Search

Machine learning (ML) models of risk prediction with coronary artery calcium (CAC) and CAC characteristics exhibit high performance, but are not inherently interpretable.To determine the direction and magnitude of impact of CAC characteristics on 10-year all-cause mortality (ACM) with explainable ML.We analyzed asymptomatic subjects in the CAC consortium. We trained ML models on 80% and tested on 20% of the data with XGBoost, using clinical characteristics ​+ ​CAC (ML 1) and additional CAC characteristics of CAC density and number of calcified vessels (ML 2). We applied SHAP, an explainable ML tool, to explore the relationship of CAC and CAC characteristics with 10-year all-cause and CV mortality.2376 deaths occurred among 63,215 patients [68% male, median age 54 (IQR 47-61), CAC 3 (IQR 0-94.3)]. ML2 was similar to ML1 to predict all-cause mortality (Area Under the Curve (AUC) 0.819 vs 0.821, p ​= ​0.23), but superior for CV mortality (0.847 vs 0.845, p ​= ​0.03). Low CAC density increased mortality impact, particularly ≤0.75. Very low CAC density ≤0.75 was present in only 4.3% of the patients with measurable density, and 75% occurred in CAC1-100. The number of diseased vessels did not increase mortality overall when simultaneously accounting for CAC and CAC density.CAC density contributes to mortality risk primarily when it is very low ≤0.75, which is primarily observed in CAC 1-100. CAC and CAC density are more important for mortality prediction than the number of diseased vessels, and improve prediction of CV but not all-cause mortality. Explainable ML techniques are useful to describe granular relationships in otherwise opaque prediction models.

Background: We examined age differences in whole-heart volumes of non-calcified and calcified atherosclerosis by coronary computed tomography angiography (CCTA) of patients with future ACS. Methods: A total of 234 patients with core-lab adjudicated ACS after baseline CCTA were enrolled. Atherosclerotic plaque was quantified and characterized from the main epicardial vessels and side branches on a 0.5 ​mm cross-sectional basis. Calcified plaque and non-calcified plaque were defined by above or below 350 Hounsfield units. Patients were categorized according to their age by deciles. Also, coronary artery calcium scores (CACS) were evaluated when available. Results: Patients were on average 62.2 ​± ​11.5 years old. On the pre-ACS CCTA, patients showed diffuse, multi-site, predominantly non-obstructive atherosclerosis across all age categories, with plaque being detected in 93.5% of all ACS cases. The proportion calcified plaque from the total plaque burden increased significantly with older presentation (10% calcification in those 80 years old). Patients with ACS 0 in 85% of the patients older than 50 years, and in 57% of patients younger than 50 years. Conclusion: The proportion of calcified plaque varied depending on patient age at the time of ACS. Only a small proportion of plaque was calcified when ACS occurred at

Coronary computed tomographic angiography (CCTA) provides a wealth of clinically meaningful information beyond anatomic stenosis alone, including the presence or absence of nonobstructive atherosclerosis and high-risk plaque features as precursors for incident coronary events. There is, however, no uniform agreement on how to identify and quantify these features or their use in evidence-based clinical decision-making. This statement from the Society of Cardiovascular Computed Tomography and North American Society of Cardiovascular Imaging addresses this gap and provides a comprehensive review of the available evidence on imaging of coronary atherosclerosis. In this statement, we provide standardized definitions for high-risk plaque (HRP) features and distill the evidence on the effectiveness of risk stratification into usable practice points. This statement outlines how this information should be communicated to referring physicians and patients by identifying critical elements to include in a structured CCTA report - the presence and severity of atherosclerotic plaque (descriptive statements, CAD-RADS (TM) categories), the segment involvement score, HRP features (e.g., low attenuation plaque, positive remodeling), and the coronary artery calcium score (when performed). Rigorous documentation of atherosclerosis on CCTA provides a vital opportunity to make recommendations for preventive care and to initiate and guide an effective care strategy for at-risk patients. (C) 2020 Published by Elsevier Inc. on behalf of Society of Cardiovascular Computed Tomography.

