Despite improvements in cardiac diagnostics and preventive treatments, cardiovascular disease is still the leading cause of death globally, causing an estimated 17.9 million deaths on a yearly basis.1 By 2030, 43.9% of the US adult population is projected to develop some form of cardiovascular disease.2 Clinical guidelines on cardiovascular disease prevention support the assessment of global cardiovascular risk using multiparametric clinical scores.3 However, cardiovascular events most often occur in asymptomatic patients who are not receiving preventive therapies, having been misclassified by conventional risk factors as low-risk patients.4,5 Heart disease and stroke fall within the top 3 leading causes of death and are common causes of long-term disability; thus, accurate identification and preventive management of at-risk subjects are crucial to decrease disease burden.6
Arterial plaque rupture with thrombosis is the principal cause of major adverse cardiovascular events (MACE) and major adverse neurovascular events (MANE). Invasive or minimally invasive treatment strategies, such as coronary angioplasty or carotid endarterectomy, are commonly used to reduce the risk of MACE and MANE, respectively, in patients with symptomatic high-grade stenosis. The treatment of asymptomatic stenosis however is more controversial given the lack of high-quality evidence on which to base treatment.7,8 Thus, the identification of patients who are at higher risk for MACE and MANE based on plaque morphology features may have relevance in prevention, as indicated by a number of studies evaluating plaque-associated events.9–11
Computed tomography (CT) angiography (CTA) has the ability to visualize vascular plaques.12 However, CTA-based manual characterization of plaque morphology is a difficult, time-consuming, and potentially imprecise process. Artificial intelligence (AI)-based tools have emerged with the goal of increasing efficiency, objectivity, repeatability, specificity, and accuracy of vascular plaque assessment and quantification.13,14 Such AI-based plaque quantitation tools typically apply a Hounsfield Unit (HU) threshold for semiautomated segmentation and plaque tissue characterization.15–18 HU thresholds are used to classify tissue, such as low attenuation, non-calcified, and calcified plaque components, according to a predefined scale. Beyond HU-based thresholds, histologically validated plaque characterization, and quantification, an emerging AI-based approach has been introduced.19 This tool characterizes specific tissue types based on a biological correlate (such as lipid-rich necrotic core [LRNC], intraplaque hemorrhage [IPH], matrix, and calcification) with accuracy validated against histology.
VALUE OF CTA
Currently, artery lumen stenosis remains the main diagnostic and prognostic measurement for coronary artery disease (CAD) used in daily practice to guide patient management.20 CTA is an increasingly utilized technique capable of not only assessing stenosis but also characterizing and quantifying the extent of carotid and coronary atherosclerosis. The position of CTA in the assessment of atherosclerotic vascular disease has been further strengthened by the most recent NICE (National Institute for Health and Care Excellence), ACC/AHA (American College of Cardiology/American Heart Association) and ESC (European Society of Cardiology) clinical guidelines.21–23 The NICE guidelines recommend coronary CTA as the first-line test if the calcium score is between 1 and 400 for the evaluation of stable CAD in chest pain pathways. While the ACC/AHA and ESC guidelines recommend the use of coronary CTA as a class I indication in intermediate-risk patients, the ACC/AHA guidelines recommend CT-based fractional flow reserve as a class IIa recommendation for additional testing in certain cases. However, the potential utility of CTA exceeds the assessment of vascular stenosis.
CTA measurement of plaque burden is superior to clinical variables for the prediction of long-term MACE and MANE.24–26 Results from the CAPIRE (“Coronary Atherosclerosis in outlier subjects: Protective and novel Individual Risk factors Evaluation”) study confirmed the prognostic role of atherosclerosis evaluation by coronary CTA beyond coronary stenosis evaluation.27 Plaque volume, particularly non-calcified plaque volume, provided superior predictive value for cardiovascular events over lumen stenosis and clinical risk profile. A demonstrated correlation between low attenuation plaque (LAP) burden and inducible myocardial ischemia further supports the prognostic role of advanced plaque evaluation by coronary CTA.28 CTA allows evaluation of the arterial wall (not just the lumen), and it has been shown that adverse plaque characteristics, such as LAP or outward expansion of the artery wall (positive remodeling), are associated with future acute coronary syndromes. In an analysis of 1769 patients, the Scottish Computed Tomography of the Heart trial demonstrated that adverse plaque features on coronary CTA, including LAP volume positive remodeling, are the strongest predictors of MACE independent of coronary artery calcium score and stenosis29; however, in this study, LRNC and IPH were not quantified. In this study, using coronary CTA for the routine clinical assessment of patients with suspected angina secondary to CAD resulted in a 3-fold reduction in the rates of normal, that is, unnecessary, invasive coronary angiography.30 Moreover, CTA has prognostic utility not only in patients with high-grade stenoses but also in those with subclinical lesions. The Progression of AtheRosclerotic PlAque Determined by Computed TomoGraphic Angiography Imaging multinational registry examined longitudinal changes of coronary atherosclerotic plaques in 2252 patients.31 In this study, plaque volume quantification by coronary CTA most accurately predicted long-term atherosclerotic cardiovascular risk, even in the absence of significant coronary stenosis.
