Artificial intelligence in coronary artery calcium score: rationale, different approaches, and outcomes.

Almost 35 years after its introduction, coronary artery calcium score (CACS) not only survived technological advances but became one of the cornerstones of contemporary cardiovascular imaging. Its simplicity and quantitative nature established it as one of the most robust approaches for atherosclerotic cardiovascular disease risk stratification in primary prevention and a powerful tool to guide therapeutic choices. Groundbreaking advances in computational models and computer power translated into a surge of artificial intelligence (AI)-based approaches directly or indirectly linked to CACS analysis. This review aims to provide essential knowledge on the AI-based techniques currently applied to CACS, setting the stage for a holistic analysis of the use of these techniques in coronary artery calcium imaging. While the focus of the review will be detailing the evidence, strengths, and limitations of end-to-end CACS algorithms in electrocardiography-gated and non-gated scans, the current role of deep-learning image reconstructions, segmentation techniques, and combined applications such as simultaneous coronary artery calcium and pulmonary nodule segmentation, will also be discussed.

Intra- and inter-reader reproducibility in quantitative coronary plaque analysis on coronary computed tomography angiography.

PURPOSE: Coronary artery plaque burden, low attenuation non-calcified plaque (LAP), and pericoronary adipose tissue (PCAT) on coronary CT angiography (CCTA), have been linked to future cardiac events. The purpose of this study was to evaluate intra- and inter reader reproducibility in the quantification of coronary plaque burden and its characteristics using an artificial intelligence-enhanced semi-automated software. MATERIALS AND METHODS: A total of 10 women and 6 men, aged 52 (IQR 49-58) underwent CCTA using a Siemens Somatom Force, Somatom Definition AS and Somatom Definition Flash scanners. Two expert readers utilized dedicated semi-automatic software (vascuCAP, Elucid Bioimaging, Wenham, MA) to assess calcified plaque, low attenuation plaque and PCAT. Readers were blinded to all clinical information and repeated their analysis at 6 weeks in random order to minimize recall bias. Data analysis was performed on the right and left coronary arteries. Intra- and inter-reader reproducibility was compared using Pearson correlation coefficient, while absolute values between analyses and readers were compared with paired non-parametric tests. This is a sub-study of the Specialized Center of Research Excellence (SCORE) clinical trial (5U54AG062334). RESULTS: A total of 64 vessels from 16 patients were analyzed. Intra-reader Pearson correlation coefficients for calcified plaque volume, LAP volume and PCAT volumes were 0.96, 0.99 and 0.92 for reader 1 and 0.94, 0.94 and 0.95 for reader 2, respectively, (all p < 0.0001). Inter-reader Pearson correlation coefficients for calcified plaque volume, LAP and PCAT volumes were 0.92, 0.96 and 0.78, and 0.99, 0.99 and 0.93 on the second analyses, all had a p value <0.0001. There was no significant bias on the corresponding Bland-Altman analyses. CONCLUSION: Volume measurement of coronary plaque burden and PCAT volume can be performed with high intra- and inter-reader agreement.

The Role of Artificial Intelligence in Cardiac Imaging.

Artificial intelligence (AI) is having a significant impact in medical imaging, advancing almost every aspect of the field, from image acquisition and postprocessing to automated image analysis with outreach toward supporting decision making. Noninvasive cardiac imaging is one of the main and most exciting fields for AI development. The aim of this review is to describe the main applications of AI in cardiac imaging, including CT and MR imaging, and provide an overview of recent advancements and available clinical applications that can improve clinical workflow, disease detection, and prognostication in cardiac disease.

Left Main Coronary Artery Calcium and Diabetes Confer Very-High-Risk Equivalence in Coronary Artery Calcium >1,000.