Aims Although there is increasing evidence supporting coronary atherosclerosis evaluation by coronary computed tomography angiography (CCTA), no data are available on age and sex differences for quantitative plaque features. The aim of this study was to investigate sex and age differences in both qualitative and quantitative atherosclerotic features from CCTA prior to acute coronary syndrome (ACS). Methods and results Within the ICONIC study, in which 234 patients with subsequent ACS were propensity matched 1:1 with 234 non-event controls, our current subanalysis included only the ACS cases. Both qualitative and quantitative advance plaque analysis by CCTA were performed by a core laboratory. In 129 cases, culprit lesions identified by invasive coronary angiography at the time of ACS were co-registered to baseline CCTA precursor lesions. The study population was then divided into subgroups according to sex and age ( Conclusion Our data underline the importance of age- and sex-related differences in coronary atherosclerosis presentation, which should be considered during CCTA-based atherosclerosis quantification.

Background Existing risk assessment tools for heart failure (HF) outcomes use structured databases with static, single-timepoint clinical data and have limited accuracy. Objective The purpose of this study was to develop a comprehensive approach for accurate prediction of 30-day unplanned readmission and all-cause mortality (ACM) that integrates clinical and physiological data available in the electronic health record system. Methods Three predictive models for 30-day unplanned readmissions or ACM were created using an extreme gradient boosting approach: (1) index admission model; (2) index discharge model; and (3) feature-aggregated model. Performance was assessed by the area under the curve (AUC) metric and compared with that of the HOSPITAL score, a widely used predictive model for hospital readmission. Results A total of 3774 patients with a primary billing diagnosis of HF were included (614 experienced the primary outcome), with 796 variables used in the admission and discharge models, and 2032 in the feature-aggregated model. The index admission model had AUC = 0.723, the index discharge model had AUC = 0.754, and the feature-aggregated model had AUC = 0.756 for prediction of 30-day unplanned readmission or ACM. For comparison, the HOSPITAL score had AUC = 0.666 (admission model: P = .093; discharge model: P = .022; feature aggregated: P = .012). Conclusion These models predict risk of HF hospitalizations and ACM in patients admitted with HF and emphasize the importance of incorporating large numbers of variables in machine learning models to identify predictors for future investigation.

Objectives This study designed and evaluated an end-to-end deep learning solution for cardiac segmentation and quantification. Background Segmentation of cardiac structures from coronary computed tomography angiography (CCTA) images is laborious. We designed an end-to-end deep-learning solution. Methods Scans were obtained from multicenter registries of 166 patients who underwent clinically indicated CCTA. Left ventricular volume (LVV) and right ventricular volume (RVV), left atrial volume (LAV) and right atrial volume (RAV), and left ventricular myocardial mass (LVM) were manually annotated as ground truth. A U-Net-inspired, deep-learning model was trained, validated, and tested in a 70:20:10 split. Results Mean age was 61.1 ± 8.4 years, and 49% were women. A combined overall median Dice score of 0.9246 (interquartile range: 0.8870 to 0.9475) was achieved. The median Dice scores for LVV, RVV, LAV, RAV, and LVM were 0.938 (interquartile range: 0.887 to 0.958), 0.927 (interquartile range: 0.916 to 0.946), 0.934 (interquartile range: 0.899 to 0.950), 0.915 (interquartile range: 0.890 to 0.920), and 0.920 (interquartile range: 0.811 to 0.944), respectively. Model prediction correlated and agreed well with manual annotation for LVV (r = 0.98), RVV (r = 0.97), LAV (r = 0.78), RAV (r = 0.97), and LVM (r = 0.94) (p Conclusions A deep-learning model rapidly segmented and quantified cardiac structures. This was done with high accuracy on a pixel level, with good agreement with manual annotation, facilitating its expansion into areas of research and clinical import.

