Advertisement

International Union of Angiology (IUA) consensus paper on imaging strategies in atherosclerotic carotid artery imaging: From basic strategies to advanced approaches

      Highlights

      • Carotid artery disease is a risk factor for ischemic stroke and a predictor of cardiovascular events.
      • Imaging assessment of carotid artery disease is critical for surveillance, risk stratification, and patient management.
      • A comprehensive assessment of carotid artery disease should aid physicians and surgeons in their decision-making.

      Abstract

      Cardiovascular disease (CVD) is the leading cause of mortality and disability in developed countries. According to WHO, an estimated 17.9 million people died from CVDs in 2019, representing 32% of all global deaths. Of these deaths, 85% were due to major adverse cardiac and cerebral events. Early detection and care for individuals at high risk could save lives, alleviate suffering, and diminish economic burden associated with these diseases.
      Carotid artery disease is not only a well-established risk factor for ischemic stroke, contributing to 10%–20% of strokes or transient ischemic attacks (TIAs), but it is also a surrogate marker of generalized atherosclerosis and a predictor of cardiovascular events. In addition to diligent history, physical examination, and laboratory detection of metabolic abnormalities leading to vascular changes, imaging of carotid arteries adds very important information in assessing stroke and overall cardiovascular risk. Spanning from carotid intima-media thickness (IMT) measurements in arteriopathy to plaque burden, morphology and biology in more advanced disease, imaging of carotid arteries could help not only in stroke prevention but also in ameliorating cardiovascular events in other territories (e.g. in the coronary arteries).
      While ultrasound is the most widely available and affordable imaging methods, computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET), their combination and other more sophisticated methods have introduced novel concepts in detection of carotid plaque characteristics and risk assessment of stroke and other cardiovascular events. However, in addition to robust progress in usage of these methods, all of them have limitations which should be taken into account. The main purpose of this consensus document is to discuss pros but also cons in clinical, epidemiological and research use of all these techniques.

      Graphical abstract

      Keywords

      1. Current scenario and evidence

      1.1 Targets of carotid imaging

      Detailed imaging assessment of extracranial carotid artery disease is critical for appropriate risk stratification and management of those presenting with cerebrovascular ischemia as well as of selected asymptomatic individuals [
      • Bos D.
      • van Dam-Nolen D.H.K.
      • Gupta A.
      • et al.
      Advances in multimodality carotid plaque imaging: AJR expert panel narrative review.
      ].
      The degree of luminal stenosis in the carotid bifurcation has historically served as the primary imaging feature for determining ischemic stroke risk and the potential need for surgery. Contemporary multimodality imaging including ultrasound, magnetic resonance imaging (MRI/MRA), CT angiography (CTA), and even positron emission tomography (PET-CT) or PET-MRI methods target more detailed visualization of carotid plaque components that indicate plaque vulnerability (e.g. maximum plaque thickness and volume, calcification, ulceration, intraplaque hemorrhage, inflammation, intraplaque neovascularization, lipid-rich necrotic core, and thin or ruptured fibrous cap) [
      • Bos D.
      • van Dam-Nolen D.H.K.
      • Gupta A.
      • et al.
      Advances in multimodality carotid plaque imaging: AJR expert panel narrative review.
      ] (Table 1).
      Table 1Main imaging methods for carotid arteries.
      Atherosclerosis vulnerable plaqueAccuracySensitivityBurdenAvailabilityStandardization
      Ultrasound++++++-++++ -
      + -
      CT angiography+++++++++ -++
      - -
      Magnetic resonance++++++++-- -+ -
      ++
      Positron emission tomography++++++++- - -+ -
      +++
      Not infrequently a carotid scan can indirectly (through flow patterns) detect the status of other vascular territories or even other abnormal findings of surrounding structures (e.g. a thyroid nodule) [
      • Chen G.
      • Xue Y.
      • Wei J.
      • Duan Q.
      The undiagnosed potential clinically significant incidental findings of neck CTA: a large retrospective single-center study.
      ].

      1.2 The role of carotid arteries imaging

      There are several reasons that require carotid imaging, the predominant being evaluation after a cerebrovascular event but also for CVD screening, risk stratification, and prevention as well as for surveillance after a carotid procedure [
      • Bos D.
      • van Dam-Nolen D.H.K.
      • Gupta A.
      • et al.
      Advances in multimodality carotid plaque imaging: AJR expert panel narrative review.
      ,
      • AbuRahma A.F.
      • Avgerinos E.D.
      • Chang R.W.
      • et al.
      Society for Vascular Surgery clinical practice guidelines for management of extracranial cerebrovascular disease.
      ].
      Imaging of the carotid bifurcation is essential in all patients with symptoms of cerebral ischemia, whether they present as a TIA or complete stroke [
      • Bos D.
      • van Dam-Nolen D.H.K.
      • Gupta A.
      • et al.
      Advances in multimodality carotid plaque imaging: AJR expert panel narrative review.
      ,
      • AbuRahma A.F.
      • Avgerinos E.D.
      • Chang R.W.
      • et al.
      Society for Vascular Surgery clinical practice guidelines for management of extracranial cerebrovascular disease.
      ]. If significant carotid artery disease is identified as the source of symptoms, these patients are candidates for a carotid intervention to prevent a secondary stroke [
      • AbuRahma A.F.
      • Avgerinos E.D.
      • Chang R.W.
      • et al.
      Society for Vascular Surgery clinical practice guidelines for management of extracranial cerebrovascular disease.
      ]. Although imaging for this indication is most often performed with a carotid duplex ultrasound, when evaluation of the vessels proximal or distal to the cervical portion of the carotid artery is required for diagnosis or to plan endovascular or surgical therapy, additional imaging with CTA, MRA or digital subtraction angiography (DSA) is indicated [
      • AbuRahma A.F.
      • Avgerinos E.D.
      • Chang R.W.
      • et al.
      Society for Vascular Surgery clinical practice guidelines for management of extracranial cerebrovascular disease.
      ].
      The use of imaging methods for screening for carotid artery disease in the general population, in particular, to identify significant disease that will prompt an intervention to prevent a stroke are controversial [
      • AbuRahma A.F.
      • Avgerinos E.D.
      • Chang R.W.
      • et al.
      Society for Vascular Surgery clinical practice guidelines for management of extracranial cerebrovascular disease.
      ]. However, targeting selected groups of neurologically asymptomatic patients is well established. These groups can be high-risk patients aged >55 years with cardiovascular risk factors, patients with a carotid bruit on clinical exam, Hollenhorst plaque on fundoscopic examination, and silent infarction on brain imaging examinations [
      • AbuRahma A.F.
      • Avgerinos E.D.
      • Chang R.W.
      • et al.
      Society for Vascular Surgery clinical practice guidelines for management of extracranial cerebrovascular disease.
      ].
      Finally, surveillance after a carotid intervention is a common practice established on the natural history of ipsilateral restenosis, contralateral disease progression, and associated stroke risk.

      2. Ultrasound

      2.1 Stenosis

      Grading and stratification of carotid stenosis is manly based on multiparametric, hemodynamic criteria on Duplex ultrasound [
      • Barlinn K.
      • Rickmann H.
      • Kitzler H.
      • et al.
      Validation of multiparametric ultrasonography criteria with digital subtraction angiography in carotid artery disease: a prospective multicenter study.
      ] (Table 2). The most important parameters are the measurement of the peak systolic and the end-diastolic flow velocity within the stenosis. The accuracy of Duplex ultrasound compared with angiography for detecting >50% and ≥70% stenosis, respectively, is very good, with a positive predictive value of >90% and a specificity of >85% [
      • Jahromi A.S.
      • Cinà C.S.
      • Liu Y.
      • Clase C.M.
      Sensitivity and specificity of color duplex ultrasound measurement in the estimation of internal carotid artery stenosis: a systematic review and meta-analysis.
      ]. Duplex ultrasound is recommended as the primary imaging modality to assess the extent and severity of extracranial carotid stenosis [
      • Aboyans V.
      • Ricco J.-B.
      • Bartelink M.-L.E.L.
      • et al.
      ESC guidelines on the diagnosis and treatment of peripheral arterial diseases, in collaboration with the European society for vascular Surgery (ESVS): document covering atherosclerotic disease of extracranial carotid and vertebral, mesenteric, renal.
      ]. Various studies have also shown that the risk of cerebrovascular events increases not only with the severity of the stenosis but also with rapid progression of the degree of stenosis [
      • Hirt L.S.
      Progression rate and ipsilateral neurological events in asymptomatic carotid stenosis.
      ,
      • Kakkos S.K.
      • Nicolaides A.N.
      • Charalambous I.
      • et al.
      Predictors and clinical significance of progression or regression of asymptomatic carotid stenosis.
      ]. Therefore, patients with an asymptomatic carotid stenosis (ACS) undergo usually annual ultrasound monitoring.
      Table 2(Duplex-)sonographic criteria for grading internal carotid artery stenosis and markers of carotid plaque vulnerability.
      Sonographic markerClassification/Feature of plaque vulnerability
      Grading of internal carotid artery stenosis based on Duplex ultrasoundDegree of stenosis as defined by NASCET
      • -
        < 50%: plaque on B-mode, aliasing on color duplex image, PSV <200 cm/s
      • -
        50–69%: PSV 200–300 cm/s, EDV <100 cm/s, PSV ratio (ICA/CCA) ≥ 2
      • -
        ≥70%: PSV >300 cm/s, EDV >100 cm/s, PSV ratio (ICA/CCA) ≥ 4
      Progression of degree of stenosis (>20%)
      Echogenicity on B-mode ultrasoundHypoechogenic (echolucent) plaque (type 1 or type 2)
      «Grey scale median (GSM)»: GSM <15 (hypoechogenic)
      Increased juxta-luminal hypoechogenic (black) area (>6 mm2)
      Heterogenic echotexture
      Plaque burden on B-mode ultrasound including 3D-ultrasoundLarge plaque area (>40 mm2)/total plaque area
      Large plaque volume/total plaque volume (3D-ultrasound)
      Carotid plaque surface on Duplex-ultrasound and CEUSPlaque surface irregularities (<1–2 mm)
      Plaque ulceration (>1–2 mm)
      Carotid intraplaque neovascularization (IPN) on CEUSIncreased IPN on semi-quantitative measurement:
      • grade 1: no vascularization
      • grade 2: limited or moderate vascularization
      • grade 3: extensive vascularization
      High IPN on semiautomatic quantitative measurement: e.g. large relative perfused area
      Peak systolic flow velocity (PSV), end-diastolic flow velocity (EDV), contrast-enhanced ultrasound (CEUS).

