If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
University of Manchester, Manchester University NHS Foundation Trust, School of Medical Sciences, Manchester Academic Health Science Centre, Faculty of Biology, Medicine and Health, Manchester, UKManchester University NHS Foundation Trust, Independent Vascular Services Ltd, Wythenshawe Hospital, Manchester Academic Health Science Centre, Manchester, UK
Stenosis severity has been the indication for carotid endarterectomy (CEA) for 4 decades, but the annual stroke risk in asymptomatic carotid stenosis >70% is under 2%. Atherosclerotic volume has emerged as a risk factor for future stroke, but needs to be measured noninvasively. Tomographic ultrasound (tUS) is a novel technology that assembles 3D images in seconds. We evaluated accuracy of measuring Carotid Plaque Volume (CPV) with tUS in patients undergoing CEA.
Consecutive patients were imaged immediately before CEA by tUS and contrast-enhanced tUS (CEtUS). CPV was measured using tUS, CEtUS, and a fused images incorporating both tUS and CEtUS by trained vascular scientists. Precise volume of the endarterectomy specimen was measured using Archimedes technique.
Mean ± sd (range) CPV in 129 endarterectomy specimens was 0.75 ± 0.43 cm3 (0.10–2.47 cm3). Mean ± sd CPV measured by tUS (n = 114) was 0.87 ± 0.51 cm3, CEtUS (n = 104) was 0.75 ± 0.45 cm3 and with fusion (n = 95) was 0.83 ± 0.49 cm3. Differences between specimen volume and CPV measured by tUS (0.13 ± 0.24 cm3), CEtUS (−0.01 ± 0.21 cm3) or fusion (−0.08 ± 0.20) were clinically insignificant. Intra-/interobserver differences were minimal.
tUS accurately measures CPV with excellent intra-/interobserver agreement. CEtUS improves accuracy if precise CPV measurement is needed for research but tUS alone would be sufficient for population screening.
Duplex ultrasound imaging has been used to detect severity of stenosis, the usual indication for carotid endarterectomy (CEA), for 4 decades. In patients with symptomatic carotid stenosis, CEA is clearly indicated to prevent stroke.
However, the indication for CEA in asymptomatic patients is far less clear. Asymptomatic carotid stenosis >70% in patients randomized to best medical therapy carries a stroke risk of only 9.52% at 5 years, or less than 2%/year.
Due to this low prevalence, the ESVS carotid guideline highlighted that screening the general population for ACD using stenosis would not be cost-effective and could lead to patient harm associated with intervention.
The guideline did recognize the importance of screening, if effective prevention strategies could be implemented, and recommended development of validated techniques to identify “high-risk for stroke” asymptomatic patients.
As measuring severity of carotid stenosis in asymptomatic people is not cost-effective and is poor at predicting stroke risk, our current approach may be compared to offering cancer surgery based on tumor size alone and neglecting all other tumor and patient features that determine prognosis. Alternatively, other enhanced ultrasound techniques could be used to detect atherosclerotic plaque features associated with plaque vulnerability and increased risk of stroke.
CPV may be an important indicator of stroke risk (plaque vulnerability) and a more appropriate indication for CEA than severity of stenosis. However, to be clinically useful, particularly in population screening, it would need to be easily and accurately measured noninvasively.
This research explores accuracy and the inter/intraobserver variation achieved by tUS in measuring CPV in patients undergoing CEA. We build on previous work by investigating whether accuracy of measuring CPV could be improved using contrast-enhanced tUS (CEtUS) to produce tUS/CEtUS fusion images that are easier to interpret than greyscale images alone.
Materials and Methods
National Research Ethics Committee (11/NW/0308) approval was obtained. Consecutive patients undergoing CEA in Greater Manchester were asked to give fully informed written consent and participate.
