Non-invasive Coronary Artery Disease Screening Based on Electrocardiogram Characteristics and Clinical Risk Factors.

Coronary artery disease (CAD), an acute and life-threatening cardiovascular disease, is a leading cause of mortality and morbidity worldwide. Coronary angiography, the principal diagnostic tool for CAD, is invasive, expensive, and requires a lot of skilled effort. The current study aims to develop an automated and non-invasive CAD detection model and improve its performance as closely as possible to clinically acceptable diagnostic sensitivity. Electrocardiogram (ECG) characteristics are observed to be altered due to CAD and can be studied to develop a screening tool for its detection. The subject's clinical information can help broadly identify the high-cardiac-risk population and serve as a primary step in diagnosing CAD. This paper presents an approach to automatically detect CAD based on clinical data, morphological ECG features, and heart rate variability (HRV) features extracted from short-duration Lead-II ECG recordings. A few popular machine-learning classifiers, including support vector machine (SVM), random forest (RF), K-nearest neighbours (KNN), Gaussian Naïve Bayes (GNB), and multi-layer perceptron (MLP), are trained on the extracted feature space, and their performance is evaluated. Classifiers built by integrating clinical data and features extracted from ECG recordings demonstrated better performance than those built on each feature set separately, and the RF classifier outperforms other considered machine learners and reports an average testing accuracy of 94% and a G-mean score of 92% with a 5-fold cross-validation training accuracy of 95(± 0.04)%.Clinical relevance- The proposed method uses a brief, single-lead ECG recording and performs similarly to current clinical practices in an explainable manner. This makes it suitable for deployment via wearable technology (like smart watch gadgets) and telemonitoring, which may facilitate an earlier and more widespread CAD diagnosis.

MedTric : A clinically applicable metric for evaluation of multi-label computational diagnostic systems.

When judging the quality of a computational system for a pathological screening task, several factors seem to be important, like sensitivity, specificity, accuracy, etc. With machine learning based approaches showing promise in the multi-label paradigm, they are being widely adopted to diagnostics and digital therapeutics. Metrics are usually borrowed from machine learning literature, and the current consensus is to report results on a diverse set of metrics. It is infeasible to compare efficacy of computational systems which have been evaluated on different sets of metrics. From a diagnostic utility standpoint, the current metrics themselves are far from perfect, often biased by prevalence of negative samples or other statistical factors and importantly, they are designed to evaluate general purpose machine learning tasks. In this paper we outline the various parameters that are important in constructing a clinical metric aligned with diagnostic practice, and demonstrate their incompatibility with existing metrics. We propose a new metric, MedTric that takes into account several factors that are of clinical importance. MedTric is built from the ground up keeping in mind the unique context of computational diagnostics and the principle of risk minimization, penalizing missed diagnosis more harshly than over-diagnosis. MedTric is a unified metric for medical or pathological screening system evaluation. We compare this metric against other widely used metrics and demonstrate how our system outperforms them in key areas of medical relevance.

LTH-ECG: Lottery Ticket Hypothesis-based Deep Learning Model Compression for Atrial Fibrillation Detection from Single Lead ECG On Wearable and Implantable Devices.

Atrial Fibrillation (AF) is a kind of arrhythmia, which is a major morbidity factor, and AF can lead to stroke, heart failure and other cardiovascular complications. Electrocardiogram (ECG) is the basic marker to test the condition of heart and it can effectively detect AF condition. Single lead ECG has the practical advantage for being small form factor and it is easy to deploy. With the sophistication of the current deep learning (DL) models, researchers have been able to construct cardiologist-level models to detect different arrhythmias including AF condition detection from single lead short-time ECG signals. However, such models are computationally expensive and require huge memory size for deployment (more than 100 MB to deploy state-of-the-art 34-layer convolutional neural network-based ECG classification model). Such models need to be significantly trimmed with insignificantly loss of its classification performance for deployment in practical applications like single lead ECG classification in wearable and implantable devices. We have found that classical deep learning model compression techniques like pruning, quantization are not capable of substantial model size reduction without compromising on the model performance. In this paper, we propose LTH-ECG, which is our novel goal-driven winning lottery ticket discovery method, where lottery ticket hypothesis (LTH)-based iterative model pruning is used with the aim of over-pruning avoidance. LTH-ECG reduces the model size by 142x times with insignificant loss of classification performance (less than 1 % test F1-score penalty). Clinical Relevance- LTH-ECG will enable practical deployment for remote screening of AF condition using single lead short-time ECG recordings such that patients can on-demand monitor AF condition remotely through wearable ECG sensing devices and report cardiological abnormality to the concerned physician. LTH-ECG acts as an early warning system for effective AF condition screening.