This case-control cohort study analyzes patients with suspected coronary atherosclerosis and control patients to identify the factors associated with higher or lower risks for adverse cardiovascular events and acute coronary syndrome.Key PointsQuestionIs the density of coronary calcified plaque associated with future development of acute coronary syndrome? FindingsIn this case-control study of 189 patients who experienced vs 189 control individuals who did not experience an acute coronary syndrome after baseline coronary computed tomography angiography imaging, the volume of plaque with more than 1000 Hounsfield unit (termed 1K plaque) was associated with lower risk for acute coronary syndrome. The specific acute coronary syndrome precursor culprit lesion had less 1K plaque compared with the most stenotic lesion in control individuals. MeaningThis study's findings suggest that 1K plaque detected by coronary computed tomography angiography is associated with lower risk of future occurrence of acute coronary syndrome.ImportancePlaque morphologic measures on coronary computed tomography angiography (CCTA) have been associated with future acute coronary syndrome (ACS). However, the evolution of calcified coronary plaques by noninvasive imaging is not known. ObjectiveTo ascertain whether the increasing density in calcified coronary plaque is associated with risk for ACS. Design, Setting, and ParticipantsThis multicenter case-control cohort study included individuals enrolled in ICONIC (Incident Coronary Syndromes Identified by Computed Tomography), a nested case-control study of patients drawn from the CONFIRM (Coronary CT Angiography Evaluation for Clinical Outcomes: An International Multicenter) registry, which included 13 study sites in 8 countries. Patients who experienced core laboratory-verified ACS after baseline CCTA (n=189) and control individuals who did not experience ACS after baseline CCTA (n=189) were included. Patients and controls were matched 1:1 by propensity scores for age; male sex; presence of hypertension, hyperlipidemia, and diabetes; family history of premature coronary artery disease (CAD); current smoking status; and CAD severity. Data were analyzed from November 2018 to March 2019. ExposuresWhole-heart atherosclerotic plaque volume was quantitated from all coronary vessels and their branches. For patients who underwent invasive angiography at the time of ACS, culprit lesions were coregistered to baseline CCTA lesions by a blinded independent reader. Low-density plaque was defined as having less than 130 Hounsfield units (HU); calcified plaque, as having more than 350 HU and subcategorized on a voxel-level basis into 3 strata: 351 to 700 HU, 701 to 1000 HU, and more than 1000 HU (termed 1K plaque). Main Outcomes and MeasuresAssociation between calcium density and future ACS risk. ResultsA total of 189 patients and 189 matched controls (mean [SD] age of 59.9[9.8] years; 247 [65.3%] were male) were included in the analysis and were monitored during a mean (SD) follow-up period of 3.9 (2.5) years. The overall mean (SD) calcified plaque volume (>350 HU) was similar between patients and controls (76.4 [101.6] mm(3) vs 99.0 [156.1] mm(3); P=.32), but patients who experienced ACS exhibited less 1K plaque (>1000 HU) compared with controls (3.9 [8.3] mm(3) vs 9.4 [23.2] mm(3); P=.02). Individuals within the highest quartile of 1K plaque exhibited less low-density plaque, as a percentage of total plaque, when compared with patients within the lower 3 quartiles (12.6% [10.4%] vs 24.9% [20.6%]; P

Importance: Distinct plaque locations and vessel geometric features predispose to altered coronary flow hemodynamics. The association between these lesion-level characteristics assessed by coronary computed tomographic angiography (CCTA) and risk of future acute coronary syndrome (ACS) is unknown. Objective: To examine whether CCTA-derived adverse geometric characteristics (AGCs) of coronary lesions describing location and vessel geometry add to plaque morphology and burden for identifying culprit lesion precursors associated with future ACS. Design, Setting, and Participants: This substudy of ICONIC (Incident Coronary Syndromes Identified by Computed Tomography), a multicenter nested case-control cohort study, included patients with ACS and a culprit lesion precursor identified on baseline CCTA (n = 116) and propensity score-matched non-ACS controls (n = 116). Data were collected from July 20, 2012, to April 30, 2017, and analyzed from October 1, 2020, to October 31, 2021. Exposures: Coronary lesions were evaluated for the following 3 AGCs: (1) distance from the coronary ostium to lesion; (2) location at vessel bifurcations; and (3) vessel tortuosity, defined as the presence of 1 bend of greater than 90° or 3 curves of 45° to 90° using a 3-point angle within the lesion. Main Outcomes and Measures: Association between lesion-level AGCs and risk of future ACS-causing culprit lesions. Results: Of 548 lesions, 116 culprit lesion precursors were identified in 116 patients (80 [69.0%] men; mean [SD], age 62.7 [11.5] years). Compared with nonculprit lesions, culprit lesion precursors had a shorter distance from the ostium (median, 35.1 [IQR, 23.6-48.4] mm vs 44.5 [IQR, 28.2-70.8] mm), more frequently localized to bifurcations (85 [73.3%] vs 168 [38.9%]), and had more tortuous vessel segments (5 [4.3%] vs 6 [1.4%]; all P

