Article Text

Original research
Comparative analysis between 1-D, 2-D and 3-D carotid web quantification
  1. Catarina Perry da Camara1,2,
  2. Raul G Nogueira1,
  3. Alhamza R Al-Bayati1,
  4. Leonardo Pisani1,
  5. Mahmoud Mohammaden1,
  6. Jason W Allen3,
  7. Fadi Nahab4,
  8. Marta Olive Gadea1,5,
  9. Michael R Frankel1,
  10. Diogo C Haussen1
  1. 1 Marcus Stroke & Neuroscience Center, Grady Memorial Hospital, Emory University School of Medicine, Atlanta, Georgia, USA
  2. 2 Department of Neuroradiology, Centro Hospitalar Universitário Lisboa Central, Lisboa, Portugal
  3. 3 Radiology and Imaging Sciences, Emory University School of Medicine, Atlanta, Georgia, USA
  4. 4 Neurology, Emory University School of Medicine, Atlanta, Georgia, USA
  5. 5 Department of Neurology, Hospital Vall d'Hebron, Barcelona, Spain
  1. Correspondence to Dr Diogo C Haussen, Neurology, Marcus Stroke and Neuroscience Center, Grady Memorial Hospital. Emory University School of Medicine, Atlanta, Georgia, USA; diogo.haussen{at}


Background Carotid webs (CaW) are now recognized as a cause of ischemic stroke in young patients. The thromboembolic potential appears related to the CaW’s morphology and consequent impact on local flow dynamics. We aim to evaluate the reliability of different measurement methods for the quantification of CaW and their relationship to symptomatic status, presence of large vessel occlusion stroke (LVOS), clot burden and final infarct volume.

Methods This was a retrospective analysis of the local comprehensive stroke center CaW database (September 2014–July 2019). CT angiograms (CTAs) were reviewed independently by two raters, blinded to the clinical information and laterality of the stroke/transient ischemic attack. CaW were quantified with 1-D (length), 2-D (area) and 3-D (volume) measurements via Osirix software. Final infarct volume was calculated on MRI. Patients with superimposed CaW thrombus and no repeat imaging were excluded.

Results Forty-eight CaW (37 symptomatic and 11 contralateral/asymptomatic) in 38 patients were included. Mean age (±SD) was 48.7 (±8.5) years, 78.9% were women and 77.1% were black. Inter-rater agreement was 0.921 (p<0.001) for 1-D, 0.930 (p<0.001) for 2-D, and 0.937 (p<0.001) for 3-D CaW measurements. When comparing symptomatic with asymptomatic CaW, mean web length was 3.2 mm versus 2.5 mm (p<0.02), median area was 5.8 versus 5.0 mm2 (p=0.35) and median volume was 15.0 versus 10.6 mm3 (p<0.04), respectively. CaW with a thinner profile (longer intraluminal projection compared with the base) were more likely to be symptomatic (0.67±0.17 vs 0.88±0.37; p=0.01). Average CaW 1-D and final infarct volume had a weak but positive association (Κ=0.230, p<0.05), while no association among web measurements and the presence of LVOS or clot burden was observed.

Conclusion CaW dimension quantification (1-D, 2-D and 3-D) is highly reproducible. Linear and volumetric measurements were more strongly associated with symptoms. The impact of CaW size on the presence of LVOS, clot burden and final infarct volume is unclear.

  • stroke
  • CT angiography
  • vessel wall
  • artery

Data availability statement

No data are available. Not applicable.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Carotid web (CaW) is an intimal fibromuscular dysplasia variant that presents as a septum-like luminal protrusion on the posterior wall of the proximal internal carotid artery.1 CaW is being increasingly recognized as a cause of anterior circulation stroke/transient ischemic attacks (TIAs) in young patients without other risk factors2–6 A suitable non-invasive diagnostic tool is computed tomographic angiography (CTA), which is fast, safe, widely available and has comparable accuracy to digital subtraction angiography (DSA).7