Tomographic 3D Ultrasound
Tomographic 3D ultrasound (tUS) computes a greyscale 3D reconstruction of vascular structures that can be viewed from any angle, but it displays no information regarding hemodynamics or blood flow. Flowing blood and liquid thrombus/soft plaques are difficult to image on greyscale (B-mode) ultrasound (Fig. 1A) which is what tUS is composed of. However, soft tissue structures like the artery wall are seen clearly. Lack of flow information creates potential errors in CPV measurement. Color Doppler (known as duplex when combined with greyscale) can be used to display blood flow but is displayed poorly in 3D by tUS. Therefore, a microbubble contrast is needed to produce clear blood flow imaging using CEtUS. CEtUS can distinguish flowing blood from echolucent thrombus, but the acoustic output power (amount of energy) must be significantly reduced to prevent microbubble destruction. This reduced energy impairs the quality of artery wall/soft tissue definition (Fig. 1B). This could limit CEtUS’ accuracy for measuring CPV. By fusing greyscale tUS images using normal acoustic output (clear artery wall structures), with the CEtUS images with reduced acoustic output (clear flowing blood), identifying plaque boundaries could make CPV more accurate (Fig. 1C).
Preoperative tUS Imaging
Imaging (tUS and CEtUS) was performed by experienced Vascular Scientists, in the same sitting, immediately before CEA without the patient moving between scans (each scan takes 10 sec). A 16-gauge cannula was inserted into a peripheral vein for contrast administration. Ultrasound imaging was undertaken using a 9 MHz linear array transducer on a Resona 7 high-definition duplex instrument (Mindray, Shenzhen, China). Severity of stenosis was recorded according to the UK Joint Committee Guidelines adopted by the Society of Vascular Technology of Great Britain and Ireland and the UK Vascular Society.
To complete the tUS images, sensors were attached to the 9 MHz transducer. Three separate, freehand B-mode, transverse section 3D tUS images were acquired from the supraclavicular fossa to just below the ear using the 9 MHz transducer (Resona 7 duplex instrument) that was magnetically tracked using the aforementioned sensors, (PIUR imaging GmbH, Vienna, Austria). The scan is conducted the same way 1 would when performing a normal transverse ultrasound sweep of the neck, just with additional sensors attached to the ultrasound transducer tracked orientation and position in time and space. With the patient perfectly still and in the same position as the aforementioned tUS scans, 2 mL of microbubble contrast (Sonovue, Bracco, Italy) was administered followed by a 5 mL 0.9% saline flush. Contrast settings were selected on the Resona 7 and 3 CEtUS scans were obtained in the same way. Using screen capture from the Resona 7, the tUS system almost instantly computes multiplaner reconstructions (MPR) with a 3D reconstruction (Fig. 2). Acquiring tUS and CEtUS images required little skill and took 20–30 sec.
Measuring the Endarterectomy Specimen
CEA was undertaken by vascular surgeons who retrieved the specimen in 1 piece where possible. The CEA specimen was collected in a dry pot within minutes of removal. The total plaque and common carotid component lengths were measured (Fig. 3A). The Carotid Research Fellow then measured the dry and immersion weight using an electronic balance and calculated the volume as described previously.
CPV was measured using 3 modalities; tUS, CEtUS and fusion. Vascular Scientists were given the total plaque and common carotid plaque lengths from the CEA specimen to ensure they measured CPV in the same segment of artery (Fig. 3A). They were blinded to the clinical history, specimen volume and each other. CPV was calculated by tracing the intima-media border and blood-intima border in each transverse section using manual planimetry (Figs. 2A and 3B). The interslice distance between each transverse section throughout the length of the carotid plaque was 1 mm (Fig. 2B, C).
For each plaque, the tUS scan was exported from PIUR imaging software and imported into specialist software (ImFusion GmbH, Munich, Germany) on the same computer. CPV was then calculated and a 3D reconstruction was generated (Video 1).
1tUS/CEtUS Fusion Imaging
tUS grey-scale images (optimizing arterial wall resolution) was fused with CEtUS (optimized blood flow imaging), in the Imfusion software (Fig. 1). Image registration (position of 1 3D scan within another) was performed using the artery lumen geometry as landmarks to achieve accurate fusion. Vascular scientists had the option to correct the registration ensuring both arterial lumens were aligned correctly for pitch, angle, and rotation. tUS and CEtUS images were then each assigned different colors (Fig. 1) and CPV measured again.