Moreover, coronary CTA has demonstrated higher diagnostic and prognostic accuracy compared with coronary artery calcium (CAC) scores in symptomatic patients without known CAD. In the Plaque Registration and Evaluation Detected In CT registry, adverse coronary plaque features, including low attenuation, positive remodeling, spotty calcification, and the napkin ring sign (low attenuation core with a higher attenuation rim), had an independent predictive value for future MACE beyond risk factors, obstructive stenosis, and the CAC score.10 The absence of coronary calcium did not guarantee the absence of atherosclerosis or adverse events. Calcium quantification alone cannot identify or characterize non-calcified plaques.
BARRIERS TO USING CTA IN ATHEROSCLEROTIC PLAQUE ASSESSMENT
While CTA can detect plaques with adverse features that place patients at risk for future MACE, there are multiple steps to take before CTA-based plaque assessment could be considered as a screening tool. First, actionable information derived from the analysis of CTAs performed for various clinical indications needs to be clearly defined. For coronary CTA to further strengthen its role in diagnostic cardiology, it would be beneficial to automate key parts of the image interpretation process to increase efficiency,32 including not only stenosis evaluation but plaque tissue characterization. Key candidates for automation are those tasks most challenging for human readers, such as resolving tissue interfaces despite partial volume effects, calcium blooming, and overlapping HU ranges—all while providing objective quantification. Furthermore, human interpretation, despite experience, is still prone to fatigue and subjective bias. In addition, there remains a shortage of expert human readers.
To overcome these limitations, software enabled by AI can facilitate certain stages of the imaging process from reconstruction33 to segmentation,34 measurement,35 and interpretation.36 As the use of diagnostic CTA continues to increase, tools to increase diagnostic specificity, accuracy, and operational efficiency should be embraced.
More information is also required to understand the lesion-level and patient-level prognostic risk of the individual components and the extent of vulnerable plaques as determined by CTA using AI-enabled plaque recognition and quantification software. Finally, and most importantly, randomized trials need to be conducted to demonstrate that this information is actionable, that is, whether treating vulnerable plaques (and patients) with either pharmacologic intensification and/or focal therapies improves clinical outcomes.
METHODOLOGIES FOR CTA PLAQUE CHARACTERIZATION
HU Thresholding Approach
The most common method to develop algorithms for CTA plaque analysis is based on HUs.15–18 HUs quantify the number of x-rays that are attenuated or absorbed by the tissue. Although HUs may have a wide range, they are typically displayed on a scale ranging from a value from 1 (black) to 256 (white) (Fig. 1). HUs are also useful in automating CTA reading and have been used to evaluate and quantify tissues and fluids. Several academic groups have developed HU-based AI algorithms for CTA to characterize atherosclerotic plaques,37 and several commercial CTA automated quantitative software packages with proprietary HU-based AI algorithms have been developed (Table 1). These products apply unstandardized, semiautomated threshold-based segmentation with generally predefined HU ranges to measure certain tissue criteria of plaque as defined on the HU scale (eg, low attenuation plaque, non-calcified plaque, calcified plaque). As described below, these technologies efficiently assess plaque volume as validated against various in vivo reference standards (representative studies highlighted in Table 2); however, they are limited in their ability to characterize plaque composition beyond calcium.
TABLE 1 -
Selected CTA Automated Quantitative Software Utilizing HU-based AI Algorithms
||Diagnostic product performs quantitative atherosclerotic plaque characterization by CT
||Coronary CTA-based plaque visualization and quantification tool developed using Object Research Systems (ORS) technology
||Visualization and measurement of vessel walls and plaque characteristics in arterial vessels using color-defined HU ranges
||Plaque measurement supported by percentile database; supports calcium scoring measurement of Agatston score, volume, and mass
|Syngo through Frontier Coronary Plaque Analysis
||Volumetric quantification and differentiation of lipid, fibrous, and calcified plaques
||Medis Medical Imaging Systems
||Degree of stenosis, plaque characterization, and perivascular adipose tissue analysis
AI indicates artificial intelligence; CTA, computed tomography angiography; HU, Hounsfield Unit.