      2.2 Features of vulnerability

      Duplex ultrasound can assess not only the degree of stenosis [
      • Barlinn K.
      • Rickmann H.
      • Kitzler H.
      • et al.
      Validation of multiparametric ultrasonography criteria with digital subtraction angiography in carotid artery disease: a prospective multicenter study.
      ,
      • Mantella L.E.
      • Liblik K.
      • Johri A.M.
      Vascular imaging of atherosclerosis: strengths and weaknesses.
      ], but also several sonomorphological characteristics which are associated with plaque vulnerability [
      • Spanos K.
      • Tzorbatzoglou I.
      • Lazari P.
      • Maras D.
      • Giannoukas A.D.
      Carotid artery plaque echomorphology and its association with histopathologic characteristics.
      ]. Hypoechogenicity including a low grey scale median (GSM) [
      • Gupta A.
      • Kesavabhotla K.
      • Baradaran H.
      • et al.
      Plaque echolucency and stroke risk in asymptomatic carotid stenosis: a systematic review and meta-analysis.
      ,
      • Nicolaides A.N.
      • Kakkos S.K.
      • Kyriacou E.
      • et al.
      Asymptomatic internal carotid artery stenosis and cerebrovascular risk stratification.
      ], large juxtaluminal hypoechogenic area [
      • Salem M.K.
      • Bown M.J.
      • Sayers R.D.
      • et al.
      Identification of patients with a histologically unstable carotid plaque using ultrasonic plaque image analysis.
      ], heterogeneous echotexture [
      • Brinjikji W.
      • Rabinstein A.A.
      • Lanzino G.
      • et al.
      Ultrasound characteristics of symptomatic carotid plaques: a systematic review and meta-analysis.
      ,
      • van Engelen A.
      • Wannarong T.
      • Parraga G.
      • et al.
      Three-dimensional carotid ultrasound plaque texture predicts vascular events.
      ], or higher plaque burden (plaque area, total plaque area [TPA], or plaque volume) [
      • Spence J.D.
      Measurement of carotid plaque burden.
      ,
      • Sillesen H.
      • Sartori S.
      • Sandholt B.
      • Baber U.
      • Mehran R.
      • Fuster V.
      Carotid plaque thickness and carotid plaque burden predict future cardiovascular events in asymptomatic adult Americans.
      ], surface irregularities and ulceration [
      • Brinjikji W.
      • Rabinstein A.A.
      • Lanzino G.
      • et al.
      Ultrasound characteristics of symptomatic carotid plaques: a systematic review and meta-analysis.
      ] on B-mode ultrasound are sonographic features of plaque vulnerability with increased embolic risk [
      • Aboyans V.
      • Ricco J.-B.
      • Bartelink M.-L.E.L.
      • et al.
      ESC guidelines on the diagnosis and treatment of peripheral arterial diseases, in collaboration with the European society for vascular Surgery (ESVS): document covering atherosclerotic disease of extracranial carotid and vertebral, mesenteric, renal.
      ] (Table 2).
      IMT represents mainly the middle layer of the carotid arterial wall and is a marker of arteriopathy [
      • Raggi P.
      • Stein J.H.
      Carotid intima-media thickness should not be referred to as subclinical atherosclerosis: a recommended update to the editorial policy at Atherosclerosis.
      ].
      According to the American Society of Echocardiography, IMT is a subclinical vascular disease rather than synonymous of subclinical atherosclerosis [
      • Stein J.H.
      • Korcarz C.E.
      • Hurst R.T.
      • et al.
      Use of carotid ultrasound to identify subclinical vascular disease and evaluate cardiovascular disease risk: a consensus statement from the American Society of Echocardiography Carotid Intima-Media Thickness Task Force endorsed by the Society for Vascular Medicine.
      ]. A 2020 review article summarized many of the advantages of measuring carotid plaque burden, which is far superior to measuring carotid IMT in many ways [
      • Spence J.D.
      Measurement of carotid plaque burden.
      ].
      Carotid plaque burden (CPB) is useful for risk stratification, treatment of atherosclerosis, research into the biology and genetics of atherosclerosis, and evaluation of new therapies against atherosclerosis.
      Measured as TPA or as 3D plaque volume, CPB is highly correlated with coronary calcium [
      • Sillesen H.
      • Sartori S.
      • Sandholt B.
      • Baber U.
      • Mehran R.
      • Fuster V.
      Carotid plaque thickness and carotid plaque burden predict future cardiovascular events in asymptomatic adult Americans.
      ], and as predictive of events [
      • Baber U.
      • Mehran R.
      • Sartori S.
      • et al.
      Prevalence, impact, and predictive value of detecting subclinical coronary and carotid atherosclerosis in asymptomatic adults: the BioImage study.
      ]; while IMT is neither [
      • Sillesen H.
      • Sartori S.
      • Sandholt B.
      • Baber U.
      • Mehran R.
      • Fuster V.
      Carotid plaque thickness and carotid plaque burden predict future cardiovascular events in asymptomatic adult Americans.
      ,
      • Baber U.
      • Mehran R.
      • Sartori S.
      • et al.
      Prevalence, impact, and predictive value of detecting subclinical coronary and carotid atherosclerosis in asymptomatic adults: the BioImage study.
      ]. A recent study reported that CPB was superior to coronary calcium for risk stratification in women [
      • Gudmundsson E.F.
      • Björnsdottir G.
      • Sigurdsson S.
      • et al.
      Carotid plaque is strongly associated with coronary artery calcium and predicts incident coronary heart disease in a population-based cohort.
      ]; it is also detected at a younger age. CPB also has significant advantages compared with coronary calcium, because it can be measured repeatedly, to assess and adjust the effects of therapy. Serial assessment of plaque burden in conjunction with life-style and pharmacological treatment according to guidelines, called “Treating Arteries” (instead of merely treating risk factors), markedly improves therapy for atherosclerosis. In part this is because showing patients images of their plaque markedly improves compliance with lifestyle changes and medication [
      • Näslund U.
      • Ng N.
      • Lundgren A.
      • et al.
      Visualization of asymptomatic atherosclerotic disease for optimum cardiovascular prevention (VIPVIZA): a pragmatic, open-label, randomised controlled trial.
      ]. Among high-risk patients with asymptomatic carotid stenosis, implementation of “Treating Arteries” was associated with a >80% reduction of the 2-year risk of stroke/myocardial infarction/vascular death [
      • Spence J.D.
      • Coates V.
      • Li H.
      • et al.
      Effects of intensive medical therapy on microemboli and cardiovascular risk in asymptomatic carotid stenosis.
      ]. In prevention clinics across Argentina, “Treating Arteries” was implemented in 2010; among patients age >65, the annual risk of cardiovascular events declined from 5.85% to 2.35% between 2011 and 2015 [
      • Pérez H.A.
      • Adeoye A.O.
      • Aballay L.
      • Armando L.A.
      • García N.H.
      An intensive follow-up in subjects with cardiometabolic high-risk.
      ]. Patients in Germany who were treated with lipid-lowering drugs on the basis of a high CPB had a much lower risk of cardiovascular events over 3.9 years, than patients treated only on the basis of serum cholesterol: (5.4% vs 23%, respectively) [
      • Adams A.
      • Bojara W.
      • Romanens M.
      Effect of statin treatment in patients with advanced carotid atherosclerosis: an observational outcome study.
      ]. It has been supported that “Treating arteries” without measuring plaque would be like treating hypertension without measuring blood pressure.” [
      • Spence J.D.
      • Hackam D.G.
      Treating arteries instead of risk factors: a paradigm change in management of atherosclerosis.
      ].
      Studies using CPB identify new causes of atherosclerosis, either through genetic studies [
      • Spence J.D.
      Genetics of atherosclerosis: the power of plaque burden and progression: invited commentary on Dong C, Beecham A, Wang L, Blanton SH, Rundek T, Sacco RL. Follow-Up association study of linkage regions reveals multiple candidate genes for carotid plaque i.
      ], or studies of new risk factors such as toxic metabolites produced by the intestinal microbiome [
      • Bogiatzi C.
      • Gloor G.
      • Allen-Vercoe E.
      • et al.
      Metabolic products of the intestinal microbiome and extremes of atherosclerosis.
      ]. Such studies will lead to new therapies for atherosclerosis, and measurement of CPB markedly reduces sample size and duration of studies to evaluate such new therapies [
      • Ainsworth C.D.
      • Blake C.C.
      • Tamayo A.
      • Beletsky V.
      • Fenster A.
      • Spence J.D.
      3D ultrasound measurement of change in carotid plaque volume: a tool for rapid evaluation of new therapies.
      ].
      New automated methods based on machine learning for measuring TPA for measuring TPA, 3D carotid plaque burden [
      • Zhou R.
      • Fenster A.
      • Xia Y.
      • Spence J.D.
      • Ding M.
      Deep learning-based carotid media-adventitia and lumen-intima boundary segmentation from three-dimensional ultrasound images.
      ] and Vessel Wall Volume [
      • Zhou R.
      • Guo F.
      • Azarpazhooh M.R.
      • et al.
      A voxel-based fully convolution network and continuous max-flow for carotid vessel-wall-volume segmentation from 3D ultrasound images.
      ] (which can be measured in persons without plaque, so it can replace IMT) will make it much easier to implement this. These new methods are very fast, reliable, and reproducible [
      • Zhou R.
      • Guo F.
      • Azarpazhooh M.R.
      • et al.
      Deep learning-based measurement of total plaque area in B-mode ultrasound images.
      ]. Fig. 1 shows comparisons of automated with manual segmentation.
      Fig. 1
      Fig. 1Automated measurement of vessel wall volume.
      (A) Automated segmentation (yellow line) was very accurate compared to manual segmentation by experts (red line). Dice-similarity-coefficient (DSC) was 93.2 ± 3.0% for the Medial-Arterial Boundary in the common carotid artery and 91.9 ± 5.0% in the bifurcation. DSC for the Lumen-Intima Boundary was 89.5 ± 6.7% and 89.3 ± 6.8% for the Common Carotid Artery and the bifurcation respectively. Automated segmentation took less than 1 s for each side. (B) Relationships of the automated and manual VWV measurements for n = 302 3DUS images in the CAIN dataset. (a) Linear correlation (r = 0.876, p = 0.0001), and (b) Bland-Altman plot of the two sets of VWV measurements. The solid red line and the dash red lines represent the bias (−3.6 mm3) and mean ± 1.96 SD, respectively. Reproduced by permission of IEEE from: Zhou R, Guo F, Azarpazhooh MR, Spence JD, Ukwatta E, Ding M and Fenster A. A Voxel-Based Fully Convolution Network and Continuous Max-Flow for Carotid Vessel-Wall-Volume Segmentation From 3D Ultrasound Images. IEEE Trans Med Imaging. 2020; 39:2844–2855. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

      2.3 Transcranial Doppler detection of embolic signals in carotid artery disease

      Transcranial Doppler ultrasound (TCD) is a non-invasive technique that can be used to detect circulating emboli/intracranial embolism. These emboli appear as short-duration, high-intensity embolic signals/intracranial embolism and are accompanied by a characteristic chirping sound. TCD circulating emboli detection has been shown to have a high sensitivity and specificity in experimental studies [
      • Russell D.
      • Madden K.P.
      • Clark W.M.
      • Sandset P.M.
      • Zivin J.A.
      Detection of arterial emboli using Doppler ultrasound in rabbits.
      ], although in clinical practice care needs to be applied in distinguishing true embolic signals (also known as high-intensity transient signals) from artefact. Consensus criteria have been developed to allow this [
      • Ringelstein E.B.
      • Droste D.W.
      • Babikian V.L.
      • et al.
      Consensus on microembolus detection by TCD. International consensus group on microembolus detection.
      ].
      Studies have demonstrated that there were differences in the prevalence of embolic signal in the different stroke subtypes with a higher occurrence in large artery stroke in comparison with cardioembolic and lacunar stroke [
      • Kaposzta Z.
      • Young E.
      • Bath P.M.
      • Markus H.S.
      Clinical application of asymptomatic embolic signal detection in acute stroke: a prospective study.
      ]. In patients with recently symptomatic carotid stenosis (SCS), during a 1-h recording from the ipsilateral middle cerebral artery (MCA) embolic signals can be identified in about 40% of individuals [
      • Molloy J.
      • Markus H.S.
      Asymptomatic embolization predicts stroke and TIA risk in patients with carotid artery stenosis.
      ]. A higher prevalence has been reported in patients with more recent symptoms, SCS compared with ACS, and plaque imaging characteristics indicate a higher risk plaque [
      • Sitzer M.
      • Müller W.
      • Siebler M.
      • et al.
      Plaque ulceration and lumen thrombus are the main sources of cerebral microemboli in high-grade internal carotid artery stenosis.
      ].
      Prospective longitudinal studies have demonstrated that asymptomatic embolization encountered by TCD predicts future stroke risk in both SCS [
      • Molloy J.
      • Markus H.S.
      Asymptomatic embolization predicts stroke and TIA risk in patients with carotid artery stenosis.
      ] and ACS [
      • Markus H.S.
      • King A.
      • Shipley M.
      • et al.
      Asymptomatic embolisation for prediction of stroke in the Asymptomatic Carotid Emboli Study (ACES): a prospective observational study.
      ,
      • King A.
      • Markus H.S.
      Doppler embolic signals in cerebrovascular disease and prediction of stroke risk: a systematic review and meta-analysis.
      ,
      • Markus H.S.
      • Droste D.W.
      • Kaps M.
      • et al.
      Dual antiplatelet therapy with clopidogrel and aspirin in symptomatic carotid stenosis evaluated using Doppler embolic signal detection: the Clopidogrel and Aspirin for Reduction of Emboli in Symptomatic Carotid Stenosis (CARESS) trial.
      ,
      • Best L.M.J.
      • Webb A.C.
      • Gurusamy K.S.
      • Cheng S.F.
      • Richards T.
      Transcranial Doppler ultrasound detection of microemboli as a predictor of cerebral events in patients with symptomatic and asymptomatic carotid disease: a systematic review and meta-analysis.
      ,
      • Topakian R.
      • King A.
      • Kwon S.U.
      • Schaafsma A.
      • Shipley M.
      • Markus H.S.
      Ultrasonic plaque echolucency and emboli signals predict stroke in asymptomatic carotid stenosis.
      ].
      The effect is additive to that obtained by plaque imaging modalities such as carotid ultrasound [
      • Goertler M.
      • Baeumer M.
      • Kross R.
      • et al.
      Rapid decline of cerebral microemboli of arterial origin after intravenous acetylsalicylic acid.
      ]. It has been suggested that embolic signal detection may be a useful method to identify ACS patients at high risk who may particularly benefit from carotid endarterectomy (CEA) [
      • Best L.M.J.
      • Webb A.C.
      • Gurusamy K.S.
      • Cheng S.F.
      • Richards T.
      Transcranial Doppler ultrasound detection of microemboli as a predictor of cerebral events in patients with symptomatic and asymptomatic carotid disease: a systematic review and meta-analysis.
      ] although this needs to be proven in a randomized intervention study. Conversely, patients with an absence of embolic signals may do well with intensive medical management alone.
      Embolic signal detection has been used to evaluate the effectiveness of antithrombotic drugs in carotid artery disease [
      • Markus H.S.
      • Droste D.W.
      • Kaps M.
      • et al.
      Dual antiplatelet therapy with clopidogrel and aspirin in symptomatic carotid stenosis evaluated using Doppler embolic signal detection: the Clopidogrel and Aspirin for Reduction of Emboli in Symptomatic Carotid Stenosis (CARESS) trial.
      ]. Studies have shown that aspirin, clopidogrel but also statins reduce embolic signals [
      • Markus H.S.
      • Droste D.W.
      • Kaps M.
      • et al.
      Dual antiplatelet therapy with clopidogrel and aspirin in symptomatic carotid stenosis evaluated using Doppler embolic signal detection: the Clopidogrel and Aspirin for Reduction of Emboli in Symptomatic Carotid Stenosis (CARESS) trial.
      ,
      • Johnston S.C.
      • Easton J.D.
      • Farrant M.
      • et al.
      Clopidogrel and aspirin in acute ischemic stroke and high-risk TIA.
      ,
      • Safouris A.
      • Krogias C.
      • Sharma V.K.
      • et al.
      Statin pretreatment and microembolic signals in large artery atherosclerosis.
      ]. Combination antiplatelet regimens such as aspirin and clopidogrel were more effective than aspirin alone in randomized controlled trials [
      • Markus H.S.
      • Droste D.W.
      • Kaps M.
      • et al.
      Dual antiplatelet therapy with clopidogrel and aspirin in symptomatic carotid stenosis evaluated using Doppler embolic signal detection: the Clopidogrel and Aspirin for Reduction of Emboli in Symptomatic Carotid Stenosis (CARESS) trial.
      ,
      • Wang Y.
      • Wang Y.
      • Zhao X.
      • et al.
      Clopidogrel with aspirin in acute minor stroke or transient ischemic attack.
      ]. This paralleled subsequent studies demonstrating their greater effectiveness in preventing recurrent stroke after minor stroke and TIA [
      • Johnston S.C.
      • Easton J.D.
      • Farrant M.
      • et al.
      Clopidogrel and aspirin in acute ischemic stroke and high-risk TIA.
      ,
      • Wang Y.
      • Wang Y.
      • Zhao X.
      • et al.
      Clopidogrel with aspirin in acute minor stroke or transient ischemic attack.
      ], and reinforcing the evidence that embolic signal detections may be a useful surrogate marker to identify the efficacy of antithrombotic agents.