Although its prevalence in the general population is not clear, previous reports have estimated an overall CaW frequency of 0.3%–0.5% ipsilateral to stroke/TIAs,8 2.5% ipsilateral to intracranial large vessel occlusions (LVOs),9 and 8.9% ipsilateral to TIAs.6 A matched case-control study showed 23% ipsilateral CaW in anterior carotid cryptogenic strokes versus none on trauma control cases.10 Patients with symptomatic CaW have been demonstrated to have pronounced rates of recurrent ischemic events, ranging from 17% to 71% and commonly occurring despite the use of antithrombotics.1 3 5 8 11

The web characteristics that may lead to higher risk of thromboembolic potential are not known.12–14 Symptomatic CaW have been demonstrated to be more conspicuous compared with asymptomatic/contralateral lesions, which could constitute a factor predisposing to higher flow disruption and thrombogenicity.5 In this study we aim to analyze different measurement methods to quantify CaW extent, and to evaluate their relationship with symptomatic status, presence of large vessel occlusion stroke (LVOS), clot burden and final infarct volumes.


This was a retrospective analysis of our prospective observational CaW database encompassing two comprehensive stroke centers spanning September 2014 to July 2019. This study was approved by the Institutional Review Board.

Patient selection

Patients diagnosed with a suspected symptomatic CaW were systematically added to the CaW database. The diagnosis of a symptomatic CaW was defined as the presence of a shelf-like linear filling defect in the posterior aspect of the carotid bulb on CTA in patients with acute ischemic stroke or TIA of undetermined etiology with a negative evaluation (as defined by TOAST criteria, Trial of Org 10172) and with the area of acute cerebral infarct in the vascular territory of the internal carotid artery affected with the CaW.15 16 All the included asymptomatic CaW derived from contralateral carotids in patients with bilateral CaW who were symptomatic.

Individuals with symptomatic CaW and a good quality CTA with contiguous axial 0.625 mm images were included in the present analysis. Patients with other potential stroke etiology (TOAST criteria of two or more potential causes) were excluded. CaW with superimposed clot were not included unless there was follow-up imaging confirming the presence of an underlying CaW and complete clot resolution. Inconspicuous shelf-like intraluminal protrusions that could indicate diminutive CaW were excluded if too small to measure by any of the methods (online supplemental figure 1).


CTA images were obtained on a GE Discovery 750HD 64 slices or GE Revolution HD 256 slices (GE Healthcare, Waukesha, Wisconsin, USA) for most (93.8%) of included webs, while outside hospital 126–256 slice scanners were used in the remaining case. Images were reviewed independently by two raters, a neuroradiologist and a neurointerventionist, blinded to the clinical information and laterality of the stroke/TIA.

CaW were quantified by linear means for length and base (1-D), as well as by area (2-D) and volume (3-D) measurements (figure 1) via Osirix MD software (Pixmeo, Switzerland) by each rater: 1-D was obtained with an oblique MIP that intersects the main axis of the CaW on thin cut axial plane at a 90° angle, allowing the measurement of the base and the length of the CaW projection into the lumen; 2-D was defined by the area of the CaW delineated free-hand in the same view as the 1-D; 3-D was calculated with thin axial CTA cuts via volumetric analysis of free-hand delineated CaW limits of each slice. The web thickness was defined as the 1-D length divided by the base (distance between the shoulders of the lesion; figure 1A).

Figure 1

Different methodologies for web measurement. (A) 1-D/linear measurement; (B) 2-D/area measurement; (C) representative cuts of the 3-D/volumetric measurement.

LVO was defined as an occlusion of the intracranial internal carotid artery or proximal middle cerebral artery (MCA) segments, from MCA origin until the limen insula (M1) and to the circular sulcus (M2). Clot burden was calculated as follows: a score of 10 points when there was contrast opacification on the entire anterior circulation on CTA; two points were subtracted for thrombus/absence of contrast opacification in the proximal M1, distal M1 or supraclinoid internal carotid artery, and one point for each M2 branches, A1 or infraclinoid internal carotid artery.17


Follow-up MRI scans (or CT when MRI was unavailable) performed in the first 72 hours were analyzed using Osirix MD software (Pixmeo) and final infarct volume (FIV) was calculated through a semi-automated volumetric analysis (plugin VoxelVolume) of free-hand delineated diffusion weighted images (DWI) lesions on each slab.