BACKGROUND: Although a coronary artery calcium (CAC) of ≥1,000 is a subclinical atherosclerosis threshold to consider combination lipid-lowering therapy, differentiating very high from high atherosclerotic cardiovascular disease (ASCVD) risk in this patient population is not well-defined. OBJECTIVES: Among persons with a CAC of ≥1,000, the authors sought to identify risk factors equating with very high-risk ASCVD mortality rates. METHODS: The authors studied 2,246 asymptomatic patients with a CAC of ≥1,000 from the CAC Consortium without a prior ASCVD event. Cox proportional hazards regression modelling was performed for ASCVD mortality during a median follow-up of 11.3 years. Crude ASCVD mortality rates were compared with those reported for secondary prevention trial patients classified as very high risk, defined by ≥2 major ASCVD events or 1 major event and ≥2 high-risk conditions (1.4 per 100 person-years). RESULTS: The mean age was 66.6 years, 14% were female, and 10% were non-White. The median CAC score was 1,592 and 6% had severe left main (LM) CAC (vessel-specific CAC ≥300). Diabetes (HR: 2.04 [95% CI: 1.47-2.83]) and severe LM CAC (HR: 2.32 [95% CI: 1.51-3.55]) were associated with ASCVD mortality. The ASCVD mortality per 100 person-years for all patients was 0.8 (95% CI: 0.7-0.9), although higher rates were observed for diabetes (1.4 [95% CI: 0.8-1.9]), severe LM CAC (1.3 [95% CI: 0.6-2.0]), and both diabetes and severe LM CAC (7.1 [95% CI: 3.4-10.8]). CONCLUSIONS: Among asymptomatic patients with a CAC of ≥1,000 without a prior index event, diabetes, and severe LM CAC define very high risk ASCVD, identifying individuals who may benefit from more intensive prevention therapies across several domains, including low-density lipoprotein-cholesterol lowering.

Implications of Bias in Artificial Intelligence: Considerations for Cardiovascular Imaging.

PURPOSE OF REVIEW: Bias in artificial intelligence (AI) models can result in unintended consequences. In cardiovascular imaging, biased AI models used in clinical practice can negatively affect patient outcomes. Biased AI models result from decisions made when training and evaluating a model. This paper is a comprehensive guide for AI development teams to understand assumptions in datasets and chosen metrics for outcome/ground truth, and how this translates to real-world performance for cardiovascular disease (CVD). RECENT FINDINGS: CVDs are the number one cause of mortality worldwide; however, the prevalence, burden, and outcomes of CVD vary across gender and race. Several biomarkers are also shown to vary among different populations and ethnic/racial groups. Inequalities in clinical trial inclusion, clinical presentation, diagnosis, and treatment are preserved in health data that is ultimately used to train AI algorithms, leading to potential biases in model performance. Despite the notion that AI models themselves are biased, AI can also help to mitigate bias (e.g., bias auditing tools). In this review paper, we describe in detail implicit and explicit biases in the care of cardiovascular disease that may be present in existing datasets but are not obvious to model developers. We review disparities in CVD outcomes across different genders and race groups, differences in treatment of historically marginalized groups, and disparities in clinical trials for various cardiovascular diseases and outcomes. Thereafter, we summarize some CVD AI literature that shows bias in CVD AI as well as approaches that AI is being used to mitigate CVD bias.

Value Creation Through Artificial Intelligence and Cardiovascular Imaging: A Scientific Statement From the American Heart Association.

Multiple applications for machine learning and artificial intelligence (AI) in cardiovascular imaging are being proposed and developed. However, the processes involved in implementing AI in cardiovascular imaging are highly diverse, varying by imaging modality, patient subtype, features to be extracted and analyzed, and clinical application. This article establishes a framework that defines value from an organizational perspective, followed by value chain analysis to identify the activities in which AI might produce the greatest incremental value creation. The various perspectives that should be considered are highlighted, including clinicians, imagers, hospitals, patients, and payers. Integrating the perspectives of all health care stakeholders is critical for creating value and ensuring the successful deployment of AI tools in a real-world setting. Different AI tools are summarized, along with the unique aspects of AI applications to various cardiac imaging modalities, including cardiac computed tomography, magnetic resonance imaging, and positron emission tomography. AI is applicable and has the potential to add value to cardiovascular imaging at every step along the patient journey, from selecting the more appropriate test to optimizing image acquisition and analysis, interpreting the results for classification and diagnosis, and predicting the risk for major adverse cardiac events.

Fusion Modeling: Combining Clinical and Imaging Data to Advance Cardiac Care.

In addition to the traditional clinical risk factors, an increasing amount of imaging biomarkers have shown value for cardiovascular risk prediction. Clinical and imaging data are captured from a variety of data sources during multiple patient encounters and are often analyzed independently. Initial studies showed that fusion of both clinical and imaging features results in superior prognostic performance compared with traditional scores. There are different approaches to fusion modeling, combining multiple data resources to optimize predictions, each with its own advantages and disadvantages. However, manual extraction of clinical and imaging data is time and labor intensive and often not feasible in clinical practice. An automated approach for clinical and imaging data extraction is highly desirable. Convolutional neural networks and natural language processing can be utilized for the extraction of electronic medical record data, imaging studies, and free-text data. This review outlines the current status of cardiovascular risk prediction and fusion modeling; and in addition gives an overview of different artificial intelligence approaches to automatically extract data from images and electronic medical records for this purpose.