Aims Right ventricular systolic dysfunction (RVSD) is an important determinant of outcomes in heart failure (HF) cohorts. While the quantitative assessment of RV function is challenging using 2D-echocardiography, cardiac magnetic resonance (CMR) is the gold standard with its high spatial resolution and precise anatomical definition. We sought to investigate the prognostic value of CMR-derived RV systolic function in a large cohort of HF with reduced ejection fraction (HFrEF). Methods and results Study cohort comprised of patients enrolled in the CarDiac MagnEtic Resonance for Primary Prevention Implantable CardioVerter DefibrillAtor ThErapy registry who had HFrEF and had simultaneous baseline CMR and echocardiography (n = 2449). RVSD was defined as RV ejection fraction (RVEF) Conclusion RV systolic dysfunction is an independent predictor of ACM in HFrEF, with a more pronounced prognostic value in select subgroups, likely reflecting the importance of RVSD in the early stages of HF progression.

Statistically and information-wise adequate data plays a critical role in training a robust deep learning model. However, collecting sufficient medical data to train a centralized model is still challenging due to various constraints such as privacy regulations and security. In this work, we develop a novel privacy-preserving federated-discriminator GAN, named FedD-GAN, that can learn and synthesize high-quality and various medical images regardless of their type, from heterogeneous datasets residing in multiple data centers whose data cannot be transferred or shared. We trained and evaluated FedD-GAN on three essential classes of medical data, each involving different types of medical images: cardiac CTA, brain MRI, and histopathology. We show that the synthesized images using our method have better quality than using a standard federated learning method and are realistic and accurate enough to train accurate segmentation models in downstream tasks. The segmentation model trained on the synthetic data only is comparable to that trained on an all-in-one real-image dataset shared from multiple data centers if possible. FedD-GAN can learn to generate a scalable and diverse synthetic database without compromising data privacy. This synthetic database could help to boost machine learning techniques in medical data analytics.

Patient-specific phenotyping of coronary atherosclerosis would facilitate personalized risk assessment and preventive treatment. We explored whether unsupervised cluster analysis can categorize patients with coronary atherosclerosis according to their plaque composition, and determined how these differing plaque composition profiles impact plaque progression. Patients with coronary atherosclerotic plaque (n = 947; median age, 62 years; 59% male) were enrolled from a prospective multi-national registry of consecutive patients who underwent serial coronary computed tomography angiography (median inter-scan duration, 3.3 years). K-means clustering applied to the percent volume of each plaque component and identified 4 clusters of patients with distinct plaque composition. Cluster 1 (n = 52), which comprised mainly fibro-fatty plaque with a significant necrotic core (median, 55.7% and 16.0% of the total plaque volume, respectively), showed the least total plaque volume (PV) progression (+ 23.3 mm3), with necrotic core and fibro-fatty PV regression (− 5.7 mm3 and − 5.6 mm3, respectively). Cluster 2 (n = 219), which contained largely fibro-fatty (39.2%) and fibrous plaque (46.8%), showed fibro-fatty PV regression (− 2.4 mm3). Cluster 3 (n = 376), which comprised mostly fibrous (62.7%) and calcified plaque (23.6%), showed increasingly prominent calcified PV progression (+ 21.4 mm3). Cluster 4 (n = 300), which comprised mostly calcified plaque (58.7%), demonstrated the greatest total PV increase (+ 50.7mm3), predominantly increasing in calcified PV (+ 35.9 mm3). Multivariable analysis showed higher risk for plaque progression in Clusters 3 and 4, and higher risk for adverse cardiac events in Clusters 2, 3, and 4 compared to that in Cluster 1. Unsupervised clustering algorithms may uniquely characterize patient phenotypes with varied atherosclerotic plaque profiles, yielding distinct patterns of progressive disease and outcome.