      2.4 Impact of contrast material

      As a complement to conventional duplex ultrasound, intravenous application of ultrasound contrast agents has greatly enriched sonographic imaging in vascular medicine [
      • Staub D.
      • Partovi S.
      • Imfeld S.
      • et al.
      Novel applications of contrast-enhanced ultrasound imaging in vascular medicine.
      ,
      • Staub D.
      • Schinkel A.F.L.
      • Coll B.
      • et al.
      Contrast-enhanced ultrasound imaging of the vasa vasorum: from early atherosclerosis to the identification of unstable plaques.
      ]. Contrast agents consist of small microbubbles, which circulate strictly intravascularly in the bloodstream for several minutes after application. Due to their non-linear reflection pattern, a contrast-specific ultrasound image is obtained, which enhances not only the vessel lumen but also the microcirculation in the vessel wall (vasa vasorum) including intraplaque neovascularization (IPN) [
      • Rafailidis V.
      • Li X.
      • Sidhu P.S.
      • Partovi S.
      • Staub D.
      Contrast imaging ultrasound for the detection and characterization of carotid vulnerable plaque.
      ].
      In the carotid artery, contrast-enhanced ultrasound (CEUS) is helpful in distinguishing complete vessel occlusion from very high-grade carotid stenosis. In addition, particularly hypoechogenic plaques can be well detected and surface irregularities and ulcerations of arteriosclerotic lesions can be better delineated [
      • Schinkel A.F.L.
      • Bosch J.G.
      • Staub D.
      • Adam D.
      • Feinstein S.B.
      Contrast-enhanced ultrasound to assess carotid intraplaque neovascularization.
      ,
      • Rafailidis V.
      • Chryssogonidis I.
      • Xerras C.
      • et al.
      A comparative study of color Doppler imaging and contrast-enhanced ultrasound for the detection of ulceration in patients with carotid atherosclerotic disease.
      ,
      • Rafailidis V.
      • Chryssogonidis I.
      • Xerras C.
      • et al.
      An ultrasonographic multiparametric carotid plaque risk index associated with cerebrovascular symptomatology: a study comparing color Doppler imaging and contrast-enhanced ultrasonography.
      ]. The most important value of CEUS lies in the detection and quantification of IPN [
      • Staub D.
      • Schinkel A.F.L.
      • Coll B.
      • et al.
      Contrast-enhanced ultrasound imaging of the vasa vasorum: from early atherosclerosis to the identification of unstable plaques.
      ], which is usually performed semiquantitatively [
      • Staub D.
      • Partovi S.
      • Schinkel A.F.L.
      • et al.
      Correlation of carotid artery atherosclerotic lesion echogenicity and severity at standard US with intraplaque neovascularization detected at contrast-enhanced US.
      ,
      • Staub D.
      • Patel M.B.
      • Tibrewala A.
      • et al.
      Vasa vasorum and plaque neovascularization on contrast-enhanced carotid ultrasound imaging correlates with cardiovascular disease and past cardiovascular events.
      ]. Such visual-based quantification has good intra- and interobserver variability, but a more objective, purely quantitative measurement of IPN ranging from measurements of maximal contrast enhancement to automated, computer-assisted quantification of the relative perfused area is desirable [
      • Li C.
      • He W.
      • Guo D.
      • et al.
      Quantification of carotid plaque neovascularization using contrast-enhanced ultrasound with histopathologic validation.
      ,
      • van den Oord S.C.H.
      • Akkus Z.
      • Bosch J.G.
      • et al.
      Quantitative contrast-enhanced ultrasound of intraplaque neovascularization in patients with carotid atherosclerosis.
      ,
      • Kaspar M.
      • Baumgartner I.
      • Staub D.
      • Drexel H.
      • Thalhammer C.
      Non-invasive ultrasound-based imaging of atherosclerosis.
      ].
      IPN on CEUS has been compared with the corresponding vascularization on histopathologic examination in patients with carotid stenosis before CEA showing a good correlation between the IPN and the extent of microvessels and inflammation within the plaque on histology [
      • Coli S.
      • Magnoni M.
      • Sangiorgi G.
      • et al.
      Contrast-enhanced ultrasound imaging of intraplaque neovascularization in carotid arteries: correlation with histology and plaque echogenicity.
      ,
      • Hoogi A.
      • Adam D.
      • Hoffman A.
      • Kerner H.
      • Reisner S.
      • Gaitini D.
      Carotid plaque vulnerability: quantification of neovascularization on contrast-enhanced ultrasound with histopathologic correlation.
      ,
      • Schmidt C.
      • Fischer T.
      • Rückert R.-I.
      • et al.
      Identification of neovascularization by contrast-enhanced ultrasound to detect unstable carotid stenosis.
      ].
      It has been demonstrated that hypoechogenic plaques, which were considered vulnerable on B-mode ultrasound, had higher IPN on CEUS than the more stable hyperechogenic lesions [
      • Staub D.
      • Partovi S.
      • Schinkel A.F.L.
      • et al.
      Correlation of carotid artery atherosclerotic lesion echogenicity and severity at standard US with intraplaque neovascularization detected at contrast-enhanced US.
      ,
      • Coli S.
      • Magnoni M.
      • Sangiorgi G.
      • et al.
      Contrast-enhanced ultrasound imaging of intraplaque neovascularization in carotid arteries: correlation with histology and plaque echogenicity.
      ]. Various retrospective studies of patients with carotid plaque revealed that those lesions with a higher embolic risk had increased plaque IPN on CEUS imaging. Thus, it was shown in a meta-analysis that the prevalence of IPN was higher in SCS compared with ACS patients [
      • van Engelen A.
      • Wannarong T.
      • Parraga G.
      • et al.
      Three-dimensional carotid ultrasound plaque texture predicts vascular events.
      ] and correlates with cardiovascular events [
      • Yan H.
      • Wu X.
      • He Y.
      • Staub D.
      • Wen X.
      • Luo Y.
      Carotid intraplaque neovascularization on contrast-enhanced ultrasound correlates with cardiovascular events and poor prognosis: a systematic review and meta-analysis.
      ]. Different prospective studies also demonstrated that in patients with a recent ischemic cerebrovascular event the risk of future ischemic stroke or TIA was significantly associated with IPN in carotid CEUS examination [
      • Li Z.
      • Xu X.
      • Ren L.
      • et al.
      Prospective study about the relationship between CEUS of carotid intraplaque neovascularization and ischemic stroke in TIA patients.
      ,
      • Cui L.
      • Xing Y.
      • Zhou Y.
      • et al.
      Carotid intraplaque neovascularisation as a predictive factor for future vascular events in patients with mild and moderate carotid stenosis: an observational prospective study.
      ]. IPN on CEUS imaging was also found to be predictive of significant and complex coronary artery disease and future cardiovascular events [
      • Mantella L.E.
      • Colledanchise K.N.
      • Hétu M.-F.
      • Feinstein S.B.
      • Abunassar J.
      • Johri A.M.
      Carotid intraplaque neovascularization predicts coronary artery disease and cardiovascular events.
      ].
      Carotid CEUS examination has the potential to improve risk stratification with respect to the occurrence of embolization by grading IPN in patients with carotid plaque and stenosis. This could be useful to monitor therapeutic interventions [
      • Magnoni M.
      • Ammirati E.
      • Moroni F.
      • Norata G.D.
      • Camici P.G.
      Impact of cardiovascular risk factors and pharmacologic treatments on carotid intraplaque neovascularization detected by contrast-enhanced ultrasound.
      ] and to better select patients with carotid stenosis, who benefit from a possible invasive therapy [
      • Rafailidis V.
      • Li X.
      • Sidhu P.S.
      • Partovi S.
      • Staub D.
      Contrast imaging ultrasound for the detection and characterization of carotid vulnerable plaque.
      ].

      3. Computed tomography

      3.1 Stenosis

      CTA has evolved along with the technological advances of CT hardware and software. Modern CTA, performed with multidetector high-speed CT hardware and evaluated with 3D reformatting software, accurately and reliably depicts carotid disease and allows for direct quantification of carotid stenosis in millimeters [
      • Anderson G.B.
      • Ashforth R.
      • Steinke D.E.
      • Ferdinandy R.
      • Findlay J.M.
      CT angiography for the detection and characterization of carotid artery bifurcation disease.
      ,
      • Leclerc X.
      • Godefroy O.
      • Pruvo J.P.
      • Leys D.
      Computed tomographic angiography for the evaluation of carotid artery stenosis.
      ,
      • Randoux B.
      • Marro B.
      • Koskas F.
      • et al.
      Carotid artery stenosis: prospective comparison of CT, three-dimensional gadolinium-enhanced MR, and conventional angiography.
      ,
      • Chen C.-J.
      • Lee T.-H.
      • Hsu H.-L.
      • et al.
      Multi-Slice CT angiography in diagnosing total versus near occlusions of the internal carotid artery: comparison with catheter angiography.
      ,
      • Koelemay M.J.W.
      • Nederkoorn P.J.
      • Reitsma J.B.
      • Majoie C.B.
      Systematic review of computed tomographic angiography for assessment of carotid artery disease.
      ,
      • Porsche C.
      • Walker L.
      • Mendelow D.
      • Birchall D.
      Evaluation of cross-sectional luminal morphology in carotid atherosclerotic disease by use of spiral CT angiography.
      ,
      • Dix J.E.
      • Evans A.J.
      • Kallmes D.F.
      • Sobel A.H.
      • Phillips C.D.
      Accuracy and precision of CT angiography in a model of carotid artery bifurcation stenosis.
      ].
      CTA is an anatomic study of arteries, allowing for direct evaluation of carotid stenosis. CTA is fast, with images of the head and neck acquired over approximately 5–15 s during contrast injection. 512 × 512 memory matrix multidetector CT scanners allow acquisitions with near-isotopic spatial resolution and an effective section thickness as small as 0.5 mm [
      • Napoli A.
      • Fleischmann D.
      • Chan F.P.
      • et al.
      Computed tomography angiography: state-of-the-art imaging using multidetector-row technology.
      ]. For evaluation of carotid arteries and the cerebral vasculature, it is possible to narrow the nominal section thickness to obtain a submillimetric dataset. This ability, combined with 3D image rendering, provides excellent accuracy for the measurement of the degree of stenosis [
      • Napoli A.
      • Fleischmann D.
      • Chan F.P.
      • et al.
      Computed tomography angiography: state-of-the-art imaging using multidetector-row technology.
      ]. In light of the relative benefits of CTA in reference to safety, time, and related lower cost than DSA, it is compelling to use CTA when the indication for angiography is not to deliver a therapeutic intervention such as stenting but to accurately characterize the degree of stenosis. Venous filling is not an artefact for neck carotid imaging, because arteries are easily recognized as distinct from veins.
      CTA evaluation is mainly based on axial sections and curved planar reformations (CPR). CTA has been shown to have a pooled sensitivity of 95% and specificity of 98% for the detection of >70% stenoses [
      • Prokop M.
      • Waaijer A.
      • Kreuzer S.
      CT angiography ofthe carotid arteries.
      ]. There are advantages of quantifying carotid stenosis by direct millimeter measurements instead of or in addition to the well-known North American Symptomatic Carotid Endarterectomy Trial (NASCET)–style ratio calculations [
      • Bartlett E.S.
      • Walters T.D.
      • Symons S.P.
      • Fox A.J.
      Quantification of carotid stenosis on CT angiography.
      ]. Multi-slice CTA can in addition detect tandem stenoses in the region of the carotid origin from the aorta, the carotid siphon, and the intracranial portion of the carotids. CT is able to provide a comprehensive evaluation of patients with acute stroke by using a combined approach of pre-contrast CT to detect hemorrhage and manifest infarction, perfusion CT measurements to differentiate between penumbra and infarct, and CTA to detect the occluded vessel as well as potential concomitant carotid abnormalities.