Thoracic Aortic Calcium Density and Area in Long-Term Atherosclerotic Cardiovascular Disease Risk Among Men Versus Women.

BACKGROUND: The development of thoracic aortic calcium (TAC) temporally precedes coronary artery calcium more often in women versus men. Whether TAC density and area confer sex-specific differences in atherosclerotic cardiovascular disease (ASCVD) risk is unknown. METHODS: We studied 5317 primary prevention patients who underwent coronary artery calcium scoring on noncontrast cardiac gated computed tomography with TAC >0. The Agatston TAC score (Agatston units), density (Hounsfield units), and area (mm2) were compared between men and women. Cox proportional hazards regression calculated adjusted hazard ratios for TAC density-area groups with ASCVD mortality, adjusting for traditional risk factors, coronary artery calcium, and TAC. Multinomial logistic regression calculated adjusted odds ratios for the association between traditional risk factors and TAC density-area groups. RESULTS: The mean age was 60.7 years, 38% were women, and 163 ASCVD deaths occurred over a median of 11.7-year follow-up. Women had higher median TAC scores (97 versus 84 Agatston units; P=0.004), density (223 versus 210 Hounsfield units; P<0.001), and area (37 versus 32 mm2; P=0.006) compared with men. There was a stepwise higher incidence of ASCVD deaths across increasing TAC density-area groups in men though women with low TAC density relative to TAC area (3.6 per 1000 person-years) had survival probability commensurate with the high-density-high-area group (4.8 per 1000 person-years). Compared with low TAC density-area, low TAC density/high TAC area conferred a 3.75-fold higher risk of ASCVD mortality in women (adjusted hazard ratio, 3.75 [95% CI, 1.13-12.44]) but not in men (adjusted hazard ratio, 1.16 [95% CI, 0.48-2.84]). Risk factors most strongly associated with low TAC density/high TAC area differed in women (diabetes: adjusted odds ratio, 2.61 [95% CI, 1.34-5.07]) versus men (hypertension: adjusted odds ratio, 1.45 [95% CI, 1.11-1.90]). CONCLUSIONS: TAC density-area phenotypes do not consistently associate with ASCVD mortality though low TAC density relative to area may be a marker of increased ASCVD risk in women.

Relationships of pericoronary and epicardial fat measurements in male and female patients with and without coronary artery disease.

INTRODUCTION: Although pericoronary adipose tissue (PCAT) is a component of the epicardial adipose tissue (EAT) depot, they may have different associations to coronary artery disease (CAD). We explored relationships between pericoronary adipose tissue mean attenuation (PCATMA) and EAT measurements in coronary CT angiography (CCTA) in patients with and without CAD. MATERIAL AND METHODS: CCTA scans of 185 non-CAD and 81 CAD patients (86.4% >50% stenosis) were included and retrospectively analyzed. PCATMA and EAT density/volume were measured and analyzed by sex, including associations with age, risk factors and tube voltage using linear regression models. RESULTS: In non-CAD and CAD, mean PCATMA and EAT volume were higher in men than in women (non-CAD: -92.5 ± 10.6HU vs -96.2 ± 8.4HU, and 174.4 ± 69.1 cm3 vs 124.1 ± 57.3 cm3; CAD: -92.2 ± 9.0HU vs -97.4 ± 9.7HU, and 193.6 ± 62.5 cm3 vs 148.5 ± 50.5 cm3 (p < 0.05)). EAT density was slightly lower in men than women in non-CAD (-96.4 ± 6.3HU vs -94.4 ± 5.5HU (p < 0.05)), and similar in CAD (-98.2 ± 5.2HU vs 98.2 ± 6.4HU). There was strong correlation between PCATMA and EAT density (non-CAD: r = 0.725, p < 0.001, CAD: r = 0.686, p < 0.001) but no correlation between PCATMA and EAT volume (non-CAD: r = 0.018, p = 0.81, CAD: r = -0.055, p = 0.63). A weak inverse association was found between EAT density and EAT volume (non-CAD: r = -0.244, p < 0.001, CAD: r = -0.263, p = 0.02). In linear regression models, EAT density was significantly associated with PCATMA in both non-CAD and CAD patients independent of risk factors and tube voltage. CONCLUSION: In CAD and non-CAD patients, EAT density, but not EAT volume, showed significant associations with PCATMA. Compared to women, men had higher PCATMA and EAT volume independently of disease status, but similar or slightly lower EAT density. Differences in trends and relations of PCATMA and EAT by sex could indicate that personalized interpretation and thresholding is needed.