Inter and Intraobserver Measurement
CPV was measured by 2 Vascular Scientists using all 3 modalities. Each scientist was fully trained in the relevant techniques and were blinded to each other's results and to the CEA volume. Each measured 1 B-Mode CPV, 1 CEUS CPV, and 1 Fusion CPV per participant. This allowed interobserver variability comparison. One Vascular Scientist repeated the 3 CPV measurements (1 B-Mode CPV, 1 CEUS CPV, and 1 Fusion CPV per participant) 3 months later in order to calculate intraobserver variability.
Bland-Altman analysis using GraphPad prism v7 (GraphPad software Inc, CA, USA) was used to determine bias between methods of CPV measurement using tUS, CEtUS, and the fusion images and between observers. Intra-Class Correlation (ICC) estimates and their respective 95% confidence intervals (95% CIs) between observers, tUS, CEtUS, and the fusion images compared to the CEA volume were calculated using SPSS version 22, (SPSS Inc, IL, USA) based on a single-measurement, absolute agreement, 2-way mixed-effects model. Time to analyze CPV was reported in minutes using median values with interquartile ranges (IQRs).
A total of 129 consecutive patients undergoing CEA underwent tUS imaging with endarterectomy specimens collected for analysis. Severe calcification obstructed ultrasound imaging in 15 patients such that volume measurements by tUS were recorded in 114 (105 symptomatic). Contrast imaging (CEtUS) was achieved in 104 of these 114 patients. Nine of the 104 images could not be fused due to tissue distortion leaving 95 with all 3 imaging modalities. Mean (±sd) specimen and common carotid lengths were 2.6 ± 0.7 cm and 1.1 ± 0.4 cm respectively. The mean (±sd) endarterectomy specimen volume calculated by modified Archimedes principle in all 129 specimens was 0.75 (0.43) cm3 with a wide range 0.10–2.47 cm3.
Accuracy of the Three tUS Techniques
Mean (±sd) CPV measured by greyscale tUS in 114 patients was 0.87 (0.51) cm3 with a range of 0.21–3.56 cm3 compared with 0.74 (0.43) cm3 for the endarterectomy specimen volumes in the same individuals. The median (IQR) time needed to analyze CPV on these tUS images was 12:31 (4:29) min. The mean difference (±sd)[95%CI] between tUS and endarterectomy specimen CPV was −0.13 (0.24)[95% CI -0.60–0.34]cm3 (Fig. 4). This difference was considered insignificant clinically when compared with the range of CPV between 0.21 and 3.56 cm3. It was also small compared with mean CPV of 0.74 cm3 and 0.97 cm3 in asymptomatic and symptomatic patients respectively from our previous study (6). The CPV measured by tUS also strongly correlated with the endarterectomy specimen volume with narrow confidence intervals (ICC = 0.85, 95% CI 0.69–0.91) (Fig. 4).
The mean (±sd) CPV measured by CEtUS was 0.75 (0.45) cm3, range 0.12–2.74 cm3 in 104 patients and compares with 0.88 (0.44) cm3 for the endarterectomy specimen volumes in the same individuals. The median (IQR) time to analyze CPV using CEtUS was 12:07 (4:51) min. The mean difference (±sd) [95% CI] between CPV measured by CEtUS and in the endarterectomy specimens was only −0.01 (0.21) [95% CI -0.42–0.39]cm3. The error was again clinically insignificant when compared to the range of CPV volumes and the difference in CPV between symptomatic and asymptomatic patients. The correlation between CPV measured by CEtUS and endarterectomy specimen volumes was even stronger at ICC = 0.89 (95% CI 0.85–0.93).