      3.2 Features of vulnerability

      Atherosclerotic disease is a complex, heterogeneous, and multifactorial condition with several types of components in the same plaque. The role of plaque imaging is to identify those imaging biomarker features of carotid plaque that are related to vulnerable plaque [
      • Saba L.
      • Saam T.
      • Jäger H.R.
      • et al.
      Imaging biomarkers of vulnerable carotid plaques for stroke risk prediction and their potential clinical implications.
      ,
      • Saba L.
      • Anzidei M.
      • Marincola B.C.
      • et al.
      Imaging of the carotid artery vulnerable plaque.
      ,
      • Cademartiri F.
      • Balestrieri A.
      • Cau R.
      • et al.
      Insight from imaging on plaque vulnerability: similarities and differences between coronary and carotid arteries—implications for systemic therapies.
      ,
      • Cau R.
      • Flanders A.
      • Mannelli L.
      • et al.
      Artificial intelligence in computed tomography plaque characterization: a review.
      ]. In particular, CTA thanks to its spatial resolution is able to assess the carotid artery lumen and the arterial wall.
      A key feature of vulnerable carotid plaque is Intraplaque hemorrhage (IPH), which is defined by the accumulation of blood components within the carotid plaque [
      • Michel J.-B.
      • Virmani R.
      • Arbustini E.
      • Pasterkamp G.
      Intraplaque haemorrhages as the trigger of plaque vulnerability.
      ]. Regarding the pathogenesis of IPH, most of the authors suggest that it is linked to the rupture of neovessels or plaque rupture itself, and some trigger events including inflammation, metabolic diseases or diabetes may precipitate this condition [
      • Michel J.-B.
      • Virmani R.
      • Arbustini E.
      • Pasterkamp G.
      Intraplaque haemorrhages as the trigger of plaque vulnerability.
      ]. IPH represents the strongest imaging feature associated with the occurrence of stroke [
      • Saam T.
      • Hetterich H.
      • Hoffmann V.
      • et al.
      Meta-analysis and systematic review of the predictive value of carotid plaque hemorrhage on cerebrovascular events by magnetic resonance imaging.
      ], and it is also more common in carotid artery ipsilateral to embolic stroke of undetermined source [
      • Singh N.
      • Moody A.R.
      • Panzov V.
      • Gladstone D.J.
      Carotid intraplaque hemorrhage in patients with embolic stroke of undetermined source.
      ]. Conflicting results have been reported about the role of CT to detect this feature. However, studies suggest that CTA is able to discriminate the presence of IPH, both directly according to attenuation at 25 HU [
      • Saba L.
      • Francone M.
      • Bassareo P.P.
      • et al.
      CT attenuation analysis of carotid intraplaque hemorrhage.
      ] and indirectly with the presence of calcified rim and soft internal plaque [
      • Eisenmenger L.B.
      • Aldred B.W.
      • Kim S.-E.
      • et al.
      Prediction of carotid intraplaque hemorrhage using adventitial calcification and plaque thickness on CTA.
      ].
      The thin fibrous cap with a lipid-rich necrotic core (LRNC) represents one of the most important features of the carotid artery vulnerable plaque model. In particular, LNRC is considered a collection of heterogeneous tissue composed of cholesterol crystals and necrotic debris of apoptotic cells [
      • Cai J.
      • Hatsukami T.S.
      • Ferguson M.S.
      • et al.
      In vivo quantitative measurement of intact fibrous cap and lipid-rich necrotic core size in atherosclerotic carotid plaque: comparison of high-resolution, contrast-enhanced magnetic resonance imaging and histology.
      ].
      The fibrous cap is a layer of fibrous connective tissue that contains macrophages and smooth-muscle cells, and particularly the morphology and thickness of the fibrous cap are indicative of rupture [
      • Cury R.C.
      • Houser S.L.
      • Furie K.L.
      • et al.
      Vulnerable plaque detection by 3.0 tesla magnetic resonance imaging.
      ]. These two imaging features are associated with the risk of stroke, especially when a thin fibrous cap covers a large LRNC [
      • Xu D.
      • Hippe D.S.
      • Underhill H.R.
      • et al.
      Prediction of high-risk plaque development and plaque progression with the carotid atherosclerosis score.
      ].
      In addition, the LRNC size correlates with future ipsilateral carotid symptoms [
      • Saba L.
      • Yuan C.
      • Hatsukami T.S.
      • et al.
      Carotid artery wall imaging: perspective and guidelines from the ASNR vessel wall imaging study group and expert consensus recommendations of the American society of neuroradiology.
      ]. CT can be used to visualize lipid components of the LNRC, thanks to lipid tissue attenuation properties, but may more hardly discriminate between LRNC and IPH, given the attenuation values of these two features [
      • Saba L.
      • Yuan C.
      • Hatsukami T.S.
      • et al.
      Carotid artery wall imaging: perspective and guidelines from the ASNR vessel wall imaging study group and expert consensus recommendations of the American society of neuroradiology.
      ]. Similarly, the evaluation of the fibrous cap with CT is not considered optimal because of the artefact related to halo-effect and edge-blur [
      • Saba L.
      • Agarwal N.
      • Cau R.
      • et al.
      Review of imaging biomarkers for the vulnerable carotid plaque.
      ].
      Another feature of plaque vulnerability is inflammation of the carotid artery plaque. Different types of inflammatory cells have been identified in the carotid plaque usually located in the plaque shoulder, cap, or both with a role in plaque “instability” [
      • Saba L.
      • Agarwal N.
      • Cau R.
      • et al.
      Review of imaging biomarkers for the vulnerable carotid plaque.
      ,
      • Cerrone G.
      • Fanni D.
      • Lai M.L.
      • et al.
      Plasma cells in the carotid plaque: occurrence and significance.
      ]. Beyond the presence of macrophages, plasma cells are also associated with the risk of rupture and the occurrence of cardiovascular events [
      • Cerrone G.
      • Fanni D.
      • Lai M.L.
      • et al.
      Plasma cells in the carotid plaque: occurrence and significance.
      ].
      Similarly, plaque neovascularization is considered a marker of plaque vulnerability, which is related to newly formed neovessel arising into the intima and is associated with plaque activity [
      • Saba L.
      • Yuan C.
      • Hatsukami T.S.
      • et al.
      Carotid artery wall imaging: perspective and guidelines from the ASNR vessel wall imaging study group and expert consensus recommendations of the American society of neuroradiology.
      ]. The presence of neovascularization in carotid plaque represents an element of instability because these microvessels are prone to rupture due to their immature and imperfect endothelial structure [
      • McCarthy M.J.
      • Loftus I.M.
      • Thompson M.M.
      • et al.
      Angiogenesis and the atherosclerotic carotid plaque: an association between symptomatology and plaque morphology.
      ]. CTA can identify the presence and the degree of neovascularization thank its ability to detect contrast plaque enhancement [
      • Saba L.
      • Lai M.L.
      • Montisci R.
      • et al.
      Association between carotid plaque enhancement shown by multidetector CT angiography and histologically validated microvessel density.
      ].
      Beyond plaque composition, vulnerable plaques tend to be associated with plaque surface morphology (i.e. smooth, irregular, or ulcerated) [
      • Saba L.
      • Anzidei M.
      • Marincola B.C.
      • et al.
      Imaging of the carotid artery vulnerable plaque.
      ]. In particular, the presence of ulceration, defined as an intimal defect larger than 1 mm in width [
      • Saba L.
      • Yuan C.
      • Hatsukami T.S.
      • et al.
      Carotid artery wall imaging: perspective and guidelines from the ASNR vessel wall imaging study group and expert consensus recommendations of the American society of neuroradiology.
      ], is considered a risk feature for cardiovascular events [
      • Barnett H.J.
      • Taylor D.W.
      • Eliasziw M.
      • et al.
      Benefit of carotid endarterectomy in patients with symptomatic moderate or severe stenosis. North American Symptomatic Carotid Endarterectomy Trial Collaborators.
      ]. The carotid plaque surface morphology can be better assessed with CTA in comparison with other non-invasive imaging modalities, as demonstrated by Saba et al. [
      • Saba L.
      • Caddeo G.
      • Sanfilippo R.
      • Montisci R.
      • Mallarini G.
      CT and ultrasound in the study of ulcerated carotid plaque compared with surgical results: potentialities and advantages of multidetector row CT angiography.
      ,
      • Saba L.
      • Caddeo G.
      • Sanfilippo R.
      • Montisci R.
      • Mallarini G.
      Efficacy and sensitivity of axial scans and different reconstruction methods in the study of the ulcerated carotid plaque using multidetector-row CT angiography: comparison with surgical results.
      ].
      Also, carotid plaque volume is a crucial determinant of plaque vulnerability. Rozie et al. demonstrated that plaque volume and the relative subcomponents of the plaque are associated with plaque vulnerability and stroke [
      • Rozie S.
      • de Weert T.T.
      • de Monyé C.
      • et al.
      Atherosclerotic plaque volume and composition in symptomatic carotid arteries assessed with multidetector CT angiography; relationship with severity of stenosis and cardiovascular risk factors.
      ]. Thanks to its excellent spatial resolution, CTA can easily evaluate this parameter [
      • Saba L.
      • Saam T.
      • Jäger H.R.
      • et al.
      Imaging biomarkers of vulnerable carotid plaques for stroke risk prediction and their potential clinical implications.
      ].
      Among the multiple parameters that have been indicated as responsible for an increased vulnerability, conflicting results have emerged in the identification of a role for calcium. Emerging research has suggested various mechanisms in calcium deposition leading to different phenotypes of carotid plaque calcification [
      • Pini R.
      • Faggioli G.
      • Fittipaldi S.
      • et al.
      Relationship between calcification and vulnerability of the carotid plaques.
      ,
      • Yang J.
      • Pan X.
      • Zhang B.
      • et al.
      Superficial and multiple calcifications and ulceration associate with intraplaque hemorrhage in the carotid atherosclerotic plaque.
      ,
      • Eisenmenger L.B.
      • Aldred B.W.
      • Kim S.E.
      • et al.
      Prediction of carotid intraplaque hemorrhage using adventitial calcification and plaque thickness on CTA.
      ,
      • Saba L.
      • Chen H.
      • Cau R.
      • et al.
      Impact analysis of different CT configurations of carotid artery plaque calcifications on cerebrovascular events.
      ,
      • Saba L.
      • Nardi V.
      • Cau R.
      • et al.
      Carotid artery plaque calcifications: lessons from histopathology to diagnostic imaging.
      ]. Yang et al. investigated the association between calcium configurations and ulceration with IPH, demonstrating that superficial, multiple, and thin calcifications were associated with IPH. The authors concluded that the size and location may represent a marker of high-risk plaque [
      • Yang J.
      • Pan X.
      • Zhang B.
      • et al.
      Superficial and multiple calcifications and ulceration associate with intraplaque hemorrhage in the carotid atherosclerotic plaque.
      ]. Table 3 summarizes the CT features of plaque vulnerability and its strengths and limitations.
      Table 3CT features of plaque vulnerability and its strengths and limitations.
      Imaging featuresSupporting evidenceLimitationsGeneral limitations
      Intraplaque hemorrhageDirectly: attenuation values ≤ 25Moderate supporting evidenceSimilar HU attenuation values between soft plaque components
      • •Radiation dose delivered to the patients
      • •Potential side effect
      • •The limit tissue constrast between soft plaque components
      • •Overstimates the degree of the stenosis due to calcium deposits
      Indirectly: calcified rim and soft internal plaque
      Lipid-rich necrotic corePresence of soft plaque componentsConflicting supporting evidenceArtefact related to halo-effect and edge-blur
      Plaque inflammationPresence of contrast plaque enhancementWeak supporting evidence
      NeovascularizationPresence of contrast plaque enhancementModerate supporting evidence
      Plaque surface morphologyAlterations of the luminal surface on the luminal profile of the plaqueStrong supporting evidencePresence of a halo or edge blur may hinder detection of smaller ulcerations
      Plaque volume and compositionSize of the carotid plaque with its subcomponentsStrong supporting evidenceLimit tissue contrast attenuation in some plaque subcomponents
      CalcificationsSize and morphology of calcium depositsStrong supporting evidence

      3.3 Ancillary findings in carotid imaging

      While evaluation of vessel patency and plaque characteristics remains the main reason to perform CT/MR-imaging of the carotid arteries, a variety of ancillary findings can be encountered (Table 4). Some are merely incidental findings with no further clinical relevance, while others represent a different etiology of the patient's complaints with clear implications for further treatment and prognosis. Although a detailed scope of all possible ancillary findings is outside the scope of this paper, some important entities will now further be discussed.
      Table 4Ancillary findings.
      ConditionComments
      CongenitalAgenesis, aplasia or hypoplasia of ICA
      Inflammatory & infectious conditionsCarotidynia
      Giant cell arteritis
      Takayasu arteritis
      Post-radiation arteritis
      Carotid dissectionTraumatic or spontaneous
      Consider underlying condition with spontaneous (e.g. FMD)
      Carotid web & floating thrombusAssociated with increased stroke risk

      3.3.1 Carotid dissection

      As in the aorta, a dissection of the carotid artery wall constitutes a disruption of the carotid intima layer, with blood flowing into the vessel wall and the creation of a true and false lumen [
      • Baumgartner R.W.
      • Arnold M.
      • Baumgartner I.
      • et al.
      Carotid dissection with and without ischemic events: local symptoms and cerebral artery findings.
      ].
      A carotid artery dissection can be spontaneous or post-traumatic [
      • Baumgartner R.W.
      • Arnold M.
      • Baumgartner I.
      • et al.
      Carotid dissection with and without ischemic events: local symptoms and cerebral artery findings.
      ]. When spontaneous, an underlying condition must be ruled-out, which can include entities like fibromuscular dysplasia, Marfan syndrome, and Ehler-Danlos syndrome [
      • Baumgartner R.W.
      • Arnold M.
      • Baumgartner I.
      • et al.
      Carotid dissection with and without ischemic events: local symptoms and cerebral artery findings.
      ,
      • Hakimi R.
      • Sivakumar S.
      Imaging of carotid dissection.
      ]. It is important to scrutinize the other cervical arteries as well, as they may exhibit morphological changes contributing to a correct diagnosis (e.g. signs of fibromuscular dysplasia in the contralateral artery) [
      • Kadian-Dodov D.
      • Gornik H.L.
      • Gu X.
      • et al.
      Dissection and aneurysm in patients with fibromuscular dysplasia: findings from the U.S. Registry for FMD.
      ].
      The pathophysiology of a carotid dissection explains its imaging findings [
      • Hakimi R.
      • Sivakumar S.
      Imaging of carotid dissection.
      ]. In contrast to the aorta, the dissection flap is seldom seen in a carotid artery dissection, as the false lumen usually thromboses and creates a semicircular non-enhancing soft tissue density surrounding the true lumen (Fig. 2) [
      • Hakimi R.
      • Sivakumar S.
      Imaging of carotid dissection.
      ]. This makes MR-imaging especially useful, as this thrombus will lead to a hyperintense signal on fat-suppressed T1-weighted images due to blood breakdown products. It can be problem-solving in cases in which the presence and extent of the dissection can be difficult to assess on CT alone, as the difference in contrast between the wall hematoma and surrounding tissues can be limited [
      • Hakimi R.
      • Sivakumar S.
      Imaging of carotid dissection.
      ].
      Fig. 2
      Fig. 2Carotid artery dissection (A and B) in a 49-year-old female patient.
      The CTA shows the filiform lumen in the right ICA (A) that is confirmed by the MR B). Infective pseudo-aneurysm in 63-year-old male patient (C and D). The CTA shows the contrast material due to into the pseudoaneurysm (white open arrows, C) and the volume rendered image (D) confirms the spatial relationship.
      A carotid dissection usually appears in the supra-bulbar internal carotid artery. In many cases, it will remain limited to the extracranial segment, but extension into the skull base can occur [
      • Hakimi R.
      • Sivakumar S.
      Imaging of carotid dissection.
      ,
      • Kadian-Dodov D.
      • Gornik H.L.
      • Gu X.
      • et al.
      Dissection and aneurysm in patients with fibromuscular dysplasia: findings from the U.S. Registry for FMD.
      ].