Numerical Analysis on Effect of Coronary Supply-Demand Equilibrium on Varying Coronary Blockage and Stress Conditions.

In this paper, we present a computational fluid dynamic (CFD) analysis to capture the effect of physical stress and stenosis severity in coronary arteries leading to changes in coronary supply demand oxygen equilibrium. We propose a coupled Od-3d coronary vessel model to predict the variation in flow dynamics of coronary as well as arterial system, modeled using an in-silico model replicating cardiovascular hemodynamics. CFD simulation were solved using subject specific CT scan for coronary and arterial flow and pressure along with metrics related to arterial wall shear stress. Simulations were performed for three heart rates (75, 90 and 120 bpm) and four stenosis states representing different stages of Coronary artery disease (CAD) namely healthy, 50%, 75%, 90% blockage in left anterior descending artery (LAD). Myocardial oxygen supply demand equilibrium were calculated for each cases using hemodynamic surrogate markers naming Diastolic pressure time index for supply and Tension time index for demand. The proposed 0d-3d coupled hemodynamic model of the coronary vessel bed along with supply-demand equilibrium estimated for different stress level and stenosis severity may provide useful insights on the dynamics of CAD manifestation and predict vulnerable regions in coronary bed for early screening and interventions.

Simulating Stenotic Conditions of the Coronary Artery in a Lumped Parameter Model of the Cardiovascular System.

Coronary flow control mechanisms maintain the average coronary blood flow (CBF) at 4% of the cardiac output (CO) in normal adults, with no prior diagnosis of coronary artery disease (CAD), under resting conditions. This paper explores a pulsatile sixth order lumped parameter (LP) model of the cardiovascular system (CVS) which utilizes the average CBF approximated from CO along with arterial blood pressure (ABP) waveform to estimate the coronary microvascular resistance using non-linear least square optimization technique. The CVS model includes a third order model of the coronary vascular bed and is shown to achieve phasic coronary flow. The coronary epicardial resistance is varied to emulate different degrees of stenosis and achieve realistic behavior of coronary microvascular resistance under these conditions.

Resource Constrained CVD Classification Using Single Lead ECG On Wearable and Implantable Devices.

Electrocardiogram (ECG) is one of the fundamental markers to detect different cardiovascular diseases (CVDs). Owing to the widespread availability of ECG sensors (single lead) as well as smartwatches with ECG recording capability, ECG classification using wearable devices to detect different CVDs has become a basic requirement for a smart healthcare ecosystem. In this paper, we propose a novel method of model compression with robust detection capability for CVDs from ECG signals such that the sophisticated and effective baseline deep neural network model can be optimized for the resource constrained micro-controller platform suitable for wearable devices while minimizing the performance loss. We employ knowledge distillation-based model compression approach where the baseline (teacher) deep neural network model is compressed to a TinyML (student) model using piecewise linear approximation. Our proposed ECG TinyML has achieved ~156x compression factor to suit to the requirement of 100KB memory availability for model deployment on wearable devices. The proposed model requires ~5782 times (estimated) less computational load than state-of-the-art residual neural network (ResNet) model with negligible performance loss (less than 1% loss in test accuracy, test sensitivity, test precision and test F1-score). We further feel that the small footprint model size of ECG TinyML (62.3 KB) can be suitably deployed in implantable devices including implantable loop recorder (ILR).