TABLE 2 -
CTA Automated Quantitative Software Approaches for Coronary Plaque Composition in Comparison to Other Diagnostic Methods
||To explore the accuracy of the automated algorithm for the evaluation of non-calcified and calcified plaques from CTA in comparison with IVUS
||In non-calcified plaque volumes, there was an excellent correlation between IVUS and multidetector CT with an automated approach
||To assess the efficacy of invasive FFR-determined hemodynamic significance of automated plaque measurements from CTA
||Automatic assessment of plaque burden improves the identification of lesion-specific hemodynamic significance by FFR compared with stenosis severity
||Myocardial perfusion SPECT
||To determine whether quantitative plaque parameters from CTA predict ischemia measured by myocardial perfusion SPECT
||Low-density non-calcified plaque contrast density difference from CTA are associated with ischemia measured by myocardial perfusion SPECT
||To validate the automatic assessment of stenosis severity and plaque constitution compared with IVUS VH
||Automatic software tool showed a good correlation with IVUS VH in characterizing plaque type
||To investigate whether CTA-based HRP definitions improve the diagnostic performance for ACS compared with IVUS-specific definitions
||CTA-based HRP definitions may improve the characterization of culprit lesions compared with traditional IVUS-specific definitions
||To evaluate the diagnostic performance of plaque analysis software between semiautomatic and automatic approach compared with IVUS
||Automatic plaque analysis showed an excellent correlation with IVUS in plaque quantification
ACS indicates acute coronary syndrome; CTA, computed tomography angiography; FFR, fractional flow reserve; IVUS, intravascular ultrasound; SPECT, single-photon emission computed tomography; VH, virtual histology.
A semiautomated product (CT SUREplaque, Canon Medical Systems, Otawara, Japan) uses color-defined HU ranges to visualize and measure vessel walls and plaque characteristics in arterial vessels.44 It provides data on plaque burden and remodeling, lumen area and diameter, plaque area and volume, and the ratio of wall area and lumen area. Another diagnostic AI-based software enhancement for CTA-based quantitative atherosclerotic plaque characterization is available (Cleerly Health, New York, NY), which can assess manually selected plaques. It can detect stenosis, along with calcified and non-calcified lesions, lesion length and volume, and LAP volume. The CT Evaluation by Artificial Intelligence for Atherosclerosis, Stenosis and Vascular Morphology (CLARIFY) study demonstrated that high-risk plaque features reported using the Cleerly AI algorithm had a high correlation to consensus from 3 readers for the determination of maximum percent stenosis (greater than or less than 70%) and CAD Reporting and Data System45 score but the correlation for individual readers was only fair.46 Furthermore, the Scottish Computed Tomography of the Heart trial recently quantified the burden of LAP by a semiautomatic coronary CTA-based plaque visualization and quantification tool (Autoplaque, Cedars Sinai, Los Angeles, CA) and concluded that LAP burden is the strongest predictor of MACE.28 Plaque components, including remodeling index, calcified plaque, non-calcified plaque, and LAP volume, were validated against subjective reader estimates and demonstrated excellent intraobserver, interobserver, and scan-rescan reproducibility. Finally, the Intuition application (TeraRecon, Durham, NC) uses a zero-click centerline creation/extraction technique to segment coronary arteries along plaque measurements supported by a percentile database. This software determines calcium scoring (Agatston score) and lesion volume measures.
While the above-referenced HU threshold-based software solutions provide a certain level of plaque characterization and improve reader efficiency and repeatability, there are several key limitations of these methods. First, it has been shown that the tube voltage setting influences HU and, consequently plaque morphology, as reported by the Progression of AtheRosclerotic PlAque Determined by Computed TomoGraphic Angiography Imaging trial.47
As shown by the study in 1236 patients, low tube voltage results in an increase in luminal HU that triggers an increase in calcified plaque and a reduction in fibrofatty and necrotic core. Therefore, kVp settings should always be considered when plaque characteristics are assessed. Second, due to the lack of human histology validation, their outputs refer to nonspecific terms such as LAP rather than specific tissue types. Third, the HU method is unable to characterize LRNC or IPH, despite LRNC being a key driver of MACE as demonstrated in numerous outcomes-based studies.15–18 Fourth, partial volume effects cannot be mitigated by the HU approach, obscuring the true interfaces between the lumen, calcium, and LRNC, thus affecting both measurement accuracy and visual interpretation. HU-based AI software may exacerbate this problem since only whole voxels are measured, which is a limitation as small structures in the vessel lumen, walls, and plaque substructures may require greater discrimination. Finally, with the lack of anatomy segmentation and primary focus on stenosis and visual reading, the outer walls are often not properly segmented but rather are estimated with a fixed radius away from the lumen constant. In most applications, a contour is drawn outside the lumen that is proportional to the lumen. Numerous hand edits are required to obtain reasonable plaque burden measurements. This contributes to inaccuracies in the measurement of atheroma cap thickness, exacerbating the inherent resolution constraints of CTA.