      3.3.2 Carotid web & thrombus

      A carotid web is identified as a small, (curvi)lineair soft tissue density protruding into the carotid lumen usually at the level of the carotid bulb [
      • Madaelil T.P.
      • Grossberg J.A.
      • Nogueira R.G.
      • et al.
      Multimodality imaging in carotid web.
      ,
      • Wojcik K.
      • Milburn J.
      • Vidal G.
      • Steven A.
      Carotid webs: radiographic appearance and significance.
      ]. According to some authors, it represents a variant of fibromuscular dysplasia and is associated with an increased risk for stroke, especially in younger patients without classic vascular risk factors [
      • Priyadarshni S.
      • Neralla A.
      • Reimon J.
      • Smithson S.
      Carotid webs: an unusual presentation of fibromuscular dysplasia.
      ].
      A thrombus presents as a non-enhancing central structure surrounded by flowing blood (the so-called “donut” sign). While rare, its presence is important as it is associated with an increased risk for stroke, but also with conditions leading to a hypercoagulable state like malignancy, infections, and pregnancy [
      • Madaelil T.P.
      • Grossberg J.A.
      • Nogueira R.G.
      • et al.
      Multimodality imaging in carotid web.
      ,
      • Wojcik K.
      • Milburn J.
      • Vidal G.
      • Steven A.
      Carotid webs: radiographic appearance and significance.
      ].

      3.3.3 Inflammatory and infectious conditions

      Carotid vasculitis can be defined as the inflammation of carotid artery walls with or without necrosis, leading to stenosis or occlusion of the lumen [
      • Abdel Razek A.A.K.
      • Alvarez H.
      • Bagg S.
      • Refaat S.
      • Castillo M.
      Imaging spectrum of CNS vasculitis.
      ]. Vasculitis may be associated with systemic connective tissue disorders or may be secondary to infection, malignancy, drugs, or radiation therapy [
      • Abdel Razek A.A.K.
      • Alvarez H.
      • Bagg S.
      • Refaat S.
      • Castillo M.
      Imaging spectrum of CNS vasculitis.
      ]. For a correct diagnosis, relevant laboratory tests are also required. The 2012 Chapel Hill Consensus Conference defined different types of vasculitis in terms of (a) the size of the involved arteries and (b) associated pathologic lesions [
      • Jennette J.C.
      • Falk R.J.
      • Bacon P.A.
      • et al.
      Revised international Chapel Hill consensus conference nomenclature of vasculitides.
      ]. The most frequent vasculitis involving carotid arteries are Takayasu arteritis and Giant cell arteritis [
      • Abdel Razek A.A.K.
      • Alvarez H.
      • Bagg S.
      • Refaat S.
      • Castillo M.
      Imaging spectrum of CNS vasculitis.
      ,
      • Jennette J.C.
      • Falk R.J.
      • Bacon P.A.
      • et al.
      Revised international Chapel Hill consensus conference nomenclature of vasculitides.
      ]. Infectious extracranial carotid disease is rare and usually caused by Staphylococcus aureus, Salmonella, and streptococcus species. When present, it can manifest as an infected aneurysm with a focal weakinging of the wall, development of a pseudo-aneurysm, and increased rupture risk (Fig. 2).
      With CTA/MRA imaging, signs of carotid vasculitis are vessel wall thickening (mostly concentric representing a key parameter in the differential diagnosis) and contrast enhancement. Usually there is no preference for the involvement of the carotid bifurcation (different from atherosclerotic disease). In the case of active vasculitis contrast enhancement of the thickened vessel wall may be seen on both CT and MR [
      • Abdel Razek A.A.K.
      • Alvarez H.
      • Bagg S.
      • Refaat S.
      • Castillo M.
      Imaging spectrum of CNS vasculitis.
      ].

      3.3.4 Other

      Any other condition or anomaly that is encountered during a carotid examination must be reported. These include rare instances like carotid body tumours or any other condition that influences clinical management.

      4. Magnetic resonance imaging

      4.1 Stroke risk assessment and characterization of low-grade carotid atherosclerosis

      Risk assessment of carotid atherosclerotic plaque for cerebrovascular ischemic events has historically relied on angiographic measures of stenosis, with thresholds for revascularization defined by randomized clinical trials that date back to the early 1990's [
      • Barnett H.J.
      • Taylor D.W.
      • Eliasziw M.
      • et al.
      Benefit of carotid endarterectomy in patients with symptomatic moderate or severe stenosis. North American Symptomatic Carotid Endarterectomy Trial Collaborators.
      ,
      MRC European Carotid Surgery Trial: interim results for symptomatic patients with severe (70-99%) or with mild (0-29%) carotid stenosis. European Carotid Surgery Trialists' Collaborative Group.
      ,
      • Walker M.D.
      • Marler J.R.
      • Goldstein M.
      • et al.
      Endarterectomy for asymptomatic carotid artery stenosis.
      ]. The established threshold for SCS is 70%, although revascularization is often considered for stenosis beginning at 50% when symptomatic and 60% when asymptomatic [
      • Barnett H.J.
      • Taylor D.W.
      • Eliasziw M.
      • et al.
      Benefit of carotid endarterectomy in patients with symptomatic moderate or severe stenosis. North American Symptomatic Carotid Endarterectomy Trial Collaborators.
      ,
      MRC European Carotid Surgery Trial: interim results for symptomatic patients with severe (70-99%) or with mild (0-29%) carotid stenosis. European Carotid Surgery Trialists' Collaborative Group.
      ,
      • Halliday A.
      • Harrison M.
      • Hayter E.
      • et al.
      10-year stroke prevention after successful carotid endarterectomy for asymptomatic stenosis (ACST-1): a multicentre randomised trial.
      ,
      • Messas E.
      • Goudot G.
      • Halliday A.
      • et al.
      Management of carotid stenosis for primary and secondary prevention of stroke: state-of-the-art 2020: a critical review.
      ]. Stenosis has worked well in these studies considering it is a surrogate for plaque burden, which is strongly associated with ischemic stroke risk [
      • Spence J.D.
      Measurement of carotid plaque burden.
      ]. However, there have been substantial technical advances in our ability to identify features of atherosclerotic plaque that can improve our precision for stratifying risk [
      • Gupta A.
      • Baradaran H.
      • Schweitzer A.D.
      • et al.
      Carotid plaque MRI and stroke risk: a systematic review and meta-analysis.
      ,
      • Wasserman B.A.
      Advanced contrast-enhanced MRI for looking beyond the lumen to predict stroke.
      ,
      • Wasserman B.A.
      • Wityk R.J.
      • Trout H.H.
      • Virmani R.
      Low-grade carotid stenosis: looking beyond the lumen with MRI.
      ]. This is especially important for atherosclerotic plaques that fall under the thresholds for angiographic detection of risk. Risk estimation for stroke from a plaque causing less than 50% stenosis must be a priority considering the high prevalence of low-grade carotid stenosis in the community [
      • Wasserman B.A.
      • Wityk R.J.
      • Trout H.H.
      • Virmani R.
      Low-grade carotid stenosis: looking beyond the lumen with MRI.
      ]. For example, in the Cardiovascular Health Study detectable carotid stenosis was present in 62% of women and 75% of men aged ≥65 years, with only 7% of men and 5% of women having stenosis ≥50% [
      • O'Leary D.H.
      • Polak J.F.
      • Kronmal R.A.
      • et al.
      Distribution and correlates of sonographically detected carotid artery disease in the cardiovascular Health study. The CHS collaborative research group.
      ]. Risk analysis of carotid plaque must also consider the accommodation of atherosclerotic plaque formation by flow-mediated outward remodeling regulated by endothelial cells to preserve lumen caliber. This endothelial response is overcome once plaque size reaches a threshold and angiographic stenosis becomes detectable. For example, there is evidence that angiographic narrowing of the extracranial internal carotid artery is not detected until plaque burden reaches 61.9% [
      • Astor B.C.
      • Sharrett A.R.
      • Coresh J.
      • Chambless L.E.
      • Wasserman B.A.
      Remodeling of carotid arteries detected with MR imaging: atherosclerosis risk in communities carotid MRI study.
      ] or 63.1% [
      • Babiarz L.S.
      • Astor B.
      • Mohamed M.A.
      • Wasserman B.A.
      Comparison of gadolinium-enhanced cardiovascular magnetic resonance angiography with high-resolution black blood cardiovascular magnetic resonance for assessing carotid artery stenosis.
      ] area stenosis measured on black blood MRI exams, highlighting the large burden of plaque that can exist in low-grade carotid atherosclerosis.

      4.2 High-risk carotid plaque features detectable on MRI

      Based on histopathologic validation studies [
      • Cai J.M.
      • Hatsukami T.S.
      • Ferguson M.S.
      • Small R.
      • Polissar N.L.
      • Yuan C.
      Classification of human carotid atherosclerotic lesions with in vivo multicontrast magnetic resonance imaging.
      ], MRI has been shown to have high accuracy in detecting key high-risk carotid plaque features. For example, using a multi-sequence protocol with a carotid coil, MRI can identify the presence of an LRNC, thinning/rupture of the fibrous cap, ulceration, and IPH, all of which are strong predictors of future stroke risk [
      • Gupta A.
      • Baradaran H.
      • Schweitzer A.D.
      • et al.
      Carotid plaque MRI and stroke risk: a systematic review and meta-analysis.
      ]. Although the literature strongly supports a potential role for multi-contrast, multi-sequence MRI to aid in risk stratification before stroke and identify culprit lesions after stroke, adoption of these techniques has been limited in the context of acute stroke imaging due to exam length, need for dedicated coil hardware and/or gadolinium, and complexity in image interpretation [
      • Saba L.
      • Moody A.R.
      • Saam T.
      • et al.
      Vessel wall-imaging biomarkers of carotid plaque vulnerability in stroke prevention trials: a viewpoint from the carotid imaging consensus group.
      ].
      In recent years, converging evidence has identified the use of a single T1-weighted, fat-suppressed sequence to identify IPH as a particularly valuable imaging strategy to incorporate into clinical practice [
      • Saba L.
      • Saam T.
      • Jäger H.R.
      • et al.
      Imaging biomarkers of vulnerable carotid plaques for stroke risk prediction and their potential clinical implications.
      ]. This sequence includes an inversion pulse to suppress the blood in the lumen. In such an approach, IPH can be identified by the presence of T1-hyperintense signal within carotid plaque when noted to be brighter than the signal intensity of adjacent background skeletal muscle [
      • Saba L.
      • Brinjikji W.
      • Spence J.D.
      • et al.
      Roadmap consensus on carotid artery plaque imaging and impact on therapy strategies and guidelines: an international, multispecialty, expert review and position statement.
      ,
      • Cappendijk V.C.
      • Cleutjens K.B.J.M.
      • Heeneman S.
      • et al.
      In vivo detection of hemorrhage in human atherosclerotic plaques with magnetic resonance imaging.
      ]. A recent meta-analysis of individual patient data of 560 patients from 7 cohort studies showed the annualized rate of ipsilateral stroke in those with carotid IPH to be markedly increased compared to those without IPH across all stenosis severity levels, including those with <50% stenosis [
      • Schindler A.
      • Schinner R.
      • Altaf N.
      • et al.
      Prediction of stroke risk by detection of hemorrhage in carotid plaques: meta-analysis of individual patient data.
      ]. For this reason, MRI-based IPH identification has significant promise not only in identifying patients who may benefit the most from surgical revascularization procedures, but also in identifying culprit low-grade plaques responsible for ischemic strokes which would otherwise be characterized as cryptogenic in nature [
      • Kamel H.
      • Merkler A.E.
      • Iadecola C.
      • Gupta A.
      • Navi B.B.
      Tailoring the approach to embolic stroke of undetermined source: a review.
      ,
      • Kamel H.
      • Navi B.B.
      • Merkler A.E.
      • et al.
      Reclassification of ischemic stroke etiological subtypes on the basis of high-risk nonstenosing carotid plaque.
      ]. Patients with an absence of IPH may do well with intensive medical management alone. Given the promise of carotid IPH as a clinically useful MRI risk marker, randomized stroke prevention trials using IPH as a selection criterion will be needed to establish whether there is evidence to support the widespread adoption of this approach in clinical treatment decision-making.