3D Cardiac Computational Model for Evaluating the Progression of Myocardial Ischemia in a Supply-Demand Paradigm.

In this paper, we present a cardiac computational framework aimed at simulating the effects of ischemia on cardiac potentials and hemodynamics. Proposed cardiac model uses an image based pipeline for modeling and analysis of the ischemic condition in-silico. We compute epicardial potential as well as body surface potential (BSP) for acute ischemic conditions based on data from animal model while varying both local coronary supply and global metabolic demand. Single lead ECG equivalent signal processed from computed BSP is used to drive a lumped hemodynamic model and derive left ventricular dynamics. Computational framework combining 3d structural information from image data and integrating electrophysiology and hemodynamics functionality is aimed to evaluate additional cardiac markers along with conventional electrical markers visible during acute ischemia and give a broader understanding of ischemic manifestation leading to pathophysiological changes. Simulation of epicardial to bodysurface potential followed by estimation of hemodynamic parameters like ejection fraction, contractility, blood pressure, etc, would help to infer subtle changes detectable beyond conventional ST segment changes.

A Lumped Parameter Model for Cardiac Output Estimation Using Arterial Blood Pressure Waveform.

This paper investigates a subject-specific lumped parameter cardiovascular model for estimating Cardiac Output (CO) using the radial Arterial Blood Pressure (ABP) waveform. The model integrates a simplified model of the left ventricle along with a linear third order model of the arterial tree and generates reasonably accurate ABP waveforms along with the Dicrotic Notch (DN). Non-linear least square optimization technique is used to obtain uncalibrated estimates of cardiovascular parameters. Thermodilution CO measurements have been used to evaluate the CO estimation accuracy. The model achieves less than 15% normalized error across 10 subjects with different shapes of ABP waveform.

Computational Model for Therapy Optimization of Wearable Cardioverter Defibrillator: Shockable Rhythm Detection and Optimal Electrotherapy.

Wearable cardioverter defibrillator (WCD) is a life saving, wearable, noninvasive therapeutic device that prevents fatal ventricular arrhythmic propagation that leads to sudden cardiac death (SCD). WCD are frequently prescribed to patients deemed to be at high arrhythmic risk but the underlying pathology is potentially reversible or to those who are awaiting an implantable cardioverter-defibrillator. WCD is programmed to detect appropriate arrhythmic events and generate high energy shock capable of depolarizing the myocardium and thus re-initiating the sinus rhythm. WCD guidelines dictate very high reliability and accuracy to deliver timely and optimal therapy. Computational model-based process validation can verify device performance and benchmark the device setting to suit personalized requirements. In this article, we present a computational pipeline for WCD validation, both in terms of shock classification and shock optimization. For classification, we propose a convolutional neural network-"Long Short Term Memory network (LSTM) full form" (Convolutional neural network- Long short term memory network (CNN-LSTM)) based deep neural architecture for classifying shockable rhythms like Ventricular Fibrillation (VF), Ventricular Tachycardia (VT) vs. other kinds of non-shockable rhythms. The proposed architecture has been evaluated on two open access ECG databases and the classification accuracy achieved is in adherence to American Heart Association standards for WCD. The computational model developed to study optimal electrotherapy response is an in-silico cardiac model integrating cardiac hemodynamics functionality and a 3D volume conductor model encompassing biophysical simulation to compute the effect of shock voltage on myocardial potential distribution. Defibrillation efficacy is simulated for different shocking electrode configurations to assess the best defibrillator outcome with minimal myocardial damage. While the biophysical simulation provides the field distribution through Finite Element Modeling during defibrillation, the hemodynamic module captures the changes in left ventricle functionality during an arrhythmic event. The developed computational model, apart from acting as a device validation test-bed, can also be used for the design and development of personalized WCD vests depending on subject-specific anatomy and pathology.

Multimodal cardiovascular model for hemodynamic analysis: Simulation study on mitral valve disorders.