The principal value of adding contrast enhancement was to enable easier and more rapid interpretation of both blood flow/intima and intima/media boundaries in carotid artery disease. The time to analyze CPV on these fusion images was marginally lower than with the other imaging methods at 11:40 (4:46) min. Although not statistically quicker than with the other techniques, the vascular scientists undertaking the analysis were more confident in their interpretation of fusion images. Mean CPV in the fused images of 95 patients was 0.83 (0.49) cm3, range 0.18–3.10 cm3 and compares with 0.75 (0.43) cm3 for the endarterectomy specimen volumes in the same individuals. The mean difference (±sd) [95% CI] between CPV measured by tUS/CEtUS fusion and the endarterectomy specimens was only −0.08 (0.20) [95% CI -0.47–0.31]cm3 (Fig. 5). The error was again clinically insignificant when compared to the range of CPV volumes and the difference in CPV between symptomatic and asymptomatic patients. The correlation between CPV measured by fusion and endarterectomy specimen volumes was even stronger at ICC = 0.90 (95% CI 0.83–0.94) (Fig. 5).
Inter and Intraobserver Agreement
The mean difference (95% CI) in CPV measured using tUS between the 2 blinded vascular scientists was minimal at only −0.05 (0.26) [95% CI -0.55–0.46] cm3. This tiny difference was clinically insignificant when compared with the range of measurements between 0.21 and 3.56 cm3 and the difference between CPV of 0.97 cm3 in symptomatic and 0.74 cm3 in asymptomatic patients (6). The correlation between the 2 observers was strong with ICC = 0.86 (95% CI 0.81–0.90) (Fig. 6).
The intraobserver variability was, as expected, even smaller with a mean difference (95% CI) between observations of only 0.09 (0.22) [95% CI -0.35–0.52]cm3 (Fig. 6). The correlation between the same observers’ measurement was even stronger with ICC = 0.89 (95% CI 0.82–0.93) (Fig. 6).
Between 2 blinded vascular scientists there was minimal mean difference (95% CI) in CPV measured using CEtUS, −0.04 (0.25) [95% CI -0.52–0.44]cm3. When compared to a range of measurement between 0.12 and 2.74 cm3, the difference was again clinically insignificant. The correlation between the 2 observers was strong with ICC = 0.85 (95% CI 0.79–0.90).
The intraobserver variability was small with a mean difference (95% CI) between observations of only 0.08 (0.24) [95%CI −0.39–0.55]cm3. The correlation between the same observers’ measurement was also strong with ICC = 0.84 (95%CI 0.76–0.89).
The vascular scientists found the fused tUS/CEtUS images significantly easier to interpret and they had greater confidence in the accuracy of their measurements. The mean difference (95%CI) in CPV measured using fused images between the 2 blinded vascular scientists was minimal at only 0.07 (0.21) [95%CI -0.36–0.48]cm3 with smaller confidence intervals than for tUS or CEtUS. Again, when compared to a range of measurement between 0.18 and 3.10 cm3, the difference was clinically insignificant. However, the fusion confidence intervals are narrower than with tUS or CEtUS imaging alone, indicating better reproducibility. The correlation between the same observers’ measurement was also strong with ICC = 0.89 (95%CI 0.82–0.93).
The intra-observer variability was also small with a mean difference (95% CI) between observations of only −0.05 (0.20) [95% CI -0.44–0.34] cm3. The correlation between the same observers’ measurement was also strong with ICC = 0.89 (95% CI 0.83–0.93).
When compared with CEA volume, tUS accurately and reproducibly measured CPV with narrow confidence intervals in this feasibility study which only recruited 129 participants. Nevertheless, tUS/CEtUS fusion imaging, which is easier to interpret and analyze, achieved even high levels of precision. Although this higher level of accuracy achieved by fusion, than plain tUS, may be of research value, it is of little clinical significance when the tUS measurement error was compared with the CPV volume range encountered in patients undergoing CEA. It was also insignificant when compared with the difference in mean CPV between patients with symptoms of cerebral ischemia and asymptomatic carotid disease, especially as there was overlap between the CPV ranges for symptomatic and asymptomatic patients.