Objectives This study sought to compare the performance of history-based risk scores in predicting obstructive coronary artery disease (CAD) among patients with stable chest pain from the SCOT-HEART study. Background Risk scores for estimating pre-test probability of CAD are derived from referral-based populations with a high prevalence of disease. The generalizability of these scores to lower prevalence populations in the initial patient encounter for chest pain is uncertain. Methods We compared 3 scores among patients with suspected CAD in the coronary computed tomographic angiography (CTA) randomized arm of the SCOT-HEART study for the outcome of obstructive CAD by coronary CTA: the updated Diamond-Forrester score (UDF), CAD Consortium clinical score (CAD2), and CONFIRM risk score (CRS). We tested calibration with goodness-of-fit, discrimination with area under the receiver-operating curve (AUC), and reclassification with net reclassification improvement (NRI) to identify low-risk patients. Results In 1,738 patients (age 58 ± 10 years and 44.0% women), overall calibration was best for UDF, with underestimation by CRS and CAD2. Discrimination by AUC was highest for CAD2 at 0.79 (95% confidence interval [CI]: 0.77 to 0.81) than for UDF (0.77 [95% CI: 0.74 to 0.79]) or CRS (0.75 [95% CI: 0.73 to 0.77]) (p Conclusions In this multicenter clinic-based cohort of patients with suspected CAD and uniform CAD evaluation by coronary CTA, CAD2 provided the best discrimination and classification, despite overestimation of obstructive CAD as evaluated by coronary CTA. CRS exhibited intermediate performance followed by UDF for discrimination and reclassification.

Conventional scoring and identification methods for coronary artery calcium (CAC) and aortic calcium (AC) result in information loss from the original image and can be time-consuming. In this study, we sought to demonstrate an end-to-end deep learning model as an alternative to the conventional methods. Scans of 377 patients with no history of coronary artery disease (CAD) were obtained and annotated. A deep learning model was trained, tested and validated in a 60:20:20 split. Within the cohort, mean age was 64.2 ± 9.8 years, and 33% were female. Left anterior descending, right coronary artery, left circumflex, triple vessel, and aortic calcifications were present in 74.87%, 55.82%, 57.41%, 46.03%, and 85.41% of patients respectively. An overall Dice score of 0.952 (interquartile range 0.921, 0.981) was achieved. Stratified by subgroups, there was no difference between male (0.948, interquartile range 0.920, 0.981) and female (0.965, interquartile range 0.933, 0.980) patients (p = 0.350), or, between age p = 0.742). There was good correlation and agreement for CAC prediction (rho = 0.876, p < 0.001), with a mean difference of 11.2% (p = 0.100). AC correlated well (rho = 0.947, p < 0.001), with a mean difference of 9% (p = 0.070). Automated segmentation took approximately 4 s per patient. Taken together, the deep-end learning model was able to robustly identify vessel-specific CAC and AC with high accuracy, and predict Agatston scores that correlated well with manual annotation, facilitating application into areas of research and clinical importance.

Myocardial perfusion imaging with single-photon emission computed tomography and positron emission tomography has aided in diagnosing the severity of coronary artery disease. Artificial intelligence in cardiovascular nuclear medicine imaging includes various methods such as rule-based reasoning, artificial neural networks, structured machine learning, and deep convolutional neural networks. These artificial intelligence methods have been applied in detecting perfusion defects and other endpoints. Deep learning methods have been applied to nuclear medicine image formation to decrease computation time, improve image quality, and reduce radiation exposure to patients. This chapter summarizes AI methodologies applied to image analysis and image reconstruction for cardiovascular nuclear medicine imaging.