Dual-Energy Computed Tomography in Cardiac Imaging.

Dual-energy computed tomography (DECT) acquires images using two energy spectra and offers a variation of reconstruction techniques for improved cardiac imaging. Virtual monoenergetic images decrease artifacts improving coronary plaque and stent visualization. Further, contrast attenuation is increased allowing significant reduction of contrast dose. Virtual non-contrast reconstructions enable coronary artery calcium scoring from contrast-enhanced scans. DECT provides advanced plaque imaging with detailed analysis of plaque components, indicating plaque stability. Extracellular volume assessment using DECT offers noninvasive detection of myocardial fibrosis. This review aims to outline the current cardiac applications of DECT, summarize recent literature, and discuss their findings.

Transient left bundle branch block associated with very high coronary artery calcium: a case report.

Coronary artery calcium (CAC) is the measure of subclinical coronary artery atherosclerosis most strongly associated with atherosclerotic cardiovascular disease (ASCVD) risk. However, CAC is rarely reported in the inpatient setting to guide chest pain management. We present a case of very high CAC in a 64-year-old woman with hypertension, type 2 diabetes, and hyperlipidemia presenting with dyspnea. Initial electrocardiogram (ECG) demonstrated normal conduction with a heart rate of 76 beats/min, but new T-wave inversions in V1-V4 and a high-sensitivity troponin-I (hsTnI) value of 6 ng/L (normal < 6 ng/L). Repeat ECG in the emergency department showed normal sinus rhythm (heart rate of 80 beats/min); however, it subsequently demonstrated a left bundle branch block (LBBB) with a repeat hsTnI of 7 ng/L. Stress testing with pharmacologic single-photon emission computerized tomography did not show scintigraphic evidence of ischemia but noted extensive CAC and a concern for balanced ischemia. Subsequent coronary computed tomography angiography (CCTA) showed nonobstructive disease and a total Agatston CAC score of 1262. Invasive evaluation with left heart catheterization was deferred given the patient's unchanged symptoms and CCTA findings. Statin therapy was intensified and aspirin, metoprolol succinate, and antihypertension therapies were continued. Initiation of glucose-lowering therapy and lipoprotein(a) testing was strongly recommended on follow-up. Our case suggests that CAC ⩾ 1000 may be incidentally associated with transient LBBB during the workup of coronary artery disease. Here, we specifically show that functional testing that incorporates measurement of CAC burden can help to improve ASCVD-preventive pharmacotherapy initiation and intensification beyond the identification of obstructive disease alone.

A Pilot Study of Neurobiological Mechanisms of Stress and Cardiovascular Risk.

Objective: Coronary heart disease is a leading cause of death and disability. Although psychological stress has been identified as an important potential contributor, mechanisms by which stress increases risk of heart disease and mortality are not fully understood. The purpose of this study was to assess mechanisms by which stress acts through the brain and heart to confer increased CHD risk. Methods: Coronary Heart Disease patients (N=10) underwent cardiac imaging with [Tc-99m] sestamibi single photon emission tomography at rest and during a public speaking mental stress task. Patients returned for a second day and underwent positron emission tomography imaging of the brain, heart, bone marrow, aorta (indicating inflammation) and subcutaneous adipose tissue, after injection of [18F]2-fluoro-2-deoxyglucose for assessment of glucose uptake followed mental stress. Patients with (N=4) and without (N=6) mental stress-induced myocardial ischemia were compared for glucose uptake in brain, heart, adipose tissue and aorta with mental stress. Results: Patients with mental stress-induced ischemia showed a pattern of increased uptake in the heart, medial prefrontal cortex, and adipose tissue with stress. In the heart disease group as a whole, activity increase with stress in the medial prefrontal brain and amygdala correlated with stress-induced increases in spleen (r=0.69, p=0.038; and r=0.69, p=0.04 respectfully). Stress-induced frontal lobe increased uptake correlated with stress-induced aorta uptake (r=0.71, p=0.016). Activity in insula and medial prefrontal cortex was correlated with post-stress activity in bone marrow and adipose tissue. Activity in other brain areas not implicated in stress did not show similar correlations. Increases in medial prefrontal activity with stress correlated with increased cardiac glucose uptake with stress, suggestive of myocardial ischemia (r=0.85, p=0.004). Conclusions: These findings suggest a link between brain response to stress in key areas mediating emotion and peripheral organs involved in inflammation and hematopoietic activity, as well as myocardial ischemia, in Coronary Heart Disease patients.

Artificial intelligence for advanced analysis of coronary plaque.