      4.3 Role of MRI in assessment of intracranial carotid and aortic arch atherosclerosis

      MRI technology has been shown to have an increasing clinical role in the evaluation of atherosclerotic plaques in locations other than the extracranial carotid artery. MRA has high sensitivity and specificity for identification of stenoses >50% involving the intracranial ICA [
      • Baradaran H.
      • Patel P.
      • Gialdini G.
      • et al.
      Quantifying intracranial internal carotid artery stenosis on MR angiography.
      ,
      • Hirai T.
      • Korogi Y.
      • Ono K.
      • et al.
      Prospective evaluation of suspected stenoocclusive disease of the intracranial artery: combined MR angiography and CT angiography compared with digital subtraction angiography.
      ] and the MCA [
      • Degnan A.J.
      • Gallagher G.
      • Teng Z.
      • Lu J.
      • Liu Q.
      • Gillard J.H.
      MR angiography and imaging for the evaluation of middle cerebral artery atherosclerotic disease.
      ]. It is routinely utilized in clinical practice to identify patients suspected of harboring intracranial stenosis. Limitations of intracranial MRA include long acquisition times and overestimation of the degree of stenosis because of flow artefact.
      A major advantage of MRI sequences in the assessment of intracranial plaques is the concomitant acquisition with parenchymal brain imaging. Recent MRI techniques also allow for further characterize plaque composition and its hemodynamic effects. High-resolution vessel wall MRA provides further characterization of the intracranial arterial wall and pathology by suppressing signal from intravascular blood. It is increasingly used to differentiate among different causes of intracranial steno-occlusive disease and to identify/characterize plaques causing minimal or no narrowing on luminal imaging in patients with otherwise unexplained ipsilateral stroke [
      • Lehman V.T.
      • Brinjikji W.
      • Kallmes D.F.
      • et al.
      Clinical interpretation of high-resolution vessel wall MRI of intracranial arterial diseases.
      ,
      • Lindenholz A.
      • van der Kolk A.G.
      • Zwanenburg J.J.M.
      • Hendrikse J.
      The use and pitfalls of intracranial vessel wall imaging: how we do it.
      ,
      • Mandell D.M.
      • Mossa-Basha M.
      • Qiao Y.
      • et al.
      Intracranial vessel wall MRI: principles and expert consensus recommendations of the American society of neuroradiology.
      ,
      • Wang Y.
      • Liu X.
      • Wu X.
      • Degnan A.J.
      • Malhotra A.
      • Zhu C.
      Culprit intracranial plaque without substantial stenosis in acute ischemic stroke on vessel wall MRI: a systematic review.
      ]. The hemodynamic effects of intracranial plaques can be measured with quantitative MRA which combines traditional MRA-Time-of-Flight (TOF) and contrast-enhanced (CE)-MRA. This technique allows quantification of the hemodynamic significance of a plaque, which may not necessarily correlate with the degree of narrowing [
      • Zhao M.
      • Amin-Hanjani S.
      • Ruland S.
      • Curcio A.P.
      • Ostergren L.
      • Charbel F.T.
      Regional cerebral blood flow using quantitative MR angiography.
      ].
      Aortic arch atheroma has been recognized as a potential cause of emboli in patients with cryptogenic stroke and MRI-based multicontrast plaque imaging was used to recognize vulnerable aortic arch plaques [
      • Wehrum T.
      • Dragonu I.
      • Strecker C.
      • et al.
      Aortic atheroma as a source of stroke - assessment of embolization risk using 3D CMR in stroke patients and controls.
      ]. The addition of 4D flow measurements identifies potential embolization pathways to the brain and is especially useful to suggest possible retrograde embolization in patients with vulnerable plaques located in the proximal descending aorta immediately distal to the origin of the large extracranial arteries [
      • Wehrum T.
      • Dragonu I.
      • Strecker C.
      • et al.
      Aortic atheroma as a source of stroke - assessment of embolization risk using 3D CMR in stroke patients and controls.
      ].

      5. Other techniques

      5.1 PET

      PET enables molecular imaging of biological and biochemical processes in vivo, whereas hybrid PET/CT [
      • Cocker M.S.
      • Spence J.D.
      • Hammond R.
      • et al.
      [(18)F]-NaF PET/CT identifies active calcification in carotid plaque.
      ] or PET/MRI [
      • Evans N.R.
      • Tarkin J.M.
      • Le E.P.
      • et al.
      Integrated cardiovascular assessment of atherosclerosis using PET/MRI.
      ,
      • Aizaz M.
      • Moonen R.P.M.
      • van der Pol J.A.J.
      • Prieto C.
      • Botnar R.M.
      • Kooi M.E.
      PET/MRI of atherosclerosis.
      ] also provides additional information on plaque morphology. The glucose analogue 18F-Fluorodeoxyglucose (FDG) is taken up by cells with a high metabolic rate, such as macrophages within an atherosclerotic plaque, and therefore enables to quantify the inflammatory activity within carotid atherosclerotic plaques [
      • Rudd J.H.F.
      • Warburton E.A.
      • Fryer T.D.
      • et al.
      Imaging atherosclerotic plaque inflammation with [18F]-fluorodeoxyglucose positron emission tomography.
      ]. In order to correct for uptake of the tracer in the blood pool, it is recommended to use the target to background ratio (TBR) to quantify FDG uptake [
      • Bucerius J.
      • Hyafil F.
      • Verberne H.J.
      • et al.
      Position paper of the cardiovascular committee of the European association of nuclear medicine (EANM) on PET imaging of atherosclerosis.
      ]. TBRmax is defined as the ratio of the maximal standardised uptake value (SUVmax) measured in the plaque and the mean SUV (SUVmean) in the blood pool [
      • Bucerius J.
      • Hyafil F.
      • Verberne H.J.
      • et al.
      Position paper of the cardiovascular committee of the European association of nuclear medicine (EANM) on PET imaging of atherosclerosis.
      ]. In a study of 49 patients that underwent an 18F-FDG PET examination before CEA, it was shown that the TBRmax correlates with the extent of CD68 staining, a measure for macrophage content of the plaque (r = 0.51, P < 0001) (Fig. 3) [
      • Cocker M.S.
      • Spence J.D.
      • Hammond R.
      • et al.
      [18F]-Fluorodeoxyglucose PET/CT imaging as a marker of carotid plaque inflammation: comparison to immunohistology and relationship to acuity of events.
      ]. Various studies demonstrated higher uptake in symptomatic compared to asymptomatic plaques, while the activity in symptomatic plaques decreases in the months after the event [
      • Cocker M.S.
      • Spence J.D.
      • Hammond R.
      • et al.
      [18F]-Fluorodeoxyglucose PET/CT imaging as a marker of carotid plaque inflammation: comparison to immunohistology and relationship to acuity of events.
      ,
      • Kwee R.M.
      • Truijman M.T.B.
      • Mess W.H.
      • et al.
      Potential of integrated [18F] fluorodeoxyglucose positron-emission tomography/CT in identifying vulnerable carotid plaques.
      ,
      • Chaker S.
      • Al-Dasuqi K.
      • Baradaran H.
      • et al.
      Carotid plaque positron emission tomography imaging and cerebral ischemic disease.
      ,
      • Poredos P.
      • Spirkoska A.
      • Lezaic L.
      • Mijovski M.B.
      • Jezovnik M.K.
      Patients with an inflamed atherosclerotic plaque have increased levels of circulating inflammatory markers.
      ,
      • Jezovnik M.K.
      • Zidar N.
      • Lezaic L.
      • Gersak B.
      • Poredos P.
      Identification of inflamed atherosclerotic lesions in vivo using PET-CT.
      ]. Moreover, several studies reported weak correlations between 18F-FDG uptake on PET and CT/MRI parameters of carotid plaque (Spearman ρ: 0.098–0.39), indicating that PET may have additive information for risk assessment [
      • Kwee R.M.
      • Truijman M.T.B.
      • Mess W.H.
      • et al.
      Potential of integrated [18F] fluorodeoxyglucose positron-emission tomography/CT in identifying vulnerable carotid plaques.
      ,
      • Truijman M.T.B.
      • Kwee R.M.
      • van Hoof R.H.M.
      • et al.
      Combined 18F-FDG PET-CT and DCE-MRI to assess inflammation and microvascularization in atherosclerotic plaques.
      ,
      • Wang J.
      • Liu H.
      • Sun J.
      • et al.
      Varying correlation between 18F-fluorodeoxyglucose positron emission tomography and dynamic contrast-enhanced MRI in carotid atherosclerosis: implications for plaque inflammation.
      ]. Importantly, 18F-FDG uptake was demonstrated to predict early post-PET stroke recurrence with a fully adjusted hazard ratio of 4.57 (95% confidence interval [CI], 1.5–13.96; p = 0.008) in a pooled cohort of 196 patients with carotid stenosis and recent stroke/transient ischemic attack with 8 post-PET stroke recurrences. Although most extensively validated, a disadvantage of 18F-FDG is that the tracer is not specific. Recently, more specific tracers for plaque inflammation have been proposed [
      • Tarkin J.M.
      • Joshi F.R.
      • Evans N.R.
      • et al.
      Detection of atherosclerotic inflammation by (68)Ga-DOTATATE PET compared to [(18)F]FDG PET imaging.
      ,
      • Gaemperli O.
      • Shalhoub J.
      • Owen D.R.J.
      • et al.
      Imaging intraplaque inflammation in carotid atherosclerosis with 11C-PK11195 positron emission tomography/computed tomography.
      ,
      • Vöö S.
      • Kwee R.M.
      • Sluimer J.C.
      • et al.
      Imaging intraplaque inflammation in carotid atherosclerosis with 18F-fluorocholine positron emission tomography-computed tomography: prospective study on vulnerable atheroma with immunohistochemical validation.
      ], but these still need to be validated in larger studies. Alternatively, uptake of [
      • Bos D.
      • van Dam-Nolen D.H.K.
      • Gupta A.
      • et al.
      Advances in multimodality carotid plaque imaging: AJR expert panel narrative review.
      ]⁸F-sodium fluoride (NaF), a marker for active microcalcific processes, was reported at the site of carotid plaque rupture and larger uptake was demonstrated in symptomatic plaques [
      • Cocker M.S.
      • Spence J.D.
      • Hammond R.
      • et al.
      [18F]-Fluorodeoxyglucose PET/CT imaging as a marker of carotid plaque inflammation: comparison to immunohistology and relationship to acuity of events.
      ,
      • Vesey A.T.
      • Jenkins W.S.A.
      • Irkle A.
      • et al.
      18)F-Fluoride and (18)F-fluorodeoxyglucose positron emission tomography after transient ischemic attack or minor ischemic stroke: case-control study.
      ,
      • Joshi N.V.
      • Vesey A.T.
      • Williams M.C.
      • et al.
      18F-fluoride positron emission tomography for identification of ruptured and high-risk coronary atherosclerotic plaques: a prospective clinical trial.
      ]. The value of 18F–NaF for risk stratification is currently under investigation in an ongoing prospective multicenter trial (PREFFIR; unique identifier: NCT02278211).
      Fig. 3
      Fig. 3Metabolic activity within atherosclerotic carotid plaque, imaged with [18F]-fluorodeoxyglucose (18FDG) hybrid PET/CT correlates with ex-vivo macrophage-specific CD68 CT angiography (axial plane) fused with [18F]-fluorodeoxyglucose PET in a patient with a symptomatic right internal carotid artery plaque (arrow).
      From the fused PET/CT images, there are small regions of calcification with a narrowing of the right internal carotid artery. The tissue to blood ratio for maximum 18FDG uptake was 4.7. Following excision and advanced immunohistology, there is strong evidence for extensive CD68 staining (rust-stained regions), a marker of macrophage expression and direct inflammatory burden. Reproduced with permission of Elsevier from: Cocker MS, Spence JD, Hammond R et al. [18F]-Fluorodeoxyglucose PET/CT imaging as a marker of carotid plaque inflammation: Comparison to immunohistology and relationship to acuity of events. International Journal of Cardiology. 2018; 271:378–386.

      5.2 Intravascular imaging platforms

      The use of intravascular technologies for intraluminal imaging of carotid atherosclerosis is currently limited to highly selected cases and includes fiber-bundle angioscopy (FBA), IVUS, and optical coherence tomography (OCT). FBA was introduced in the early 1980s and initially applied to assess plaque disruption, luminal thrombus, and stent placement [,
      • Uchida Y.
      Recent advances in coronary angioscopy.
      ]. Despite the initial enthusiasm given the unprecedented images of the arterial lumen and surface, the poor image quality (<10,000 pixels even with FBA), the large size, and the excessive stiffness of the cameras significantly limited adoption [
      • Savastano L.E.
      • Seibel E.J.
      Scanning fiber angioscopy: a multimodal intravascular imaging platform for carotid atherosclerosis.
      ]. Recent advances in photonics and optics allowed the development of Scanning Fiber Angioscopy, a miniature laser-based platform capable of generating high resolution (∼12 μm, or >250,000 pixels) structural, biochemical, and biological vascular videos in real time (Fig. 4A) [
      • Savastano L.E.
      • Seibel E.J.
      Scanning fiber angioscopy: a multimodal intravascular imaging platform for carotid atherosclerosis.
      ,
      • Savastano L.E.
      • Zhou Q.
      • Smith A.
      • et al.
      Multimodal laser-based angioscopy for structural, chemical and biological imaging of atherosclerosis.
      ]. IVUS was introduced in the late 1980s and employs an intravascular piezoelectric transducer that creates waves that propagate through blood and tissue. IVUS generates cross-sectional imaging without the need of clearing the intravascular blood, but the resolution is poor (100–150 μm). IVUS has been clinically used to characterize the structure of carotid plaques by virtual histology, measure the degree of stenosis, and evaluate stent apposition and plaque protrusion in CAS (Fig. 4B) [
      • Kan P.
      • Mokin M.
      • Abla A.A.
      • et al.
      Utility of intravascular ultrasound in intracranial and extracranial neurointerventions: experience at university at buffalo neurosurgery-millard fillmore gates circle hospital.
      ,
      • Sangiorgi G.
      • Bedogni F.
      • Sganzerla P.
      • et al.
      The Virtual histology in CaroTids Observational RegistrY (VICTORY) study: a European prospective registry to assess the feasibility and safety of intravascular ultrasound and virtual histology during carotid interventions.
      ,
      • Diethrich E.B.
      • Pauliina Margolis M.
      • Reid D.B.
      • et al.
      Virtual histology intravascular ultrasound assessment of carotid artery disease: the Carotid Artery Plaque Virtual Histology Evaluation (CAPITAL) study.
      ]. OCT shines a near-infrared laser sideways and a small portion of this light (scattering) that reflects from sub-surface tissues is collected and processed by interferometry. Automated pullback in a bloodless lumen results in cross-sectional images of arteries. The use of OCT in carotid arteries continues to be very limited in the evaluation of disrupted plaques, stent apposition, and tissue prolapse in CAS [
      • Funatsu N.
      • Enomoto Y.
      • Egashira Y.
      • et al.
      Tissue protrusion with attenuation is associated with ischemic brain lesions after carotid artery stenting.
      ,
      • de Donato G.
      • Pasqui E.
      • Alba G.
      • et al.
      Clinical considerations and recommendations for OCT-guided carotid artery stenting.
      ].
      Fig. 4
      Fig. 4(A) Laser-angioscopy showing an acutely disrupted plaque with red blood cell rich intraluminal thrombus resulting in critical stenosis; (B) IVUS images with doppler mode of a symptomatic calcified plaque causing severe irregular stenosis; (C) OCT images showing stent apposition in a carotid artery. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

