Coronary artery disease detection using artificial intelligence techniques: A survey of trends, geographical differences and diagnostic features 1991–2020

doi:10.1016/j.compbiomed.2020.104095

Computers in Biology and Medicine

Volume 128, January 2021, 104095

https://doi.org/10.1016/j.compbiomed.2020.104095 Get rights and content

Highlights

•
This research gives a comprehensive insight to the researchers of this field.
•
The researchers can find the algorithms or feature categories that have been less investigated.
•
It shows that in which countries this field was more interested for researchers.
•
The importance of features according to the published papers is also reported in this research.
•
It shows the importance of different features in various countries.

Abstract

While coronary angiography is the gold standard diagnostic tool for coronary artery disease (CAD), but it is associated with procedural risk, it is an invasive technique requiring arterial puncture, and it subjects the patient to radiation and iodinated contrast exposure. Artificial intelligence (AI) can provide a pretest probability of disease that can be used to triage patients for angiography. This review comprehensively investigates published papers in the domain of CAD detection using different AI techniques from 1991 to 2020, in order to discern broad trends and geographical differences. Moreover, key decision factors affecting CAD diagnosis are identified for different parts of the world by aggregating the results from different studies. In this study, all datasets that have been used for the studies for CAD detection, their properties, and achieved performances using various AI techniques, are presented, compared, and analyzed. In particular, the effectiveness of machine learning (ML) and deep learning (DL) techniques to diagnose and predict CAD are reviewed. From PubMed, Scopus, Ovid MEDLINE, and Google Scholar search, 500 papers were selected to be investigated. Among these selected papers, 256 papers met our criteria and hence were included in this study. Our findings demonstrate that AI-based techniques have been increasingly applied for the detection of CAD since 2008. AI-based techniques that utilized electrocardiography (ECG), demographic characteristics, symptoms, physical examination findings, and heart rate signals, reported high accuracy for the detection of CAD. In these papers, the authors ranked the features based on their assessed clinical importance with ML techniques. The results demonstrate that the attribution of the relative importance of ML features for CAD diagnosis is different among countries. More recently, DL methods have yielded high CAD detection performance using ECG signals, which drives its burgeoning adoption.

Introduction

The most common cardiovascular disease (CVD) is coronary artery disease (CAD), in which atherosclerosis causes narrowing of the artery lumen, resulting in diminished blood flow being supplied to the distal myocardium, i.e. ischemia. It is the leading cause of death worldwide and its prevalence has increased in low- and middle-income countries in recent years [1,2]. The reference standard for CAD diagnosis is invasive coronary angiography. The presence of a 50% or more diameter stenosis of the coronary artery lumen on angiography is generally defined as “significant” or “clinically important” stenosis, i.e. significant CAD [3]. While this anatomical threshold may not adequately characterize the functional state of ischemia of the myocardium supplied by the coronary artery, it is nonetheless widely accepted [4]. Invasive coronary angiography is associated with procedural risks, radiation, and potentially nephrotoxic iodinated contrast exposure. Therefore, accurate, reliable, noninvasive, and cost-effective methods for CAD diagnosis are highly sought by healthcare services payers in order to reduce the need for the number of negative invasive coronary angiography tests.

In the last three decades, artificial intelligence (AI) applications have been increasingly incorporated into clinical diagnostic tools to improve their accuracy [5,6]. In the same period, AI algorithms have also been widely applied to CVD datasets for data-driven decision-making [[7], [8], [9]]. AI-based systems can help to facilitate decision-making by automating and standardizing the interpretation, and inference processes, as well as improving the diagnostic accuracy [6]. The clinical utility of these advanced technologies depends on the skill of individuals who acquire, analyze, and interpret the data [10]. It is important to note that the type of raw data can intimately impact the quality and performance of AI methods. Therefore, to increase the quality of diagnosis, close collaboration between AI developers and clinical experts is essential. A brief overview of AI methods, advantages and disadvantages of CAD detection are presented in the supplementary information Fig. s2. Recently Alizadehsani et al. [11] published a review of the application of machine learning techniques for the automated detection of CAD using ECG signals. This current work is a review of secular trends from 1991 to 2020 as well as geographical differences in the use of AI techniques for CAD diagnosis, and focuses in particular on the dominant CAD features being employed in published works of AI methods. In our survey, the investigated features and the level of importance assigned to common features used for CAD prediction are diverse and vary significantly among countries. In addition, the performances of AI, which encompasses machine learning (ML) and deep learning (DL), are compared and discussed. It should be noted that in this research, we are describing trends. The reader can be referred to other publications for an in-depth discussion of ML and DL [12].