Statistical analysis

Demographic and clinical data were reported as mean (±SD) for normal variables or median (IQR) for non-parametric distributions. Intraclass correlation coefficient was used to evaluate the inter-rater agreement and Fisher r-to-z transformation was utilized to compare the agreement in different measures. Two sample t-test/Mann-Whitney U test were employed to compare imaging characteristics (length, area, volume, ratio length/base) between symptomatic and asymptomatic CaW and between patients with or without LVOS. Kendal τ was used to test correlation between web dimensions and clot burden and FIV. Statistical significance was considered when p-value≤0.05. Statistical analysis was performed using IBM SPSS Statistics, version 26 (IBM, Armonk, New York, USA).


Out of 51 patients with CaW, 38 fit the inclusion criteria (online supplemental figure 1). Fourteen (36.8%) out of the 38 patients had bilateral webs (a symptomatic and an asymptomatic side) leading to 52 distinct lesions. Four webs had to be excluded due to either residual superimposed clot on follow-up CTA that were impossible to accurately measure or due to diminute/non-measurable web, totaling 48 webs (37 symptomatic and 11 asymptomatic) for the analysis. Mean age was 48.7 (±8.5) years, 78.9% were women, 77.1% were black, and 92.1% had a stroke (table 1).

Table 1

Demographics, risk factors and web characteristics. Data are numbers (%)

CaW characteristics

The symptomatic CaW were associated with a LVO in 56.3% (n=27). Out of 37 symptomatic CaW, 8 (21.6%) had a superimposed clot on the first CTA (table 2) and were confirmed to have resolution on following CTA. The mean (SD) length of all webs was 3.05 mm (±1.13), median base was 3.77 mm (IQR 2.91–4.63), median area was 5.21 mm2 (4.04–8.43) and median volume was 12.9 mm3 (7.8–20.4) (table 2).

Table 2

Web characteristics and clinical findings. Data are numbers (%) unless indicated otherwise

CTA inter-rater agreement for different quantification modalities

The reliability of length, area and volume measurements of CaW was good to excellent. Intraclass correlation coefficient (ICC) was 0.921, p<0.001 (95% CI 0.848 to 0.959) and 0.915, p<0.001 (95% CI 0.836 to 0.956) for length and base 1-D measurements, respectively; 0.930, p<0.001 (95% CI 0.866 to 0.964) for 2-D and 0.937, p<0.001 (95% CI 0.879 to 0.967) for 3-D measurements. The methods had comparable inter-rater agreement performance (z-score comparing 3-D vs 2-D ICC was 0.23, p=0.82; while 3D vs 1D was 0.49, p=0.62).

Symptomatic vs asymptomatic CaW

Symptomatic mean web length (1-D) and median web volume (3-D) were larger (1-D: 3.2 vs 2.5 mm, p=0.02; 3-D: 15.0 vs 10.6 mm3; p=0.04) than the asymptomatic/contralateral webs, while the median area was not statistically different (2-D: 5.8 vs 5.1 mm2; p=0.43) (table 3). CaW with a thinner profile (longer intraluminal projection in relation to the base) were more likely symptomatic (p=0.01).

Table 3

Web characteristics according to symptoms status

Correlation between CaW quantification and LVOS, clot burden and FIV

All the symptomatic webs had an MRI available to calculate final infarct volume. There was a weak but positive association between average CaW 1-D measurements and FIV (Κ=0.230, p<0.05), although we found no association with other measures (2-D: K=0.199, p=0.086; 3-D: K=0.031, p=0.793). No association among web measurements and the presence of LVOS (1-D: 3.2 mm in patients with LVO vs 2.7 mm in patients without LVO, p=0.29) or clot burden was observed (online supplemental figure 2).


This study was able to provide insight on different methods for quantification of CaW dimensions and their relation to symptoms and stroke characteristics. We found that one-dimensional, two-dimensional and three-dimensional methods are equally reproducible, and that symptomatic CaW are longer, thinner CaW and more voluminous compared with asymptomatic lesions.