      6. A look into the future

      6.1 Artificial intelligence

      As stroke is the second leading cause of global mortality, this demonstrated the need for improved tools in the management of occlusive vascular disease [
      • Roth G.A.
      • Johnson C.
      • Abajobir A.
      • et al.
      Global, regional, and national burden of cardiovascular diseases for 10 causes, 1990 to 2015.
      ,
      • Feigin V.L.
      • Norrving B.
      • Mensah G.A.
      Global burden of stroke.
      ]. Patients with cardiovascular disease leading to stroke often require significant medical imaging in the acute, sub-acute, and chronic settings, using a range of imaging modalities. Vascular imaging is then used as a key source of information in the determination of appropriate clinical management. In the era of modern medicine, AI is an evolving field that is experiencing a steady development in vascular imaging [
      • Cau R.
      • Flanders A.
      • Mannelli L.
      • et al.
      Artificial intelligence in computed tomography plaque characterization: a review.
      ,
      • Cau R.
      • Cherchi V.
      • Micheletti G.
      • et al.
      Potential role of artificial intelligence in cardiac magnetic resonance imaging.
      ,
      • Cau R.
      • Cherchi V.
      • Micheletti G.
      • et al.
      Potential role of artificial intelligence in cardiac magnetic resonance imaging.
      ,
      • Cau R.
      • Faa G.
      • Nardi V.
      • et al.
      Long-COVID diagnosis: from diagnostic to advanced AI-driven models.
      ]. In daily clinical practice, plaque assessment is performed through manual measurement of the degree of the stenosis and visual evaluation of plaque composition [
      • Aboyans V.
      • Ricco J.-B.
      • Bartelink M.-L.E.L.
      • et al.
      ESC guidelines on the diagnosis and treatment of peripheral arterial diseases, in collaboration with the European society for vascular Surgery (ESVS): document covering atherosclerotic disease of extracranial carotid and vertebral, mesenteric, renal.
      ,
      • Shishikura D.
      Noninvasive imaging modalities to visualize atherosclerotic plaques.
      ]. However, a manual evaluation has various limitations, including a long analysis time and is highly dependent on the operator. AI may evaluate carotid plaques with their vulnerable features to decide whether invasive investigation and treatment are necessary.
      US is the modality of choice for initial evaluation and confirmation of carotid artery disease: characteristics of the carotid plaque in patients with carotid stenosis can identify those patients with relatively higher risk for stroke and help select patients who may benefit from intervention over medical treatment alone or vice versa. Symptomatic plaques tend to produce more tight stenosis, be more hypoechoic, have a large juxtaluminal black area close to the lumen without a visible echogenic cap, and discrete white hyperechoic areas compared with asymptomatic plaques [
      • Paraskevas K.I.
      • Nicolaides A.N.
      • Kakkos S.K.
      Asymptomatic carotid stenosis and risk of stroke (ACSRS) study: what have we learned from it?.
      ,
      • Kakkos S.K.
      • Griffin M.B.
      • Nicolaides A.N.
      • et al.
      The size of juxtaluminal hypoechoic area in ultrasound images of asymptomatic carotid plaques predicts the occurrence of stroke.
      ]. Additional and more precise information can be derived from software implementations applied to ultrasound: 3D/4D, Intravascular Ultrasound (IVUS), CEUS, Sonoelastography, Vector Doppler, Grayscale Median (GSM), Radiofrequency, etc. The large datasets obtained from all these imaging sources are traditionally interpreted qualitatively by clinicians but are highly heterogeneous, varying due to differences in patient, imaging technology, and site scanning protocols. Current measurement methods are time-consuming and do not utilize the power of knowledge-based paradigms such as artificial intelligence (AI), a branch of computer science that includes machine learning (ML), deep learning (DL), and convolutional neural networks (CNNs) (Fig. 5). AI excels at automatically recognizing complex patterns and providing quantitative assessment for imaging data, showing high potential to assist physicians in acquiring more accurate and reproducible results [
      • Shen Y.-T.
      • Chen L.
      • Yue W.-W.
      • Xu H.-X.
      Artificial intelligence in ultrasound.
      ,
      • Boyd C.
      • Brown G.
      • Kleinig T.
      • et al.
      Machine learning quantitation of cardiovascular and cerebrovascular disease: a systematic review of clinical applications.
      ]. Recently, a DL–based model was applied for carotid IMT and lumen measurement [
      • Biswas M.
      • Saba L.
      • Chakrabartty S.
      • et al.
      Two-stage artificial intelligence model for jointly measurement of atherosclerotic wall thickness and plaque burden in carotid ultrasound: a screening tool for cardiovascular/stroke risk assessment.
      ]. It was the first artificial intelligence-based approach to ultrasound-based carotid artery segmentation and carotid IMT (cWT þ CP) measurement that used 13 layers of convolution layers for feature extraction and three up-sample layers for segmentation. Deep learning resulted in a useful tool for carotid ultrasound-based characterization and classification of symptomatic and asymptomatic plaques in a more recent paper [
      • Saba L.
      • Sanagala S.S.
      • Gupta S.K.
      • et al.
      Ultrasound-based internal carotid artery plaque characterization using deep learning paradigm on a supercomputer: a cardiovascular disease/stroke risk assessment system.
      ] where implementation with a supercomputer configuration was more precise and faster if compared with other AI systems. Further and more accurate measurements can be obtained when an AI-based model utilizing DL methodology is used on image patches rather than full-size images, mainly to have better local control in small regions rather than the whole image at once [
      • Shen Y.-T.
      • Chen L.
      • Yue W.-W.
      • Xu H.-X.
      Artificial intelligence in ultrasound.
      ,
      • Biswas M.
      • Saba L.
      • Chakrabartty S.
      • et al.
      Two-stage artificial intelligence model for jointly measurement of atherosclerotic wall thickness and plaque burden in carotid ultrasound: a screening tool for cardiovascular/stroke risk assessment.
      ,
      • Biswas M.
      • Kuppili V.
      • Araki T.
      • et al.
      Deep learning strategy for accurate carotid intima-media thickness measurement: an ultrasound study on Japanese diabetic cohort.
      ]. A new method consisting of a novel design of 10 types of solo deep learning (SDL) and hybrid deep learning (HDL) models focused on automated plaque segmentation in the internal carotid artery (ICA) has shown to be very useful in identifying plaques at risk of rupture: the system is very fast and precise (it takes <1 s per image) and it may therefore be practical to introduce such an AI-based system to detect rupture-prone plaques (or vulnrable plaque detection) [
      • Jain P.K.
      • Sharma N.
      • Giannopoulos A.A.
      • Saba L.
      • Nicolaides A.
      • Suri J.S.
      Hybrid deep learning segmentation models for atherosclerotic plaque in internal carotid artery B-mode ultrasound.
      ].
      Fig. 5
      Fig. 5Venn diagram illustrating the hierarchy of the artificial intelligence fields.
      In addition to the improved ability to define so-called vulnerable plaques to enable the best therapeutic approach, AI has also proved to be useful for the evaluation of carotid artery stenting (CAS) prognosis and in the prediction of persistent hemodynamic depression after carotid angioplasty [
      • Cheng C.-A.
      • Chiu H.-W.
      An artificial neural network model for the evaluation of carotid artery stenting prognosis using a national-wide database.
      ,
      • Jeon J.P.
      • Kim C.
      • Oh B.-D.
      • Kim S.J.
      • Kim Y.-S.
      Prediction of persistent hemodynamic depression after carotid angioplasty and stenting using artificial neural network model.
      ]. Of note, the application of AI to ultrasonographic diagnostics for better diagnosis and possibly new classification and standardization methods [
      • Saba L.
      • Jamthikar A.
      • Gupta D.
      • et al.
      Global perspective on carotid intima-media thickness and plaque: should the current measurement guidelines be revisited?.
      ] requires close collaboration among computer scientists, clinical investigators, clinicians, and other users in order to identify the most relevant problems to be solved and the best approach and data sources to achieve this.
      An AI-based approach has also proven its usefulness in CT. Acharya et al. investigated a supervised-learning model to classify carotid artery images into symptomatic and asymptomatic using a combination of local binary model and wavelet energy features [
      • Acharya U.R.
      • Sree S.V.
      • Mookiah M.R.K.
      • et al.
      Computed tomography carotid wall plaque characterization using a combination of discrete wavelet transform and texture features: a pilot study.
      ]. The authors reported sensitivities, specificities, and accuracies of 0.88, 0.865, and 0.902, respectively [
      • Acharya U.R.
      • Sree S.V.
      • Mookiah M.R.K.
      • et al.
      Computed tomography carotid wall plaque characterization using a combination of discrete wavelet transform and texture features: a pilot study.
      ]. Dos Santos et al. proposed a fully-automated, user-independent tool for the segmentation and analysis of atherosclerosis in the extracranial carotid arteries, reporting performance of 83% with accuracy, sensitivity, and specificity values of 71%, 83%, and 25%, respectively with an average difference between manual and automated analysis of 37% (p = 027) and an average analysis time of 1381 s per patient [
      • Caetano dos Santos F.L.
      • Kolasa M.
      • Terada M.
      • Salenius J.
      • Eskola H.
      • Paci M.
      VASIM: an automated tool for the quantification of carotid atherosclerosis by computed tomography angiography.
      ]. AI models have been also developed to simplify plaque characterization and predict histological plaque composition. Hanning et al. tested an ML-based analysis of admission non-contrast CT and CTA to predict thrombus composition with its fractions of fibrin and red blood cells [
      • Hanning U.
      • Sporns P.B.
      • Psychogios M.N.
      • et al.
      Imaging-based prediction of histological clot composition from admission CT imaging.
      ]. This analysis included 112 patients who underwent thrombectomy due to carotid or middle cerebral artery occlusion, evaluating both vessel walls, thrombi, and peri-vascular tissue response. The ML-based algorithm demonstrated an AUC of 0.83 for differentiating thrombi with a high fraction of red blood cells (sensitivity and specificity of 77% and 74%, respectively) and an AUC of 0.84 for differentiating fibrin-rich thrombi (sensitivity and specificity of 81% and 73%, respectively) [
      • Hanning U.
      • Sporns P.B.
      • Psychogios M.N.
      • et al.
      Imaging-based prediction of histological clot composition from admission CT imaging.
      ]. Another research investigated the ability of a DL-based model to identify symptomatic patients from asymptomatic patients and further discriminate between culprit and non-culprit carotid arteries in symptomatic patients [
      • Le E.P.V.
      • Evans N.R.
      • Tarkin J.M.
      • et al.
      Contrast CT classification of asymptomatic and symptomatic carotids in stroke and transient ischaemic attack with deep learning and interpretability.
      ]. This proposed model was 92% accurate in differentiating between symptomatic and asymptomatic patients, and 71% accurate in discriminating between culprit versus non-culprit carotid arteries in symptomatic patients [
      • Le E.P.V.
      • Evans N.R.
      • Tarkin J.M.
      • et al.
      Contrast CT classification of asymptomatic and symptomatic carotids in stroke and transient ischaemic attack with deep learning and interpretability.
      ]. The relationship between carotid vessel image parameters and stroke risk was also investigated by Lal et al. using an artificial intelligence algorithm for risk stratification in carotid atherosclerosis incorporating a combination of carotid plaque geometry, plaque composition, patient demographics, and clinical information [
      • Lal B.K.
      • Kashyap V.S.
      • Patel J.B.
      • et al.
      Novel application of artificial intelligence algorithms to develop a predictive model for major adverse neurologic events in patients with carotid atherosclerosis.
      ] AI is able to mesh a large amount of quantitative imaging data to clinical parameters, that may be a new frontier of AI in carotid plaque risk assessment improving diagnosis and decision-making in daily clinical practice.
      AI is transforming most healthcare domains including carotid MRI. AI is increasingly used to reduce manual effort in carotid MRI measurements. Using a convolutional neural network (CNN) based algorithm called DeepMAD to separately segment the carotid lumen and outerwall contours on 2D T1w turbo spin-echo MRI, Wu et al. identified slices with atherosclerotic plaque [
      • Wu J.
      • Xin J.
      • Yang X.
      • et al.
      Deep morphology aided diagnosis network for segmentation of carotid artery vessel wall and diagnosis of carotid atherosclerosis on black-blood vessel wall MRI.
      ]. Similarly, Samber et al. used two separate CNNs for lumen and outerwall segmentation of 2D T2w turbo spin echo MRI [
      • Samber D.D.
      • Ramachandran S.
      • Sahota A.
      • et al.
      Segmentation of carotid arterial walls using neural networks.
      ]. Chen et al. demonstrated a CNN algorithm called LATTE for segmentation [
      • Chen L.
      • Sun J.
      • Canton G.
      • et al.
      Automated artery localization and vessel wall segmentation using tracklet refinement and polar conversion.
      ] of the carotid vessel wall on 3D-MERGE [
      • Balu N.
      • Yarnykh V.L.
      • Chu B.
      • Wang J.
      • Hatsukami T.
      • Yuan C.
      Carotid plaque assessment using fast 3D isotropic resolution black-blood MRI.
      ] black-blood MRI using a polar transformation centered on the carotid after vessel identification. To make the segmentation robust to inter-scanner differences, domain adaptation for LATTE was developed and shown to improve the identification of advanced plaque [
      • Chen L.
      • Zhao H.
      • Jiang H.
      • et al.
      Domain adaptive and fully automated carotid artery atherosclerotic lesion detection using an artificial intelligence approach (LATTE) on 3D MRI.
      ]. Thus, quick, and automated screening for carotid plaque using 3D-MERGE is made possible by LATTE. The next frontier for carotid AI applications lies in plaque component segmentation. CNN-based segmentation of plaque components such as lipid-core, calcification, and intra-plaque hemorrhage on multi-contrast 2D MRI is able to better match the human expert's plaque component segmentation than non-CNN methods [
      • Dong Y.
      • Pan Y.
      • Zhao X.
      • Li R.
      • Yuan C.
      • Xu W.
      Identifying carotid plaque composition in MRI with convolutional neural networks.
      ]. Zhang et al. compared several ML methods [
      • Zhang R.
      • Zhang Q.
      • Ji A.
      • et al.
      Identification of high-risk carotid plaque with MRI-based radiomics and machine learning.
      ] for plaque component segmentation using a specific sequence called SNAP [
      • Wang J.
      • Börnert P.
      • Zhao H.
      • et al.
      Simultaneous noncontrast angiography and intraplaque hemorrhage (SNAP) imaging for carotid atherosclerotic disease evaluation.
      ]. However, these methods are 2D MRI based and need to be modified for use with 3D carotid MRI. With future development, multi-contrast 3D plaque component segmentation may allow complete carotid plaque analysis and quantification with minimal user intervention thereby reducing clinician workloads and expanding the applications of carotid plaque MRI.
      CNNs have also found applications in carotid MRA. Koktzoglou et al. demonstrated that non-contrast carotid MRI can be accelerated to below 3 min when combined with denoising of MRA using CNNs [
      • Koktzoglou I.
      • Huang R.
      • Ong A.L.
      • Aouad P.J.
      • Aherne E.A.
      • Edelman R.R.
      Feasibility of a sub-3-minute imaging strategy for ungated quiescent interval slice-selective MRA of the extracranial carotid arteries using radial k-space sampling and deep learning-based image processing.
      ]. Ziegler et al. used the DeepMedic CNN on CE-MRA to segment the carotid artery into common, internal, and external carotid segments [
      • Ziegler M.
      • Alfraeus J.
      • Bustamante M.
      • et al.
      Automated segmentation of the individual branches of the carotid arteries in contrast-enhanced MR angiography using DeepMedic.
      ].
      Information contained in the grey-scale differences among tissues is easily summarized by human-derived features. Radiomics can extract traditional grey scale level features from images to improve the diagnostic capabilities of carotid MRI. Zhang et al. showed that adding radiomic features of carotid plaque to traditional plaque features improved the model's ability to predict symptom status [
      • Zhang R.
      • Zhang Q.
      • Ji A.
      • et al.
      Identification of high-risk carotid plaque with MRI-based radiomics and machine learning.
      ]. Application of such radiomics specific to the carotids requires segmentation of the carotid lumen and outer wall. Hence future combination of CNN based segmentation methods combined with radiomics may enable a comprehensive and automated analysis of both carotid MRI and other clinical variables to predict patient outcomes.