The value of the work lies in the extensive overview of the literature and the comprehensive survey of publication trends, geographical differences, and dissection into different features and their significance.

Section snippets

Material and methods

The present review aims to study the application of AI techniques for CAD detection, so as to identify issues that require attention to further advance the field. Accordingly, we performed a comprehensive review of the relevant AI literature. The strengths and weaknesses of previous studies are highlighted, thus serving as a base on which future studies can select as well as build upon AI techniques to improve CAD detection. The search criteria are discussed in the next sub-section.

We performed

Results

In this section, the statistical analysis of the papers published is reported. Important features and their categories are analyzed. Each category has a set of features extracted by a specific method. Some of the common categories are demographic characteristics, laboratory results, ECG, and echocardiographic findings.

Fig. 2a shows the number of articles published per year on CAD detection. The number has increased sharply from 2012 compared to the period from 1991 to 2011. Moreover, Fig. 2b

Discussion

In this research, our goal was the investigation of papers that used AI for CAD detection. More than 500 papers related to this goal are identified on initial screening. Among them, 256 papers are selected based on their relevance to the desired topic and quality of results. Our results indicate that most studies on CAD detection by ML algorithms were conducted in 2019 followed by 2017. According to Fig. 2a, it is obvious that most studies are conducted in these two years. More importantly, the

Conclusions

This paper reviewed publications on CAD detection using AI techniques over the last 30 years. Although AI-based methods cannot fully replace coronary angiography, they can be used as reliable, fast, and economical adjunctive screening tools for improving the diagnostic process. These tools can facilitate and expedite the referral of patients for invasive diagnostic coronary angiography while reducing the number of patients with no significant CAD undergoing expensive and risky invasive tests.

Contributors

RA, MR, MA, and FK had a continuation to prepare the first draft and NS, SN, AK, RP, AB, RB, MP, URA, RST, and SK contributed to edit the final draft. RA, RP, SK, MA, AK, AS, RST, and MP contributed to all analysis of the data and produced the outcomes accordingly. RA and MR found the papers and then extracted data. NS provided overall guidance and managed the project.

Patient consent for publication

Not required.

Funding

The authors have not declared a specific grant for this research from any funding agency.

Declaration of competing interest

The authors have no conflicts to disclose.

References (38)

R. Lozano et al.
Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of Disease Study 2010
Lancet
(2012)
K. Turk-Adawi et al.
Cardiovascular disease in the Eastern Mediterranean region: epidemiology and risk factor burden
Nat. Rev. Cardiol.
(2017)
P.W. Serruys et al.
Percutaneous coronary intervention versus coronary-artery bypass grafting for severe coronary artery disease, new England
J. Med.
(2009)
A. Abidov et al.
Integration of automatically measured transient ischemic dilation ratio into interpretation of adenosine stress myocardial perfusion SPECT for detection of severe and extensive CAD
J. Nucl. Med.
(2004)
L. Oakden-Rayner et al.
Precision Radiology: predicting longevity using feature engineering and deep learning methods in a radiomics framework
Sci. Rep.
(2017)
F.S. Collins et al.
A new initiative on precision medicine
N. Engl. J. Med.
(2015)
C. Krittanawong et al.
Artificial intelligence in precision cardiovascular medicine
J. Am. Coll. Cardiol.
(2017)
E. Butun et al.
One dimensional deep capsule networks for coronary artery disease detection using ECG signals
Phys. Med.
(2020)
U.R. Acharya et al.
Entropies for automated detection of coronary artery disease using ECG signals: a review
Biocybern. Biomed. Eng.
(2018)
S. Narula et al.
Machine-learning algorithms to automate morphological and functional assessments in 2D echocardiography
J. Am. Coll. Cardiol.
(2016)

R. Alizadehsani et al.

Machine learning-based coronary artery disease diagnosis: a comprehensive review

Comput. Biol. Med.