OBJECTIVES This study sought to identify culprit lesion (CL) precursors among acute coronary syndrome (ACS) patients based on qualitative and quantitative computed tomography-based plaque characteristics.BACKGROUND Coronary computed tomography angiography (CTA) has been validated for patient-level prediction of ACS. However, the applicability of coronary CTA to CL assessment is not known.METHODS Utilizing the ICONIC (Incident COroNary Syndromes Identified by Computed Tomography) study, a nested casecontrol study of 468 patients with baseline coronary CTA, the study included ACS patients with invasive coronary angiography-adjudicated CLs that could be aligned to CL precursors on baseline coronary CTA. Separate blinded core laboratories adjudicated CLs and performed atherosclerotic plaque evaluation. Thereafter, the study used a boosted ensemble algorithm (XGBoost) to develop a predictive model of CLs. Data were randomly split into a training set (80%) and a test set (20%). The area under the receiver-operating characteristic curve of thismodel was compared with that of diameter stenosis (model 1), high-risk plaque features (model 2), and lesion-level features of CL precursors from the ICONIC study (model 3). Thereafter, the machine learning (ML) model was applied to 234 non-ACS patients with 864 lesions to determine model performance for CL exclusion.RESULTS CL precursors were identified by both coronary angiography and baseline coronary CTA in 124 of 234 (53.0%) patients, with a total of 582 lesions (containing 124 CLs) included in the analysis. The ML model demonstrated significantly higher area under the receiver-operating characteristic curve for discriminating CL precursors (0.774; 95% confidence interval [CI]: 0.758 to 0.790) compared with model 1 (0.599; 95% CI: 0.599 to 0.599; p < 0.01), model 2 (0.532; 95% CI: 0.501 to 0.563; p < 0.01), and model 3 (0.672; 95% CI: 0.662 to 0.682; p < 0.01). When applied to the non-ACS cohort, the ML model had a specificity of 89.3% for excluding CLs.CONCLUSIONS In a high-risk cohort, a boosted ensemble algorithm can be used to predict CL from non-CL precursors on coronary CTA. (c) 2020 by the American College of Cardiology Foundation.

Author(s): Ferraro, Richard A; van Rosendael, Alexander R; Lu, Yao; Andreini, Daniele; Al-Mallah, Mouaz H; Cademartiri, Filippo; Chinnaiyan, Kavitha; Chow, Benjamin JW; Conte, Edoardo; Cury, Ricardo C; Feuchtner, Gudrun; de Araujo Goncalves, Pedro; Hadamitzky, Martin; Kim, Yong-Jin; Leipsic, Jonathon; Maffei, Erica; Marques, Hugo; Plank, Fabian; Pontone, Gianluca; Raff, Gilbert L; Villines, Todd C; Lee, Sang-Eun; Al'Aref, Subhi J; Baskaran, Lohendran; Cho, Iksung; Danad, Ibrahim; Gransar, Heidi; Budoff, Matthew J; Samady, Habib; Stone, Peter H; Virmani, Renu; Narula, Jagat; Berman, Daniel S; Chang, Hyuk-Jae; Bax, Jeroen J; Min, James K; Shaw, Leslee J; Lin, Fay Y | Abstract: AimsHigh-risk plaque (HRP) and non-obstructive coronary artery disease independently predict adverse events, but their importance to future culprit lesions has not been resolved. We sought to determine in patients prior to confirmed acute coronary syndrome (ACS) the association between lesion percent diameter stenosis (%DS), and the absolute number and prevalence of HRP. The secondary objective was to examine the relative importance of non-obstructive HRP in future culprit lesions.Methods and resultsWithin the ICONIC study, a nested case-control study of patients undergoing coronary computed tomographic angiography (coronary CT), we included ACS cases with culprit lesions confirmed by invasive coronary angiography and coregistered to baseline coronary CT. Quantitative CT was used to evaluate obstructive (≥50%) and non-obstructive (l50%) diameter stenosis, with HRP defined as ≥2 features of spotty calcification, positive remodelling, or low-attenuation plaque at baseline. A total of 234 patients with downstream ACS over 54 (interquartile range 5-525.5) days exhibited 198/898 plaques with HRP on coronary CT. While HRP was less prevalent in non-obstructive (19.7%, 161/819) than obstructive lesions (46.8%, 37/79, P l 0.001), non-obstructive plaque comprised 81.3% (161/198) of HRP lesions overall. Among the 128 patients with identifiable culprit lesion precursors, the adjusted hazard ratio (HR) was 1.85 [95% confidence interval (CI) 1.26-2.72] for HRP, with no interaction between %DS and HRP (P = 0.82). Compared to non-obstructive HRP lesions, obstructive lesions without HRP exhibited a non-significant HR of 1.41 (95% CI 0.61-3.25, P = 0.42).ConclusionsWhile HRP is more prevalent among obstructive lesions, non-obstructive HRP lesions outnumber those that are obstructive and confer risk clinically approaching that of obstructive lesions without HRP.