      6.2 Radiomics

      Since the 1990s, the improvement of resolution, which allows the identification of increasingly smaller lesions, and the availability of imaging modalities that provide morphological and functional information have introduced new scenarios and new diagnostic possibilities. The introduction of new imaging technologies such as ultrasound contrast agents, microvascular flow, elastography, and specific imaging processing techniques allows us to obtain improved morphological/functional quantitative information compared to those only derived from B-mode.
      Precision medicine requires a clear understanding of each patient's heterogeneity and individual situation. Radiological images are often analysed and interpreted by the radiologist only qualitatively (visual evaluation). However, digital images are composed of individual pixels to which discrete brightness or colour values are assigned. They can be efficiently processed, objectively evaluated, and made available at many places at the same time by means of appropriate communication networks and protocols, such as PACS and DICOM protocols. In a digital image, a large amount of numerical data is not analysed by the radiologist. This “hidden” information can be used to create a “radiological plot”, which can provide much more information on tissue than simple visual observations by providing objective data. The amount of data associated with digital imaging has increased and produced a large amount of electronic data (“Big Data”). In personalized and precision medicine, the data, analysed with complex mathematical algorithms and the use of artificial intelligence methods (Fig. 6) [
      • Tang A.
      • Tam R.
      • Cadrin-Chênevert A.
      • et al.
      Canadian association of radiologists white paper on artificial intelligence in radiology.
      ], can provide quantitative information on pathophysiological phenomena to improve diagnostic accuracy, prognostic, and predictive imaging capacity.
      Fig. 6
      Fig. 6Basic representation of an artificial neural network with neurons similar to those within a brain.
      The left layer of the neural network is called the input layer and contains neurons that encode the values of the input pixels. The right most layer is called the output layer, which contains the output neurons. The middle contains the “n” number of hidden layers, which perform mathematical transformations of the data.
      Artificial intelligence techniques consist of ML systems. The computer receives data and analyses the existing relationships using analysis systems that reproduce the functioning of the nervous system.
      The term “radiomics” was defined by Lambin in 2012 [
      • Lambin P.
      • Rios-Velazquez E.
      • Leijenaar R.
      • et al.
      Radiomics: extracting more information from medical images using advanced feature analysis.
      ] as the high-throughput extraction of image features from diagnostic images. The final product is a quantitative feature, measurable and minable, defined as an “imaging biomarker”. Biomarkers are indicators of normal biological processes, pathological changes, or pharmaceutical responses to a therapeutic intervention [
      • ES of R.
      ESR
      White paper on imaging biomarkers.
      ,
      • Neri E.
      • Del Re M.
      • Paiar F.
      • et al.
      Radiomics and liquid biopsy in oncology: the holons of systems medicine.
      ]. Therefore, radiomics represents diagnostic and predictive support that, together with other clinical and genetic investigations, allows the formulation of personalized therapies and the evaluation of treatment response.
      The radiomic data are extracted and processed with bioinformatics tools. They can be combined with other patient data (clinical, biohumoral, genetic) to develop models to improve diagnostic, prognostic, and predictive accuracies. Although radiomics is a natural extension of computer-aided diagnosis and detection (CAD) systems, it is significantly different from them. CAD systems are usually used for the detection or diagnosis of disease [
      • Doi K.
      Computer-aided diagnosis in medical imaging: historical review, current status and future potential.
      ,
      ] and are directed towards delivering a single answer (presence or absence of disease). Radiomics is a process designed to extract a large number of quantitative features from digital images to generate pathophysiological hypotheses and provide information on the phenotype and microenvironment. These features, in conjunction with other information, can be correlated with clinical outcomes and used for clinical support. Radiomics has the potential to help with the diagnosis and visualization of lesion heterogeneity and may prove critical in the assessment of prognosis, prediction of response to treatment, and monitoring of disease status. The “omics” concept readily applies to quantitative tomographic imaging on multiple levels (one multi-layer or three-dimensional image from one patient may easily contain millions of voxels). Complex images with high-dimensional data are generated, corresponding to measurable biological characteristics.
      Radiomics depicts the goal of precision medicine, in which stable, reproducible, and validated molecular biomarkers are used to predict “the right treatment for the right patient at the right time” [
      ,
      • Kumar V.
      • Gu Y.
      • Basu S.
      • et al.
      Radiomics: the process and the challenges.
      ].
      The radiomic process can be divided into five phases [
      • Aerts H.J.W.L.
      The potential of radiomic-based phenotyping in precision medicine: a review.
      ]: 1) Image acquisition and reconstruction, 2) Segmentation and rendering, 3) Feature extraction and qualification, 4) Construction of a database, and 5) Modelling and validation.
      The first step in the radiomics algorithm begins with the choice of an image acquisition protocol. This varies according to the clinical end-point. However, image acquisition parameters, including radiation dose, scanning protocol, reconstruction algorithm, and slice thickness, vary widely in routine clinical practice. Therefore, a comparison of the features extracted from different methods of image acquisition is not possible. The radiomic features are generally sensitive to the acquisition protocols used, only some are stable despite the different image reconstruction settings. Significant efforts are required to identify univocal acquisition and reconstruction protocols and to match them between different scanners.
      In most patients with carotid stenotic lesions, the volumes of interest can be identified. Furthermore, the subvolumes within atherosclerotic plaque, representative of plaque heterogeneity, can be analysed separately. With this approach, images with different acquisition parameters can be combined to yield regions with specific combinations of plaque features (cell density, necrotic core, hemorrhage, atherosclerotic fibrous cap, flow velocity, etc). Once the volumes of interest have been identified, the segmentation strategy must be chosen. This point is critical as the resulting feature values depend on the adopted segmentation methods, which should be stable and reproducible. Usually, manual segmentation by expert readers is considered the gold standard, but it is a time-consuming process with high inter-operator variability. Consequently, the best compromise has been identified in CAD systems that work semi-automatically, with subsequent human manual correction. The use of semi-automated methods has also paved the way for three-dimensional (3D) segmentation [
      • Kumar V.
      • Gu Y.
      • Basu S.
      • et al.
      Radiomics: the process and the challenges.
      ]. Volumetric segmentation allows a comprehensive view of the total lesion and burden, a more complete description of the shape, and a greater number of points included in the computation of statistical features, leading to more reliable results that do not suffer from sampling errors. Moreover, computer-aided approaches reduce the manual workload, allow fast, and reproducible volumetric segmentation in large cohorts of patients [
      • Kumar V.
      • Gu Y.
      • Basu S.
      • et al.
      Radiomics: the process and the challenges.
      ]. From the identified atherosclerotic plaque, multiple quantitative image features can be extracted, including features that describe the characteristics of the region under analysis, such as the histogram of signal intensity, shape, and texture, and descriptors of the position and its relationships with surrounding tissues. Features can be “semantic” or “agnostic”. Semantic features are those commonly used by radiologists to describe regions of interest with qualitative descriptors such as size and shape. The agnostic features are mathematically extracted indicators that are generally not part of the traditional lexicon of radiologists and can be divided into first- and second-order statistical features [
      • Aerts H.J.W.L.
      The potential of radiomic-based phenotyping in precision medicine: a review.
      ]. The first-order features describe the intensity histogram by extracting features such as the maximum and average values in addition to the causality and asymmetry of the distribution and have the limitation of not providing information on the spatial relations between voxels. This information can be obtained from the statistical features of the second order, which, using texture analysis methods, describe the relations between the signal in a voxel and the signal in the neighbouring voxels. Overall, each category produces various quantitative parameters that reflect the specific aspects of a lesion. The power of a predictive classifier model is dependent on having sufficient data. A reasonable rule of thumb is that 10 samples (patients) are needed for each feature in a model based on binary classifiers. Furthermore, the best models are those that can accommodate additional clinical or genomic covariates. Radiomics can be performed with as few as 100 patients, although larger data sets provide more power. Radiomic and non-radiomic features should be combined with the prediction target to create a single dataset. After identification, features can be included or excluded from the model. Radiomic features that are well-correlated with routine clinical feature (such as symptoms) or features not correlated with the clinical end-point are excluded. A predictive model of clinical outcomes is constructed using the features extracted from the data set. Radiomics produces models for assessing the risk of stroke or for estimating the probability of patient survival. Independent validation datasets are needed to confirm the prognostic value of the same radiomic features. Model performance is measured using receiver operating characteristic (ROC) curve analysis, which measures accuracy throughout the range of possible model values and identifies the best cut-off value. The validation of a model must be accompanied by the verification of its reproducibility by repeating the analyses with the same procedures on different data sets. It is therefore important to have a comprehensive and detailed medical image database [
      • Gillies R.J.
      • Kinahan P.E.
      • Hricak H.
      Radiomics: images are more than pictures, they are data.
      ]. Multiple articles have focused on ML approaches for the role of image processing in thw prediction of cardiovascular event and demonstrated that can improve the accuracy of cardiovascular disease prediction and a better predictive capacity than some traditional risk scores [
      • Yip S.S.F.
      • Aerts H.J.W.L.
      Applications and limitations of radiomics.
      ,
      • Ambale-Venkatesh B.
      • Yang X.
      • Wu C.O.
      • et al.
      Cardiovascular event prediction by machine learning: the multi-ethnic study of atherosclerosis.
      ,
      • Han D.
      • Kolli K.K.
      • Al'Aref S.J.
      • et al.
      Machine learning framework to identify individuals at risk of rapid progression of coronary atherosclerosis: from the PARADIGM registry.
      ,
      • Hu X.
      • Reaven P.D.
      • Saremi A.
      • et al.
      Machine learning to predict rapid progression of carotid atherosclerosis in patients with impaired glucose tolerance.
      ,
      • Motwani M.
      • Dey D.
      • Berman D.S.
      • et al.
      Machine learning for prediction of all-cause mortality in patients with suspected coronary artery disease: a 5-year multicentre prospective registry analysis.
      ,
      • Quesada J.A.
      • Lopez-Pineda A.
      • Gil-Guillén V.F.
      • et al.
      Machine learning to predict cardiovascular risk.
      ,
      • Groenendyk J.W.
      • Mehta N.N.
      Applying the ordinal model of atherosclerosis to imaging science: a brief review.
      ,
      • Terrada O.
      • Cherradi B.
      • Raihani A.
      • Bouattane O.
      A novel medical diagnosis support system for predicting patients with atherosclerosis diseases.
      ,
      • van Rosendael A.R.
      • Maliakal G.
      • Kolli K.K.
      • et al.
      Maximization of the usage of coronary CTA derived plaque information using a machine learning based algorithm to improve risk stratification; insights from the CONFIRM registry.