(2019)

S. Raschka et al.

Python Machine Learning

(2017)

E. Nasarian et al.

Association between work-related features and coronary artery disease: a heterogeneous hybrid feature selection integrated with balancing approach

Pattern Recogn. Lett.

(2020)

R. Alizadehsani et al.

Model uncertainty quantification for diagnosis of each main coronary artery stenosis

Soft Comput.

(2020)

R. Alizadehsani et al.

A database for using machine learning and data mining techniques for coronary artery disease diagnosis

Sci. Data

(2019)

R. Alizadehsani et al.

Hybrid genetic-discretized algorithm to handle data uncertainty in diagnosing stenosis of coronary arteries

Expet Syst.

(2020)

S.M.J. Jalali et al.

Parsimonious evolutionary-based model development for detecting artery disease

D. Dicker et al.

Global, regional, and national age-sex-specific mortality and life expectancy, 1950–2017: a systematic analysis for the global burden of disease study 2017

Lancet

(2018)

D.P. Zipes

Braunwald's Heart Disease: A Textbook of Cardiovascular Medicine

(2018)

Cited by (75)

Computational detection and interpretation of heart disease based on conditional variational auto-encoder and stacked ensemble-learning framework
2024, Biomedical Signal Processing and Control
Worldwide, cardiovascular disease is the leading cause of death. Based on clinical data, a Machine Learning (ML) system can detect cardiac disease in its early stages, which enables a reduction in mortality rates. However, imbalanced and high dimensionality data have been a persistent challenge in ML, impeding accurate predictive data analysis in many real-world applications, such as the detection of cardiovascular disease. To address this, computational methods targeting heart disease detection have been developed. However, their performance is still inadequate. Hence, this study presents a new stack predictor for the heart disease model (termed SPFHD). SPFHD employs five common tree-based ensemble learning algorithms as base models for heart disease detection. In addition, the predictions from the base models are integrated using a support vector machine algorithm to enhance the accuracy of heart disease detection. A new conditional variational autoencoder (CVAE) based method is developed to overcome the imbalance issue, which performs better than the conventional balancing methods. Finally, the SPFHD model is tuned by Bayesian optimization. The results show that the proposed SPFHD model outperforms the state-of-art methods over four datasets achieving higher f1-score of 4.68 %, 4.55 %, 2 %, and 1 % for HD clinical, Z-Alizadeh Sani, Statlog, and Cleveland, respectively. Moreover, this new framework offers vital interpretations which assist in understanding model success by leveraging the powerful SHapley Additive explanation (SHAP) algorithm. This highlights the most significant attributes for detecting heart disease and overcoming the limitations of current 'Black-box' methods that cannot reveal causal relationships between features.
A classification and regression tree algorithm for heart disease modeling and prediction
2023, Healthcare Analytics
Heart disease remains the leading cause of death, such that nearly one-third of all deaths worldwide are estimated to be caused by heart-related conditions. Advancing applications of classification-based machine learning to medicine facilitates earlier detection. In this study, the Classification and Regression Tree (CART) algorithm, a supervised machine learning method, has been employed to predict heart disease and extract decision rules in clarifying relationships between input and output variables. In addition, the study’s findings rank the features influencing heart disease based on importance. When considering all performance parameters, the 87% accuracy of the prediction validates the model’s reliability. On the other hand, extracted decision rules reported in the study can simplify the use of clinical purposes without needing additional knowledge. Overall, the proposed algorithm can support not only healthcare professionals but patients who are subjected to cost and time constraints in the diagnosis and treatment processes of heart disease.
The crosstalk between Nrf2 and NF-κB pathways in coronary artery disease: Can it be regulated by SIRT6?
2023, Life Sciences
Coronary artery disease (CAD) is the leading cause of death worldwide. Oxidative stress and inflammation are major mechanisms responsible for the progression of CAD. Nuclear transcription factor erythroid-2 related factor 2 (Nrf2) is a transcription factor that modulates the cellular redox status. Nrf2 upregulation increases the expression of antioxidant genes, decreases the expression of Nuclear factor-kappa B (NF-kB), and increases free radical metabolism. Activated NF-kB increases the production of inflammatory cytokines causing endothelial dysfunction. The two pathways of Nrf2 and NF-kB can regulate the expression of each other. Foremost, the Nrf2 pathway can decrease the level of active NF-κB by increasing the level of antioxidants and cytoprotective enzymes. Furthermore, the Nrf2 pathway prevents IκB-α degradation, an inhibitor of NF-kB, and thus inhibits NF-κB mediated transcription. Also, NF-kB transcription inhibits Nrf2 activation by reducing the antioxidant response element (ARE) transcription.
Sirtuin 6 (SIRT6) is a member of the Sirtuins family that was found to protect against cardiovascular diseases. SIRT6 can suppress the production of Reactive oxygen species (ROS) through deacetylation of NRF2 which results in NRF2 activation. Furthermore, SIRT6 can inhibit the inflammatory process through the downregulation of NF-kB transcription.