OBJECTIVES: To develop, demonstrate and evaluate an automated deep learning method for multiple cardiovascular structure segmentation. BACKGROUND: Segmentation of cardiovascular images is resource-intensive. We design an automated deep learning method for the segmentation of multiple structures from Coronary Computed Tomography Angiography (CCTA) images. METHODS: Images from a multicenter registry of patients that underwent clinically-indicated CCTA were used. The proximal ascending and descending aorta (PAA, DA), superior and inferior vena cavae (SVC, IVC), pulmonary artery (PA), coronary sinus (CS), right ventricular wall (RVW) and left atrial wall (LAW) were annotated as ground truth. The U-net-derived deep learning model was trained, validated and tested in a 70:20:10 split. RESULTS: The dataset comprised 206 patients, with 5.130 billion pixels. Mean age was 59.9 ± 9.4 yrs., and was 42.7% female. An overall median Dice score of 0.820 (0.782, 0.843) was achieved. Median Dice scores for PAA, DA, SVC, IVC, PA, CS, RVW and LAW were 0.969 (0.979, 0.988), 0.953 (0.955, 0.983), 0.937 (0.934, 0.965), 0.903 (0.897, 0.948), 0.775 (0.724, 0.925), 0.720 (0.642, 0.809), 0.685 (0.631, 0.761) and 0.625 (0.596, 0.749) respectively. Apart from the CS, there were no significant differences in performance between sexes or age groups. CONCLUSIONS: An automated deep learning model demonstrated segmentation of multiple cardiovascular structures from CCTA images with reasonable overall accuracy when evaluated on a pixel level.

Background Rapid coronary plaque progression (RPP) is associated with incident cardiovascular events. To date, no method exists for the identification of individuals at risk of RPP at a single point in time. This study integrated coronary computed tomography angiography–determined qualitative and quantitative plaque features within a machine learning (ML) framework to determine its performance for predicting RPP. Methods and Results Qualitative and quantitative coronary computed tomography angiography plaque characterization was performed in 1083 patients who underwent serial coronary computed tomography angiography from the PARADIGM (Progression of Atherosclerotic Plaque Determined by Computed Tomographic Angiography Imaging) registry. RPP was defined as an annual progression of percentage atheroma volume ≥1.0%. We employed the following ML models: model 1, clinical variables; model 2, model 1 plus qualitative plaque features; model 3, model 2 plus quantitative plaque features. ML models were compared with the atherosclerotic cardiovascular disease risk score, Duke coronary artery disease score, and a logistic regression statistical model. 224 patients (21%) were identified as RPP. Feature selection in ML identifies that quantitative computed tomography variables were higher‐ranking features, followed by qualitative computed tomography variables and clinical/laboratory variables. ML model 3 exhibited the highest discriminatory performance to identify individuals who would experience RPP when compared with atherosclerotic cardiovascular disease risk score, the other ML models, and the statistical model (area under the receiver operating characteristic curve in ML model 3, 0.83 [95% CI 0.78–0.89], versus atherosclerotic cardiovascular disease risk score, 0.60 [0.52–0.67]; Duke coronary artery disease score, 0.74 [0.68–0.79]; ML model 1, 0.62 [0.55–0.69]; ML model 2, 0.73 [0.67–0.80]; all P P =0.128). Conclusions Based on a ML framework, quantitative atherosclerosis characterization has been shown to be the most important feature when compared with clinical, laboratory, and qualitative measures in identifying patients at risk of RPP.

The purpose of this review is to summarize the work published by the Journal of Cardiovascular Computed Tomography (JCCT) for the year 2019, highlighting original research and new guidelines.