Our data clearly demonstrates tUS was sufficiently accurate for clinical purposes but 15 patients were removed due to acoustic calcium shadowing (the same patients where a clinical duplex would have struggled to confirm the degree of stenosis) where tUS/CEtUS image fusion was easier to interpret and achieved a higher degree of precision. Fusion imaging may be needed for studies on the indication for carotid surgery and would be essential for evaluating potential medications to treat atherosclerosis. However, administering contrast intravenously is inappropriate for population screening.
For all tUS measures of atherosclerotic plaque volume, it is important that the proximal and distal limits of the plaque be established for each individual patient. Defining length of the atherosclerotic plaque and the proximal and distal limits for planimetry slices to measure CPV has previously been shown to be a source of operator variability.
We measured CPV in 1 mm slices based on the common carotid length proximal to the bifurcation and the total plaque length. Although theoretically observers may start these measurements from different levels for the carotid bifurcation, the minimal differences between our interobserver measurements were reassuring.
Another source of potential error could be the Archimedes measurement of specimen volume as soft adherent thrombus could theoretically be lost during collection. It is possible the liquefied core of a degenerating plaque could escape during endarterectomy. Although there were inevitably errors in the measurement of both specimen volume and CPV by tUS, it was reassuring these errors were minimal when compared with the range of plaque volumes encountered in patients undergoing CEA. There was also a huge range of carotid specimen volumes in asymptomatic patients and those with a history of recent cerebral ischemia.
The manual planimetry techniques to analyze CPV in this study were time consuming although median (IQR) image analysis could still be completed in only 12:31 (4:29) min/patient. If CPV measurement is to be used for population screening, it would be important to reduce the need for skilled analysis and the time required.
CEtUS and fusion involve additional skill probably limiting their use to research. However, tUS alone was simple and as there was considerable variability in CPV and as it is unlikely there will be a precise cut-off between those at risk of stroke and those that are not, it is likely CPV measured by tUS alone would be adequate as an indication for CEA.
These results confirm tUS was a simple, accurate and noninvasive method for detecting and measuring CPV at carotid bifurcations. Other features of plaque vulnerability, such as texture analysis using 3D grey-scale median or hemodynamic information via computational fluid dynamics need to be explored. Widespread use of CPV as an indication for treatment would be facilitated by the rapid analysis that could be achieved by Artificial Intelligence. Additionally, future research may address the calcium shadowing limitation through fusion of different B-mode tUS scans taken from opposing scan planes (anterior, lateral, and posterior) which may give the ability to peer round any shadow and provide a full picture.
In the meantime, a major prospective cohort study recruiting asymptomatic carotid patients will be essential to identify CPV thresholds that predict cardiovascular events, including ischaemic stroke. For now, accurate and noninvasive measurement of CPV has become part of our clinical assessment of patients with asymptomatic carotid disease.
CPV can be accurately and noninvasively measured using tUS with excellent inter/intraobserver agreement. tUS/CEtUS fusion imaging achieved precise levels of accuracy in the measurement of CPV needed for research. However, the marginal loss of accuracy by using tUS alone was clinically irrelevant when compared with the range of CPVs seen in patients undergoing CEA and the difference in CPV between asymptomatic and symptomatic disease. The role of CPV in predicting future cardiovascular events in patients with asymptomatic carotid disease can only be established by major cohort studies that may also include other biomarkers and ultrasound features.
We thank Dr Julie Morris and Dr John Belcher for statistical advice. The University of Manchester acknowledges the support of the National Institute for Health Research Clinical Research Network (NIHR CRN).
Preliminary results presented at the Vascular Society of Great Britain and Ireland, Charing Cross International Vascular Symposium and the European Society of Vascular Surgery annual meetings.
Funding: This project was funded by a European Union's Horizon 2020 Fast Track to Innovation Programme under grant agreement No 760380.
Conflicts of interest: The sponsors (The University of Manchester), The Funders (European Union's Horizon 2020 FTI fund) and manufacturer (PIUR Imaging GmbH) had no role in the design, execution, interpretation, or writing of the study. Charles McCollum was a founder of PIUR imaging.