Therefore, targeting sirtuins could be a therapeutic strategy to treat CAD.
This review describes the potential role of SIRT6 in regulating the crosstalk between NRF2 and NF-kB signaling pathways in CAD.
Deep learning algorithm for predicting subacromial motion trajectory: Dynamic shoulder ultrasound analysis
2023, Ultrasonics
Subacromial motion metrics can be extracted from dynamic shoulder ultrasonography, which is useful for identifying abnormal motion patterns in painful shoulders. However, frame-by-frame manual labeling of anatomical landmarks in ultrasound images is time consuming. The present study aims to investigate the feasibility of a deep learning algorithm for extracting subacromial motion metrics from dynamic ultrasonography. Dynamic ultrasound imaging was retrieved by asking 17 participants to perform cyclic shoulder abduction and adduction along the scapular plane, whereby the trajectory of the humeral greater tubercle (in relation to the lateral acromion) was depicted by the deep learning algorithm. Extraction of the subacromial motion metrics was conducted using a convolutional neural network (CNN) or a self-transfer learning-based (STL)-CNN with or without an autoencoder (AE). The mean absolute error (MAE) compared with the manually-labeled data (ground truth) served as the main outcome variable. Using eight-fold cross-validation, the average MAE was proven to be significantly higher in the group using CNN than in those using STL-CNN or STL-CNN+AE for the relative difference between the greater tubercle and lateral acromion on the horizontal axis. The MAE for the localization of the two aforementioned landmarks on the vertical axis also seemed to be enlarged in those using CNN compared with those using STL-CNN. In the testing dataset, the errors in relation to the ground truth for the minimal vertical acromiohumeral distance were 0.081–0.333 cm using CNN, compared with 0.002–0.007 cm using STL-CNN. We successfully demonstrated the feasibility of a deep learning algorithm for automatic detection of the greater tubercle and lateral acromion during dynamic shoulder ultrasonography. Our framework also demonstrated the capability of capturing the minimal vertical acromiohumeral distance, which is the most important indicator of subacromial motion metrics in daily clinical practice.
A literature embedding model for cardiovascular disease prediction using risk factors, symptoms, and genotype information
2023, Expert Systems with Applications
Citation Excerpt :
Hence, the performance of feature selection differs by the data set chosen as reported in (Alizadehsani et al., 2019). The limitations of the previous cohort data-based CVD risk factor identification methods have been detailed (Alizadehsani et al., 2019, 2021). Previous methods can be categorized as: (1) cohort-based CVD risk factor identification (Ambale-Venkatesh et al., 2017; Knowles & Ashley, 2018; Nakanishi et al., 2018; Vrablik et al., 2021), and (2) literature-based CVD risk factor identification and information management (Alizadehsani et al., 2019, 2021).
Accurate prediction of cardiovascular disease (CVD) requires multifaceted information consisting of not only a patient’s medical history, but genomic data, symptoms, lifestyle, and risk factors which are often not incorporated into a decision making process as the data are vast, difficult to obtain, and require complex algorithms. However, given the vast amount of publicly available information and recent advances in machine and deep learning techniques, acquiring multifaceted data for accurate diagnosis and medical decision making via artificial intelligence has become more attainable. In this work we illustrate a literature embedding model, which identifies various risk factors, symptoms, mechanisms, and genes associated with CVD using a query word (i.e., “stroke”), and then based on the collected information the model predicts whether or not a person will be susceptible to CVD. For this purpose, we collected published literature from PubMed using search keywords consisting of a word such as “heart” and 19,264 human gene names, then trained our literature embedding model using the collected abstracts. For the intrinsic evaluation, we analyzed whether or not the captured words and genes were correctly identified as risk factors and associated symptoms for the input query words. For the extrinsic evaluation, we used our embedding model as feature selection and dimensionality reduction tasks on cohort data for CVD prediction. Our model accurately (average accuracy of >96 %) captured associated risk factors, symptoms, and genes for a given input query word (intrinsic evaluation). Using the selected features and reduced dimensions, our method provided better performance for CVD prediction with less computational time when compared with other popular methods (extrinsic evaluation). Our model provided outstanding results for both evaluation tasks, which enables accurate risk factor identification for CVD prediction. Our model has the potential to facilitate easier collation of multifaceted information for better data mining of vast publicly available data so that efficient and accurate risk factors and symptoms can be identified, which enables better-informed decisions for CVD prediction and treatment.
AttGRU-HMSI: enhancing heart disease diagnosis using hybrid deep learning approach
2024, Scientific Reports