BackgroundHeart failure (HF) is a major cause of morbidity and mortality. However, much of the clinical data is unstructured in the form of radiology reports, while the process of data collection and curation is arduous and time-consuming.PurposeWe utilized a machine learning (ML)-based natural language processing (NLP) approach to extract clinical terms from unstructured radiology reports. Additionally, we investigate the prognostic value of the extracted data in predicting all-cause mortality (ACM) in HF patients.Materials and methodsThis observational cohort study utilized 122,025 thoracoabdominal computed tomography (CT) reports from 11,808 HF patients obtained between 2008 and 2018. 1,560 CT reports were manually annotated for the presence or absence of 14 radiographic findings, in addition to age and gender. Thereafter, a Convolutional Neural Network (CNN) was trained, validated and tested to determine the presence or absence of these features. Further, the ability of CNN to predict ACM was evaluated using Cox regression analysis on the extracted features.Results11,808 CT reports were analyzed from 11,808 patients (mean age 72.8 ± 14.8 years; 52.7% (6,217/11,808) male) from whom 3,107 died during the 10.6-year follow-up. The CNN demonstrated excellent accuracy for retrieval of the 14 radiographic findings with area-under-the-curve (AUC) ranging between 0.83-1.00 (F1 score 0.84-0.97). Cox model showed the time-dependent AUC for predicting ACM was 0.747 (95% confidence interval [CI] of 0.704-0.790) at 30 days.ConclusionAn ML-based NLP approach to unstructured CT reports demonstrates excellent accuracy for the extraction of predetermined radiographic findings, and provides prognostic value in HF patients.

Background and aims Studies evaluating the relationship between dense coronary calcium (DC) and myocardial ischemia have had incongruent results. We sought to clarify whether DC, as detected by computed coronary tomographic angiography (CCTA), is an independent predictor of ischemia as measured by invasive fractional flow reserve (FFR). Methods In total, 249 (399 lesions) stable patients undergoing CCTA and invasive FFR were enrolled for this post-hoc analysis. DC was defined as plaque with ≥350 HU using quantification software, and ischemia was defined as FFR ≤0.80. We evaluated the relationship of dense calcium volume (DCV), lesion plaque volume (LPV), non-calcified plaque volume (NCV), and area stenosis (AS) with ischemia using logistic regression reporting odds ratios (OR) with 95% confidence intervals (95% CI). Results Mean age was 63.0 ± 8.6 years, and 73 (29.3%) were female. Mean DCV was higher in lesions with FFR ≤0.80 (57.0 ± 54.7 mm 3 vs. 37.6 ± 49.5 mm 3 , [ p p p p = 0.69) was independently associated with ischemia. Conclusions Dense calcium is not an independent predictor of ischemia, but rather a marker of aggregate LPV, which in turn, is predictive of ischemia.

Hypertension is an atherosclerosis factor and is associated with cardiovascular risk. We investigated the relationship between hypertension and the presence, extent, and severity of coronary atherosclerosis in coronary computed tomographic angiography and cardiac events risk. Of 17 181 patients enrolled in the CONFIRM registry (Coronary CT Angiography Evaluation for Clinical Outcomes: An International Multicenter Registry) who underwent ≥64-detector row coronary computed tomographic angiography, we identified 14 803 patients without known coronary artery disease. Of these, 1434 hypertensive patients were matched to 1434 patients without hypertension. Major adverse cardiac events risk of hypertension and non-hypertensive patients was evaluated with Cox proportional hazards models. The prognostic associations between hypertension and no-hypertension with increasing degree of coronary stenosis severity (nonobstructive or obstructive ≥50%) and extent of coronary artery disease (segment involvement score of 1–5, >5) was also assessed. Hypertension patients less commonly had no coronary atherosclerosis and more commonly had nonobstructive and 1-, 2-, and 3-vessel disease than the no-hypertension group. During a mean follow-up of 5.2±1.2 years, 180 patients experienced cardiac events, with 104 (2.0%) occurring in the hypertension group and 76 (1.5%) occurring in the no-hypertension group (hazard ratios, 1.4; 95% confidence intervals, 1.0–1.9). Compared with no-hypertension patients without coronary atherosclerosis, hypertension patients with no coronary atherosclerosis and obstructive coronary disease tended to have higher risk of cardiac events. Similar trends were observed with respect to extent of coronary artery disease. Compared with no-hypertension patients, hypertensive patients have increased presence, extent, and severity of coronary atherosclerosis and tend to have an increase in major adverse cardiac events.