Plaque Assessment Based on Radiomics
Radiomics enables complete phenotype quantification by extracting a wide number of computational quantitative features from medical images.48 With this process, the amount of information increases, facilitating to display even more accurate details from radiologic images. Previous studies showed that the detection of vulnerable plaques with radiomics is achievable. Moreover, it may provide a more precise identification of advanced atherosclerotic lesions compared with conventional methods including sodium-fluoride positron emission tomography (NaF18-PET), intravascular ultrasound (IVUS), and optical coherence tomography (OCT).49
In the identification of atheromatous lesions, radiomics-based machine learning was superior over visual and histologic-based assessment,50 which leads to the conclusion that with the help of this novel analytic method, the diagnostic accuracy of coronary CT angiography can be further improved. Furthermore, a specific radiomic phenotype could be found in culprit lesions compared with non-culprit and stable CAD lesions. As a result of these findings, CTA-based radiomic plaque analysis has the potential to provide a tool to identify culprit and non-culprit lesions in myocardial infarction and in stable CAD.51
Histopathological Validation Approach
Imaging modalities such as radiography, CT, and magnetic resonance imaging are fundamental diagnostic tools and play a critical role in clinical practice. Each of these modalities, however, is limited in the assessment of plaque composition when validated against a histology gold standard.
Specifically, a plaque characterization algorithm should be able to quantify plaque morphology using histologically validated logic to mitigate calcium blooming, HU overlap, and partial volume artifacts in 3D space, thus delivering semiautomatic, histologically, and contextually accurate plaque structures within accurate inner and outer artery wall boundaries. There are several key advantages of histologically validated plaque assessment and quantitation compared with HU thresholding. First, a histologically validated approach can uniquely characterize LRNC and IPH, 2 key drivers of MACE. AI-based plaque composition recognition and composition may further enhance efficiency as even trained readers may have difficulty differentiating or quantifying different tissue types. Second, such an approach mitigates partial volume effects and enables the delineation of true interfaces between the lumen, calcium, and LRNC, affecting both visual interpretation and measurement accuracy. Finally, outer walls are segmented with an accurate radius, delivering precise measurements with minimal human editing, mostly at vessel branching points, including cap thickness.
The histologic validation of an AI-based CTA image processing algorithm (ElucidVivo [formerly known as vascuCAP], Elucid Bioimaging, Boston, MA) has been demonstrated in 333 histologic sections of human carotid atherosclerotic plaques from specimens derived during endarterectomy procedures.19 Histologic plaque features, including calcification, LRNC, and matrix, were assessed with hematoxylin-eosin staining. Beyond simple HU-based tissue typing, the software uses deterministic algorithms that account for density distributions at super-resolution, trained and validated against histopathological specimens, to quantify true tissue types. Quantitative plaque analysis of CTA compared with the reference standard histologic analysis (Fig. 2) showed significant correlations between plaque calcification, LRNC, and matrix. This study demonstrates the potential of the CTA AI-based software to categorize and quantify plaque tissue characteristics with reliable accuracy and low reader variability.19 As a result, the FDA has approved ElucidVivo to characterize specific tissue types corroborated with histopathology and arterial structural measurements, which were correlated against phantoms with specific performance metrics in its FDA label.
A major advantage of this tool is its validation with carotid artery samples from living patients. Carotid artery endarterectomy procedures enable tissue collection from living patients, and samples can be collected during surgery following a CT angiography, which are more readily obtainable than coronary artery tissues and thus were used for training and validation of the ElucidVivo data set. Validation of tissue excised from live patients rather than post-mortem avoids artifacts arising from tissue degradation after death. It may be questioned whether image recognition compared against carotid tissue applies to plaque in other vascular beds. While it is true that the frequency and relative contribution of plaque progression, rupture versus erosion, and thrombosis may vary in coronary compared with the carotid arteries and other arterial beds, in fact, plaque characteristics such as a large atheromatous lipid-rich core, thin fibrous cap, outward remodeling, infiltration of the plaque with macrophages and lymphocytes and thinning of the media predispose to thrombosis in both carotid and coronary artery disease.52,53 The clinical differences in risk and disease manifestation across the coronary and carotid artery beds are accounted for by the use of additional factors, including vessel size, plaque geometry and rheology, collateral flow, and the downstream circulations (myocardium vs. brain) to augment the tissue characterizations. Furthermore, carotid atherosclerosis has been shown to be an independent predictor for coronary artery-derived MACE (even in patients without pre-existing CAD54), suggesting a common underlying pathogenesis, which was further supported by the Multi-Ethnic Study of Atherosclerosis (MESA).55
The histology-validated AI-driven software tool offers several benefits. Whereas standard CTA is capable of differentiating lipid-rich regions from fibrous tissue and calcification,56,57 its ability to classify plaque regions is limited to categories such as “hard” (ie, calcified) and “soft” (ie, non-calcified) components, determined based on visual evaluation without the aid of advanced algorithms or histologic correlation. ElucidVivo bridges this gap by mitigating specific issues in CT reconstruction known to affect accurate measurements of atherosclerotic plaque composition in routinely acquired CTA (Fig. 3).