View all citing articles on Scopus

View full text

Coronary artery disease detection using artificial intelligence techniques: A survey of trends, geographical differences and diagnostic features 1991–2020

Highlights

Abstract

Introduction

Section snippets

Material and methods

Results

Discussion

Conclusions

Contributors

Patient consent for publication

Funding

Declaration of competing interest

Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of Disease Study 2010

Lancet

Cardiovascular disease in the Eastern Mediterranean region: epidemiology and risk factor burden

Nat. Rev. Cardiol.

Percutaneous coronary intervention versus coronary-artery bypass grafting for severe coronary artery disease, new England

J. Med.

Integration of automatically measured transient ischemic dilation ratio into interpretation of adenosine stress myocardial perfusion SPECT for detection of severe and extensive CAD

J. Nucl. Med.

Precision Radiology: predicting longevity using feature engineering and deep learning methods in a radiomics framework

Sci. Rep.

A new initiative on precision medicine

N. Engl. J. Med.

Artificial intelligence in precision cardiovascular medicine

J. Am. Coll. Cardiol.

One dimensional deep capsule networks for coronary artery disease detection using ECG signals

Phys. Med.

Entropies for automated detection of coronary artery disease using ECG signals: a review

Biocybern. Biomed. Eng.

Machine-learning algorithms to automate morphological and functional assessments in 2D echocardiography

J. Am. Coll. Cardiol.

Machine learning-based coronary artery disease diagnosis: a comprehensive review

Comput. Biol. Med.

Python Machine Learning

Association between work-related features and coronary artery disease: a heterogeneous hybrid feature selection integrated with balancing approach

Pattern Recogn. Lett.

Model uncertainty quantification for diagnosis of each main coronary artery stenosis

Soft Comput.

A database for using machine learning and data mining techniques for coronary artery disease diagnosis

Sci. Data

Hybrid genetic-discretized algorithm to handle data uncertainty in diagnosing stenosis of coronary arteries

Expet Syst.

Parsimonious evolutionary-based model development for detecting artery disease

Global, regional, and national age-sex-specific mortality and life expectancy, 1950–2017: a systematic analysis for the global burden of disease study 2017

Lancet

Braunwald's Heart Disease: A Textbook of Cardiovascular Medicine