Valvular heart diseases are a prevalent cause of cardiovascular morbidity and mortality worldwide, affecting a wide spectrum of the population. In-silico modeling of the cardiovascular system has recently gained recognition as a useful tool in cardiovascular research and clinical applications. Here, we present an in-silico cardiac computational model to analyze the effect and severity of valvular disease on general hemodynamic parameters. We propose a multimodal and multiscale cardiovascular model to simulate and understand the progression of valvular disease associated with the mitral valve. The developed model integrates cardiac electrophysiology with hemodynamic modeling, thus giving a broader and holistic understanding of the effect of disease progression on various parameters like ejection fraction, cardiac output, blood pressure, etc., to assess the severity of mitral valve disorders, naming Mitral Stenosis and Mitral Regurgitation. The model mimics an adult cardiovascular system, comprising a four-chambered heart with systemic, pulmonic circulation. The simulation of the model output comprises regulated pressure, volume, and flow for each heart chamber, valve dynamics, and Photoplethysmogram signal for normal physiological as well as pathological conditions due to mitral valve disorders. The generated physiological parameters are in agreement with published data. Additionally, we have related the simulated left atrium and ventricle dimensions, with the enlargement and hypertrophy in the cardiac chambers of patients with mitral valve disorders, using their Electrocardiogram available in Physionet PTBI dataset. The model also helps to create 'what if' scenarios and relevant analysis to study the effect in different hemodynamic parameters for stress or exercise like conditions.

Detection of Counterfeit Medicines Using Hyperspectral Sensing.

Worldwide revenue of pharmaceutical market is more than 1200 billion USD [1] and that of counterfeit medicines is around 200 billion USD [2][3]. Counterfeit medicines can be detected by technical experts using visual inspection or through sophisticated lab and relevant methods. However, such methods require time, sample preparation and technical expertise with lab setup. These methods are not feasible and scalable to be used in the field by the general public. The objective of our research work was to detect counterfeit medicines using simpler and faster method using hyperspectral sensing. In this experiment, a visible - near infrared (350nm - 1050nm) hyperspectral device was used to capture spectral signature of the medicines. We used 24 medicine tablets of different companies. To imitate counterfeit medicines, tablet powders were adulterated by adding different levels of calcium carbonate. Spectral signatures were captured from original stage to all stages of adulterations and analyzed using machine learning (multilayer perceptron classifier). Result shows that we are able to achieve more than 90% classification accuracy. Portable hyperspectral sensing combined with medicines spectral database can be a good field level test method for detection of counterfeit medicines, as it is very fast, easy to use and does not require technical expertise.

Detection of atrial fibrillation and other abnormal rhythms from ECG using a multi-layer classifier architecture.

OBJECTIVE: Atrial fibrillation (AF) and other types of abnormal heart rhythm are related to multiple fatal cardiovascular diseases that affect the quality of human life. Hence the development of an automated robust method that can reliably detect AF, in addition to other non-sinus and sinus rhythms, would be a valuable addition to medicine. The present study focuses on developing an algorithm for the classification of short, single-lead electrocardiogram (ECG) recordings into normal, AF, other abnormal rhythms and noisy classes. APPROACH: The proposed classification framework presents a two-layer, three-node architecture comprising binary classifiers. PQRST markers are detected on each ECG recording, followed by noise removal using a spectrogram power based novel adaptive thresholding scheme. Next, a feature pool comprising time, frequency, morphological and statistical domain ECG features is extracted for the classification task. At each node of the classification framework, suitable feature subsets, identified through feature ranking and dimension reduction, are selected for use. Adaptive boosting is selected as the classifier for the present case. The training data comprises 8528 ECG recordings provided under the PhysioNet 2017 Challenge. F1 scores averaged across the three non-noisy classes are taken as the performance metric. MAIN RESULT: The final five-fold cross-validation score achieved by the proposed framework on the training data has high accuracy with low variance (0.8254 [Formula: see text] 0.0043). SIGNIFICANCE: Further, the proposed algorithm has achieved joint first place in the PhysioNet/Computing in Cardiology Challenge 2017 with a score of 0.83 computed on a hidden test dataset.