In addition, unlike incumbent software, the histologically validated approach corrects HU values for partial volume effects (eg, blooming artifacts from calcified plaque) and enhances discrimination of LRNC and IPH using an iterative optimization algorithm informed but not constrained by partially overlapping HU ranges for different tissue types. To address the imaging artifacts from calcified plaque, which hinder the accuracy of sub-voxel measurements, ElucidVivo compensates for the imaging system point spread function with an algorithm that probabilistically estimates the most likely fine structure given the magnitude of scanner blur. Such image-based determination of blur coupled with sub-voxel analysis of plaque component densities leads to more accurate scoring of coronary artery calcification, useful by itself, but also allowing the quantification of subtle changes in LRNC and consequently cap thickness (Table 3).58,61,62
TABLE 3 -
Clinical Studies Demonstrating the Application of Histologically Validated Plaque Analysis
||To characterize LRNC in the coronary arteries in psoriasis before and after biological therapy over 1 y
||Provided evidence of a potential reduction in LRNC with the treatment of systemic inflammation
||Determining the mechanism of action for drugs
||To evaluate the effect of Icosapent Ethyl on changes in coronary plaque morphology
||Provided evidence that Icosapent Ethyl lowers serum lipid levels, causing reductions in features that contribute to instability, migrating the plaques to more stable phenotypes
||Determining the mechanism of action for drugs
||To develop a predictive model for major adverse neurological events (MANE) in patients with carotid atherosclerotic plaques
||ElucidVivo-based risk stratification in carotid atherosclerosis is a better predictor of MANE than the traditionally utilized degree of stenosis alone
||Implementing this predictive model on asymptomatic patients in the clinical setting will help identify and inform treatment for those at high risk for future MANE
LRNC indicates lipid-rich necrotic core.
There are several potential limitations and evidence gaps with the histology-driven approach. Histologic validation was performed on carotid arteries only. Coronary artery analysis was implemented by adapting similar histologic characteristics based on carotid artery tissue composition, which may be influenced by differences in vessel and plaque sizes. Sampling live tissue is a complex and invasive process, and some fixation artifacts may be unavoidable. Histologic validation of CTA images poses difficulties when attempting to match (co-register) the location of a 2D histologic slice within the 3D CTA image. This has been mitigated with software solutions (HistoMatch, Elucid Bioimaging) developed to correlate histologic slices, with locations in the CT image, enabling quantitative plaque analysis.19 Most importantly, the goal of detailed classification and quantification of vascular plaque components is to use this information to inform patient prognosis and management, as well as the development of novel therapies. Studies have provided evidence supporting the clinical utility of histologically validated plaque analysis (Table 2).58–60 Nonetheless, additional prospective investigations are required to demonstrate the superior prognostic utility of CTA-based plaque composition analysis compared with HU-based analysis, and to determine the hazard ratios of different thresholds of plaque-level, vessel-level and patient-level extent of total disease and high-risk plaque that places individual atheromas and patients at increased risk for future cardiovascular events. Ultimately, this knowledge will need to be applied to randomized trials to demonstrate that such information is actionable - that is, directs pharmacologic intensification, revascularization, or other therapies that improve clinical outcomes.
In the last decade, significant evidence has been generated to support the prognostic utility of atherosclerosis evaluation by coronary CTA. Compared with standard approaches using HU thresholding, histology-based plaque analysis software enables the more accurate and detailed identification and quantification of plaque extent and composition. Specifically, the ability of histologically validated plaque quantification to identify high-risk plaque components (eg, LRNC and IPH) may be useful in identifying vulnerable plaques that place patients at risk for future adverse cardiovascular events. The same software has been validated to enhance the accuracy of plaque quantification, structural measurements including cap thickness, and dynamic 3-dimensional visual representation. The AI-based analysis tools extend what can be discerned by the experienced CTA reader. Given these attributes, plaque quantification software has the potential to become increasingly adopted in clinical care. Additional studies are warranted to determine whether information from advanced plaque assessment tools such as ElucidVivo can be applied to therapeutic interventions to improve clinical outcomes for patients with atherosclerotic cardiovascular disease.
1. Neuhaus E, Weiss K, Bastkowski R, et al. Accelerated aortic 4d flow cardiovascular magnetic resonance using compressed sensing: Applicability, validation and clinical integration. J Cardiovasc Magn Reson. 2019;21:65.
2. Virani SS, Alonso A, Aparicio HJ, et al. Heart disease and stroke statistics-2021 update: A report from the American heart association. Circulation. 2021;143:e254–e743.