CaW may be responsible for a large proportion of young patients otherwise classified as having cryptogenic strokes.2 10 18 19 Therefore, it is relevant to understand which lesion characteristics may predispose an individual to higher thromboembolic potential in order to optimize treatment. It has been previously observed that individuals with symptomatic CaW had more conspicuous lesions compared with their contralateral asymptomatic/incidental webs through a simple linear measurement method.5 CaW were then demonstrated to have longer length in patients with TIA who had recurrent ischemic events compared with those that did not.6 Atherosclerotic carotid plaque volume, contrary to intima-media thickness or percent stenosis, has been suggested to constitute a better predictor of stroke risk.20 Similarly, we evaluated if CaW volumetric analysis could be more reproducible and if it could be more robustly associated with symptomatic versus asymptomatic status; however we observed that the CaW lesion volume performed similarly to the simpler one-dimensional method despite its increased technical complexity.

The hypothesized pathophysiology of CaW as a cause of stroke8 through local flow disruption and thrombogenicity has been corroborated by the demonstration of prominent contrast stagnation on conventional angiography and by the presence of superimposed clots.5 A study investigating flow patterns on CaW revealed larger recirculation zones (reversed flow areas) and regional higher wall sheer stress metrics, which contributes to platelet aggregation and activation, potentially explaining the thrombogenic potential.21 We speculated that more conspicuous webs could have a more significant hemodynamic effect and consequently lead to higher thrombogenic potential. We found no association between web size with LVO status or thrombus burden, although we observed that webs with longer length had larger FIV. Although FIV may relate to clot size, variables such as collateral potential and treatment strategies can exert an important modifying effect. The lack of correlation between thrombus load and web size is not unexpected. Despite the different pathophysiology, it has been reported that patients with milder degrees of coronary atherosclerotic stenosis were observed to have higher thrombus burden compared with higher degrees of steno-occlusive disease.22

Our study has several limitations inherent to retrospective analyses and the limited sample size, especially for subgroup analyses, which is due to this relatively rare condition. The large proportion of LVO in this series could indicate selection bias and affect the analysis correlating clot burden and web size, although it has been demonstrated that patients with cryptogenic stroke and ipsilateral CaW have a higher chance of presenting with a LVOS.10

CaW dimension quantification (1-D, 2-D and 3-D) is highly reproducible. Linear and volumetric measurements were the methods more strongly related to symptomatic versus asymptomatic status. The impact of CaW size on the presence of LVOS, clot burden and final infarct volume is unclear. Further studies are warranted.

Data availability statement

No data are available. Not applicable.

Ethics statements

Patient consent for publication


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Twitter @PerrydaCamaraMD, @pisanileonardo, @diogohaussen

  • Contributors CPC: study conception, design of the work, acquisition of data, statistical analysis, interpretation of data, drafting of the manuscript. RGN, ARA, LP, MM, JWA, FN, MOG, MRF: data acquisition, critical revision of manuscript. DCH: study conception, design of the work, acquisition of data, interpretation of data, critical revision of manuscript. DH is the guarantor.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests CPC, ARA, LP, MM, JWA, FN, MOG, MRF: none. RGN: principal Investigator, Stryker Neurovascular (DAWN trial, no compensation; Trevo 2 trial), Cerenovus/Neuravi (ENDOLOW trial, no compensation); consultant to Stryker Neurovascular; steering committee member, Stryker Neurovascular (no compensation), Medtronic (SWIFT trial, SWIFT Prime trial, no compensation), Cerenovus/Neuravi (ARISE 2 trial, no compensation); angiographic core lab, Medtronic (STAR trial); executive committee member, Penumbra (no compensation); physician advisory board, Cerenovus/Neuravi, Phenox, Anaconda, Genentech, Biogen, Prolong Pharmaceuticals, Allm (no compensation), Viz-AI; stock options, Viz-AI. DCH: consultant for Stryker, Cerenovus, Vesalio; Viz-AI; stock options.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.