SDD = 0.74, p for mean difference = 0.62; and p for CCC = 0.06) (Table 2).
Binary stenosis grading
4D/3DC-US showed a sensitivity of 97% with a specificity of 92% to detect 50% stenosis (Table 3). Sensitivity and specificity for 70% stenoses were 8% to 10% lower (Table 3). Cohen’s kappa was 0.75 (Table 2). Performance of DUS velocity grading is highlighted in Table 3. Accuracies of DUS were similar. No significant difference in sensitivity and specificity was noted for 4D/3DC-US and DUS (Table 3). 4D/3DC-US assigned 12 (12%) stenoses correctly that were erroneously classified by DUS regarding 70% stenosis and 8 (8%) stenoses at 50%.
ROC analysis showed an AUC of 0.99 for 50% NASCET stenosis and 0.93 for 70% (Figure 5). There were no significant differences of the AUCs between the 2 cohorts (n = 52 or 103) for diagnosing 50% or 70% stenosis (Table 2).
In an interobserver subset of 52 lesions, Cohen’s kappa for the CA raters was 0.62 and 0.68 for the 4D/3DC-US observers (Table 4). The 4D/3DC-US measures of percentage of stenosis had a mean difference of 4.2%, and the CA had a difference of 1.5% (Table 4, Figure 6). The SDD was 8.9% between 4D/3DC-US and 7.2% between CA raters. Coefficient of variation was 13.3% in 4D/3DC-US versus 10.8% in CA. No significant differences between CA and 4D/3DC-US interrater analysis using the CCC were found (Table 4).
Comparison of the CA interobserver with the 4D/3DC-US to CA intermethod analysis
Bland-Altman analysis between 2 CA readers did not differ from corresponding results between the CA and 4D/3DC-US comparison (p for comparison of bias = 0.88 and SDD 0.35). CCCs between the CA and 4D/3DC-US readers did not differ from those of the CA readers (p = 0.34).
Our evaluation demonstrates that 4D/3DC-US measures and classifies the percentage of stenoses accurately and in a highly reproducible manner as compared with CA, the established reference standard (1,6,9–12). Stenosis quantification with 4D/3DC-US has good intermethod agreement with CA that does not differ from the agreement between 2 independent CA raters assessing the same study. Accuracies were similar to those achieved with DUS grading.
The degree of ICAS and its progression over time are accepted markers predictive of outcomes and are routinely used in therapeutic decision making (1,4,5,15). The approach is feasible; 4D/3DC-US could be applied for screening in 86% of our study patients to quantify NASCET stenosis. This technique allowed rapid 3D acquisition of vessel segments within seconds, and the whole procedure including 3D MPR and stenosis measurement was performed in <5 min. We demonstrated a good overall discriminatory capability of 4D/3DC-US for binary stenosis detection by ROC analysis (Table 2). Considering the no harm maxim in diagnosing ≥50% symptomatic stenoses before a possible revascularization procedure, the high specificity and a very high AUC of ROC analysis would support a potential application of 4D/3DC-US (Table 2). In screening asymptomatic patients for clinically more relevant ≥70% stenoses, 4D/3DC-US showed slightly lowered sensitivity, specificity, and AUC of ROC analysis as compared with ≥50% stenoses in our specific cohort, whereas overall sensitivity and specificity did not significantly differ from those of DUS. Furthermore, 4D/3DC-US slightly improved misdiagnosis by DUS for 50% and 70% stenoses.
Summarizing our comparison with CA and DUS, our data suggest that 4D/3DC-US sonomorphometry improves on the weakness of the established DUS modality for diagnosing ≥50% stenoses (6–10). We speculate that 4D/3DC-US may at least have a role in a combined approach with DUS, thus enhancing specificity. A combined approach needed to be addressed in a larger cohort that would allow analysis of the performance of 4D/3DC-US in patients with indeterminate DUS scans, as previously defined (17). Current diagnostic consensus criteria for noninvasive DUS recommend a combination of plaque estimation and a set of quantitative spectral blood flow velocity parameters (7). Our grading with 4D/3DC-US of 70% NASCET compares well with DUS results reported in a previous meta-analysis (18).
The high accuracy and reproducibility of 4D/3DC-US may at least in part be explained by the possibility of rendering the target lesion and the healthy distal segment separately within the 3D dataset as recorded at multiple 4D/3DC-US probe angulations. Therefore, 4D/3DC-US has a potential to overcome previous limitations of 2DC-US stenosis grading: it allows for selection of the best views for percentage of stenosis measuring online and off-line according to the well-standardized NASCET method, thereby contributing to reproducibility in detecting and quantifying the relevant vessel sections by independent observers (10,18). Our 4D/3DC-US offers imaging features that allow instant delineation of diastolic luminal flow from plaque borders and immediate 3D multiplanar reconstruction to render the optimal plane according to the NASCET method (Figure 2) (10,14). The multiplanar acquisition of the whole dataset of a carotid artery in <1 s is contrasted with the former apparatus of 3D image generation that made the images more prone to pulsation artifacts (11,13). Other features of the 4D/3DC-US analysis are that examinations and analyses can be performed separately, thus offering new perspectives for later reconstruction.