3. Piepoli MF, Hoes AW, Agewall S, et al. 2016 European guidelines on cardiovascular disease prevention in clinical practice: The sixth joint task force of the European society of cardiology and other societies on cardiovascular disease prevention in clinical practice (constituted by representatives of 10 societies and by invited experts)developed with the special contribution of the European Association For Cardiovascular Prevention & Rehabilitation (EACPR). Eur Heart J. 2016;37:2315–2381.
4. Conroy RM, Pyorala K, Fitzgerald AP, et al. Estimation of ten-year risk of fatal cardiovascular disease in Europe: The score project. Eur Heart J. 2003;24:987–1003.
5. D’Agostino RB Sr, Vasan RS, Pencina MJ, et al. General cardiovascular risk profile for use in primary care: The framingham heart study. Circulation. 2008;117:743–753.
6. Arnett DK, Blumenthal RS, Albert MA, et al. 2019 acc/aha guideline on the primary prevention of cardiovascular disease: A report of the American college of cardiology/American heart association task force on clinical practice guidelines. J Am Coll Cardiol. 2019;74:e177–e232.
7. Barkat M, Roy I, Antoniou SA, et al. Systematic review and network meta-analysis of treatment strategies for asymptomatic carotid disease. Sci Rep. 2018;8:4458.
8. Czarnecki A, Qiu F, Elbaz-Greener G, et al. Variation in revascularization practice and outcomes in asymptomatic stable ischemic heart disease. JACC Cardiovasc Interv. 2019;12:232–41.
9. van Assen M, Varga-Szemes A, Schoepf UJ, et al. Automated plaque analysis
for the prognostication
of major adverse cardiac events. Eur J Radiol. 2019;116:76–83.
10. Yamamoto H, Kihara Y, Kitagawa T, et al. Coronary plaque characteristics in computed tomography
and 2-year outcomes: The predict study. J Cardiovasc Comput Tomogr. 2018;12:436–443.
11. Stone PH, Saito S, Takahashi S, et al. Prediction of progression of coronary artery disease
and clinical outcomes using vascular profiling of endothelial shear stress and arterial plaque characteristics: The prediction study. Circulation. 2012;126:172–181.
12. Kolossvary M, Szilveszter B, Merkely B, et al. Plaque imaging with ct-a comprehensive review on coronary ct angiography based risk assessment. Cardiovasc Diagn Ther. 2017;7:489–506.
13. Cau R, Flanders A, Mannelli L, et al. Artificial intelligence in computed tomography
plaque characterization: A review. Eur J Radiol. 2021;140:109767.
14. Al-Mallah MH, Sakr S. Artificial intelligence for plaque characterization: A scientific exercise looking for a clinical application. Atherosclerosis. 2019;288:158–159.
15. Feuchtner G, Kerber J, Burghard P, et al. The high-risk criteria low-attenuation plaque <60 hu and the napkin-ring sign are the most powerful predictors of mace: A long-term follow-up study. Eur Heart J Cardiovasc Imaging. 2017;18:772–779.
16. Matsumoto H, Watanabe S, Kyo E, et al. Standardized volumetric plaque quantification and characterization from coronary ct angiography: A head-to-head comparison with invasive intravascular ultrasound. Eur Radiol. 2019;29:6129–6139.
17. Saba L, Francone M, Bassareo PP, et al. Ct attenuation analysis of carotid intraplaque hemorrhage. AJNR Am J Neuroradiol. 2018;39:131–137.
18. van Rosendael AR, Narula J, Lin FY, et al. Association of high-density calcified 1k plaque with risk of acute coronary syndrome. JAMA Cardiol. 2020;5:282–290.
19. Sheahan M, Ma X, Paik D, et al. Atherosclerotic plaque tissue: Noninvasive quantitative assessment of characteristics with software-aided measurements from conventional ct angiography. Radiology. 2018;286:622–631.
20. Patel MR, Calhoon JH, Dehmer GJ, et al. ACC/AATS/AHA/ASE/ASNC/SCAI/SCCT/STS 2017 appropriate use criteria for coronary revascularization in patients with stable ischemic heart disease: A report of the American college of cardiology appropriate use criteria task force, American association for thoracic surgery, American heart association, American society of echocardiography, American society of nuclear cardiology, society for cardiovascular angiography and interventions, society of cardiovascular computed tomography
, and society of thoracic surgeons. J Thorac Cardiovasc Surg. 2019;157:e131–e161.
21. Skinner JS, Smeeth L, Kendall JM, et al. Nice guidance. Chest pain of recent onset: Assessment and diagnosis of recent onset chest pain or discomfort of suspected cardiac origin. Heart. 2010;96:974–978.
22. Gulati M, Levy PD, Mukherjee D, et al. 2021 aha/acc/ase/chest/saem/scct/scmr guideline for the evaluation and diagnosis of chest pain: Executive summary: A report of the American college of cardiology/American heart association joint committee on clinical practice guidelines. Circulation. 2021;144:e368–e454.