The field of coronary plaque analysis is advancing including more quantitative analysis of coronary artery diseases such as plaque burden, high-risk plaque features, computed tomography-derived fractional flow reserve, and radiomics. Although these biomarkers have shown great promise for the diagnosis and prognosis of cardiac patients in a research setting, many of these advanced analyses are labour and time intensive and therefore hard to implement in daily clinical practice. Artificial intelligence (AI) is playing an increasing role in supporting the quantification of these new biomarkers. AI offers the opportunity to increase efficiency, reduce human error and reader variability and to increase the accuracy of diagnosis and prognosis by automating many processing and supporting clinicians in their decision-making. With the use of AI these novel analysis approaches for coronary artery disease can be made feasible for clinical practice without increasing cost and workload and potentially improve patient care.

Quantitation of diffuse myocardial ischemia with mental stress and its association with cardiovascular events in individuals with recent myocardial infarction.

Microcirculatory dysfunction during psychological stress may lead to diffuse myocardial ischemia. We developed a novel quantification method for diffuse ischemia during mental stress (dMSI) and examined its relationship with outcomes after a myocardial infarction (MI). We studied 300 patients ≤ 61 years of age (50% women) with a recent MI. Patients underwent myocardial perfusion imaging with mental stress and were followed for 5 years. dMSI was quantified from cumulative count distributions of rest and stress perfusion. Focal ischemia was defined in a conventional fashion. The main outcome was a composite outcome of recurrent MI, heart failure hospitalizations, and cardiovascular death. A dMSI increment of 1 standard deviation was associated with a 40% higher risk for adverse events (HR 1.4, 95% CI 1.2-1.5). Results were similar after adjustment for viability, demographic and clinical factors and focal ischemia. In sex-specific analysis, higher levels of dMSI (per standard deviation increment) were associated with 53% higher risk of adverse events in women (HR 1.5, 95% CI 1.2-2.0) but not in men (HR 0.9, 95% CI 0.5-1.4), P 0.001. A novel index of diffuse ischemia with mental stress was associated with recurrent events in women but not in men after MI.

Evaluation of pericoronary adipose tissue attenuation on CT.

Pericoronary adipose tissue (PCAT) is the fat deposit surrounding coronary arteries. Although PCAT is part of the larger epicardial adipose tissue (EAT) depot, it has different pathophysiological features and roles in the atherosclerosis process. While EAT evaluation has been studied for years, PCAT evaluation is a relatively new concept. PCAT, especially the mean attenuation derived from CT images may be used to evaluate the inflammatory status of coronary arteries non-invasively. The most commonly used measure, PCATMA, is the mean attenuation of adipose tissue of 3 mm thickness around the proximal right coronary artery with a length of 40 mm. PCATMA can be analyzed on a per-lesion, per-vessel or per-patient basis. Apart from PCATMA, other measures for PCAT have been studied, such as thickness, and volume. Studies have shown associations between PCATMA and anatomical and functional severity of coronary artery disease. PCATMA is associated with plaque components and high-risk plaque features, and can discriminate patients with flow obstructing stenosis and myocardial infarction. Whether PCATMA has value on an individual patient basis remains to be determined. Furthermore, CT imaging settings, such as kV levels and clinical factors such as age and sex affect PCATMA measurements, which complicate implementation in clinical practice. For PCATMA to be widely implemented, a standardized methodology is needed. This review gives an overview of reported PCAT methodologies used in current literature and the potential use cases in clinical practice.

Evolving Role of Calcium Density in Coronary Artery Calcium Scoring and Atherosclerotic Cardiovascular Disease Risk.

Coronary artery calcium (CAC) is a specific marker of coronary atherosclerosis that can be used to measure calcified subclinical atherosclerotic burden. The Agatston method is the most widely used scoring algorithm for quantifying CAC and is expressed as the product of total calcium area and a quantized peak calcium density weighting factor defined by the calcification attenuation in HU on noncontrast computed tomography. Calcium density has emerged as an important area of inquiry because the Agatston score is upweighted based on the assumption that peak calcium density and atherosclerotic cardiovascular disease (ASCVD) risk are positively correlated. However, recent evidence demonstrates that calcium density is inversely associated with lesion vulnerability and ASCVD risk in population-based cohorts when accounting for age and plaque area. Here, we review calcium density by focusing on 3 main areas: 1) CAC scan acquisition parameters; 2) pathophysiology of calcified plaques; and 3) epidemiologic evidence relating calcium density to ASCVD outcomes. Through this process, we hope to provide further insight into the evolution of CAC scoring on noncontrast computed tomography.