This study described stenosis quantification by 4D/3DC-US with a dedicated probe that acquires volume datasets without the need for freehand movement of the transducer. Previous studies on 3DC Doppler US on much lower numbers of ICAS resulted in significantly lower correlation coefficients (R) with angiography compared with the overall R = 0.89 with our 4D/3DC method: R = 0.57; n = 48 stenoses; p for difference <0.0001 (12); R = 0.74; n = 62 stenoses; p for difference = 0.0041 (11). Furthermore, these studies resulted in inhomogeneous sensitivity and specificity (11,12). At least 1 former study was not focused on ICAS alone, including external and common carotid artery stenoses or tandem stenoses (13).
Taken together the 4D/3DC-US innovation may offer new possibilities for patients’ screening or as a method complementary to established imaging modalities, especially in patients with contraindications (e.g., contrast agent intolerance) to standard diagnostic tools.
Bearing the selection criteria in mind, our data would justify further evaluation in real-life observational studies. We considered CA in this paper as a gold standard even though stenosis measurement determined on the basis of CA lacks 100% reproducibility. Actually, we observed that the SDDs between CA and 4D/3DC-US measurements were of similar magnitude as the SDDs between the 2 CA observers (8.1 vs. 7.2), findings suggesting that replacing CA with 4D/3DC-US may have the same impact as replacing it with the results determined by a second reader.
The main limitation of the 4D/3DC-US modality is the influence of acoustic shadowing caused by dense calcifications. This issue can be an underestimated confounder even in our cohort and may partially explain the lowered specificities for the 70% stenoses because it is known that arteries with higher degrees of stenoses are more prone to calcifications (19). The relatively lower sensitivity to diagnose 70% stenosis compared with 50% may also be a cohort-specific selection effect with a higher prevalence of more severe stenoses.
This systematic evaluation of 4D/3DC-US suggests strong potential clinical validity and reliability. Larger multicenter trials will be required to determine whether 4D/3DC-US is superior to other established noninvasive modalities such as 2D DUS, magnetic resonance imaging, or computed tomography angiography.
The use of 4D US–guided 3DC-US is an innovative tool for stenosis grading with high reproducibility and overall agreement when compared with CA and DUS.
COMPETENCY IN MEDICAL KNOWLEDGE: The use of 4D/3DC-US allows rapid, reproducible detection and quantification of NASCET percentage of stenosis determinations of the internal carotid artery in most patients with high correlation, agreement, and binary classification accuracy that do not significantly differ from those seen with CA. The 4D/3DC-US method offers fast and reliable imaging and off-line interpretation with overall diagnostic performance similar to that of DUS.
TRANSLATIONAL OUTLOOK: Future studies should address whether 4D/3DC-US is superior to other established noninvasive modalities such as 2D DUS, magnetic resonance imaging, or computed tomography angiography or whether some individual modalities are preferred over others, depending on the ultrasonographic characteristics of the suspected stenosis. The use of 4D/3DC-US may provide a valuable tool for the remote diagnosis of carotid stenosis in clinical practice.
Dr. Messé has received research support from GlaxoSmithKline and Bayer; and is on the Clinical Events Committee of the SALUS trial sponsored by Direct Flow Medical. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- catheter angiography
- concordance correlation coefficient
- duplex ultrasonography
- 4-dimensionally real-time gray-scale–guided 3-dimensional color-Doppler ultrasonography
- internal carotid artery stenosis
- minimal lumen diameter
- multiplanar rendering
- standard deviation of differences
- 3-dimensional color-Doppler ultrasonography
- Received August 3, 2016.
- Revision received February 21, 2017.
- Accepted February 23, 2017.
- 2018 American College of Cardiology Foundation
- Brott T.G.,
- Halperin J.L.,
- Abbara S.,
- et al.
- Sabeti S.,
- Schlager O.,
- Exner M.,
- et al.
- Hirt L.S.
- von Reutern G.,
- Goertler M.,
- Bornstein N.M.,
- et al.
- Wessels T.,
- Harrer J.U.,
- Stetter S.,
- et al.
- Bucek R.A.,
- Reiter M.,
- Dirisamer A.,
- et al.
- North American Symptomatic Carotid Endarterectomy Trial
- Macharzina R.R.,
- Claus C.,
- Messé S.R.,
- et al.
- Altman D.G.