23. Knuuti J, Wijns W, Saraste A, et al. 2019 esc guidelines for the diagnosis and management of chronic coronary syndromes. Eur Heart J. 2020;41:407–477.
24. Bauer RW, Thilo C, Chiaramida SA, et al. Noncalcified atherosclerotic plaque burden at coronary ct angiography: A better predictor of ischemia at stress myocardial perfusion imaging than calcium score and stenosis severity. AJR Am J Roentgenol. 2009;193:410–418.
25. Bittencourt MS, Hulten E, Ghoshhajra B, et al. Prognostic value of nonobstructive and obstructive coronary artery disease
detected by coronary computed tomography
angiography to identify cardiovascular events. Circ Cardiovasc Imaging. 2014;7:282–291.
26. Ferencik M, Mayrhofer T, Bittner DO, et al. Use of high-risk coronary atherosclerotic plaque detection for risk stratification of patients with stable chest pain: A secondary analysis of the promise randomized clinical trial. JAMA Cardiol. 2018;3:144–152.
27. Andreini D, Magnoni M, Conte E, et al. Coronary plaque features on CTA can identify patients at increased risk of cardiovascular events. JACC Cardiovasc Imaging. 2020;13:1704–1717.
28. Williams MC, Kwiecinski J, Doris M, et al. Low-attenuation noncalcified plaque on coronary computed tomography
angiography predicts myocardial infarction: Results from the multicenter scot-heart trial (Scottish computed tomography
of the heart). Circulation. 2020;141:1452–1462.
29. Williams MC, Moss AJ, Dweck M, et al. Coronary artery plaque characteristics associated with adverse outcomes in the scot-heart study. J Am Coll Cardiol. 2019;73:291–301.
30. SCOT-HEART Investigators, Newby DE, Adamson PD, et al. Coronary CT angiography and 5-year risk of myocardial infarction. N Engl J Med. 2018;379:924–933.
31. Indraratna P, Khasanova E, Gulsin GS, et al. Plaque progression: Where, why, and how fast? A review of what we have learned from the analysis of patient data from the paradigm registry. J Cardiovasc Comput Tomogr. 2022;16:294–302.
32. Yacoub B, Varga-Szemes A, Schoepf UJ, et al. Impact of artificial intelligence assistance on chest ct interpretation times: A prospective randomized study. AJR Am J Roentgenol. 2022;219:743–751.
33. Benz DC, Benetos G, Rampidis G, et al. Validation of deep-learning image reconstruction for coronary computed tomography
angiography: Impact on noise, image quality and diagnostic accuracy. J Cardiovasc Comput Tomogr. 2020;14:444–451.
34. Morris ED, Ghanem AI, Dong M, et al. Cardiac substructure segmentation with deep learning for improved cardiac sparing. Med Phys. 2020;47:576–586.
35. Muscogiuri G, Chiesa M, Trotta M, et al. Performance of a deep learning algorithm for the evaluation of cad-rads classification with CCTA. Atherosclerosis. 2020;294:25–32.
36. Motwani M, Dey D, Berman DS, et al. Machine learning for prediction of all-cause mortality in patients with suspected coronary artery disease
: A 5-year multicentre prospective registry analysis. Eur Heart J. 2017;38:500–507.
37. Lee H, Emrich T, Schoepf UJ, et al. Artificial intelligence in cardiac ct: Automated calcium scoring and plaque analysis
. Current Cardiovascular Imaging Reports. 2020;13. https://doi.org/10.1007/s12410-020-09549-9
38. Dey D, Schepis T, Marwan M, et al. Automated three-dimensional quantification of noncalcified coronary plaque from coronary CT angiography: Comparison with intravascular us. Radiology. 2010;257:516–522.
39. Diaz-Zamudio M, Dey D, Schuhbaeck A, et al. Automated quantitative plaque burden from coronary ct angiography noninvasively predicts hemodynamic significance by using fractional flow reserve in intermediate coronary lesions. Radiology. 2015;276:408–415.
40. Diaz-Zamudio M, Fuchs TA, Slomka P, et al. Quantitative plaque features from coronary computed tomography
angiography to identify regional ischemia by myocardial perfusion imaging. Eur Heart J Cardiovasc Imaging. 2017;18:499–507.
41. de Graaf MA, Broersen A, Kitslaar PH, et al. Automatic quantification and characterization of coronary atherosclerosis with computed tomography
coronary angiography: Cross-correlation with intravascular ultrasound virtual histology. Int J Cardiovasc Imaging. 2013;29:1177–1190.
42. Bittner DO, Mayrhofer T, Puchner SB, et al. Coronary computed tomography
angiography-specific definitions of high-risk plaque features improve detection of acute coronary syndrome. Circ Cardiovasc Imaging. 2018;11:e007657.
43. Park HB, Lee BK, Shin S, et al. Clinical feasibility of 3D automated coronary atherosclerotic plaque quantification algorithm on coronary computed tomography
angiography: Comparison with intravascular ultrasound. Eur Radiol. 2015;25:3073–3083.
44. Kiran Munnur R, Andrews J, Kataoka Y, et al. Serial coronary plaque assessment using computed tomography
coronary angiography. Circ Cardiovasc Imaging. 2019;12:e008404.
45. Cury RC, Abbara S, Achenbach S, et al. Cad-rads(TM) coronary artery disease
- reporting and data system. An expert consensus document of the society of cardiovascular computed tomography
(SCCT), the American college of radiology (ACR) and the north American society for cardiovascular imaging (NASCI). Endorsed by the American college of cardiology. J Cardiovasc Comput Tomogr. 2016;10:269–281.
46. Choi AD, Marques H, Kumar V, et al. Ct evaluation by artificial intelligence for atherosclerosis, stenosis and vascular morphology (clarify): A multi-center, international study. J Cardiovasc Comput Tomogr. 2021;15:470–476.
47. Takagi H, Leipsic JA, Indraratna P, et al. Association of tube voltage with plaque composition on coronary CT angiography: Results from paradigm registry. JACC Cardiovasc Imaging. 2021;14:2429–2440.
48. Aerts HJ. The potential of radiomic-based phenotyping in precision medicine: A review. JAMA Oncol. 2016;2:1636–1642.
49. Kolossvary M, Park J, Bang JI, et al. Identification of invasive and radionuclide imaging markers of coronary plaque vulnerability using radiomic analysis of coronary computed tomography
angiography. Eur Heart J Cardiovasc Imaging. 2019;20:1250–1258.
50. Kolossvary M, Karady J, Kikuchi Y, et al. Radiomics versus visual and histogram-based assessment to identify atheromatous lesions at coronary CT angiography: an ex vivo study. Radiology. 2019;293:89–96.
51. Lin A, Kolossvary M, Cadet S, et al. Radiomics-based precision phenotyping identifies unstable coronary plaques from computed tomography
angiography. JACC Cardiovasc Imaging. 2022;15:859–871.
52. Sigala F, Oikonomou E, Antonopoulos AS, et al. Coronary versus carotid artery plaques. Similarities and differences regarding biomarkers morphology and prognosis. Curr Opin Pharmacol. 2018;39:9–18.
53. Ibrahimi P, Jashari F, Nicoll R, et al. Coronary and carotid atherosclerosis: How useful is the imaging? Atherosclerosis. 2013;231:323–333.
54. Steinvil A, Sadeh B, Bornstein NM, et al. Impact of carotid atherosclerosis on the risk of adverse cardiac events in patients with and without coronary disease. Stroke. 2014;45:2311–2317.
55. Gepner AD, Young R, Delaney JA, et al. Comparison of coronary artery calcium presence, carotid plaque presence, and carotid intima-media thickness for cardiovascular disease prediction in the multi-ethnic study of atherosclerosis. Circ Cardiovasc Imaging. 2015;8:e002262.
56. Marwan M, Taher MA, El Meniawy K, et al. In vivo CT detection of lipid-rich coronary artery atherosclerotic plaques using quantitative histogram analysis: a head to head comparison with ivus. Atherosclerosis. 2011;215:110–115.
57. Schlett CL, Maurovich-Horvat P, Ferencik M, et al. Histogram analysis of lipid-core plaques in coronary computed tomographic angiography: Ex vivo validation against histology. Invest Radiol. 2013;48:646–653.
58. Choi H, Uceda DE, Dey AK, et al. Treatment of psoriasis with biologic therapy is associated with improvement of coronary artery plaque lipid-rich necrotic core: Results from a prospective, observational study. Circ Cardiovasc Imaging. 2020;13:e011199.
59. Lakshmanan S, Buckler AJ, Bhatt DL, et al. Effect of icosapent ethyl on changes in coronary plaque characteristics at 9 months in patients with elevated triglycerides on statin therapy: Insights from evaporate. Circulation. 2020;142:A15389.
60. Lal BK, Kashyap VS, Patel JB, et al. Novel application of artificial intelligence algorithms to develop a predictive model for major adverse neurologic events in patients with carotid atherosclerosis. J Vasc Surg. 2020;72:e176–e177.
61. Lee SH, Hong D, Dai N, et al. Anatomic and hemodynamic plaque characteristics for subsequent coronary events. Front Cardiovasc Med. 2022;9:871450.
62. Varga-Szemes A, Schoepf UJ, Maurovich-Horvat P, et al. Coronary plaque assessment of vasodilative capacity by CT angiography effectively estimates fractional flow reserve. Int J Cardiol. 2021;331:307–315.