A unified framework for multi-lead ECG characterization using Laplacian Eigenmaps.

Background.The analysis of multi-lead electrocardiographic (ECG) signals requires integrating the information derived from each lead to reach clinically relevant conclusions. This analysis could benefit from data-driven methods compacting the information in those leads into lower-dimensional representations (i.e. 2 or 3 dimensions instead of 12).Objective.We propose Laplacian Eigenmaps (LE) to create a unified framework where ECGs from different subjects can be compared and their abnormalities are enhanced.Approach.We conceive a normal reference ECG space based on LE, calculated using signals of healthy subjects in sinus rhythm. Signals from new subjects can be mapped onto this reference space creating a loop per heartbeat that captures ECG abnormalities. A set of parameters, based on distance metrics and on the shape of loops, are proposed to quantify the differences between subjects.Main results.This methodology was applied to find structural and arrhythmogenic changes in the ECG. The LE framework consistently captured the characteristics of healthy ECGs, confirming that normal signals behaved similarly in the LE space. Significant differences between normal signals, and those from patients with ischemic heart disease or dilated cardiomyopathy were detected. In contrast, LE biomarkers did not identify differences between patients with cardiomyopathy and a history of ventricular arrhythmia and their matched controls.Significance.This LE unified framework offers a new representation of multi-lead signals, reducing dimensionality while enhancing imperceptible abnormalities and enabling the comparison of signals of different subjects.

Artificial intelligence for non-invasive glycaemic-events detection via ECG in a paediatric population: study protocol.

Purpose: Paediatric Type 1 Diabetes (T1D) patients are at greater risk for developing severe hypo and hyperglycaemic events due to poor glycaemic control. To reduce the risk of adverse events, patients need to achieve the best possible glycaemic management through frequent blood glucose monitoring with finger prick or Continuous Glucose Monitoring (CGM) systems. However, several non-invasive techniques have been proposed aiming at exploiting changes in physiological parameters based on glucose levels. The overall objective of this study is to validate an artificial intelligence (AI) based algorithm to detect glycaemic events using ECG signals collected through non-invasive device. Methods: This study will enrol T1D paediatric participants who already use CGM. Participants will wear an additional non-invasive wearable device for recording physiological data and respiratory rate. Glycaemic measurements driven through ECG variables are the main outcomes. Data collected will be used to design, develop and validate the personalised and generalized classifiers based on a deep learning (DL) AI algorithm, able to automatically detect hypoglycaemic events by using few ECG heartbeats recorded with wearable devices. Results: Data collection is expected to be completed approximately by June 2023. It is expected that sufficient data will be collected to develop and validate the AI algorithm. Conclusion: This is a validation study that will perform additional tests on a larger diabetes sample population to validate the previous pilot results that were based on four healthy adults, providing evidence on the reliability of the AI algorithm in detecting glycaemic events in paediatric diabetic patients in free-living conditions. Trial registration: ClinicalTrials.gov identifier: NCT03936634. Registered on 11 March 2022, retrospectively registered, https://www.clinicaltrials.gov/ct2/show/NCT05278143?titles=AI+for+Glycemic+Events+Detection+Via+ECG+in+a+Pediatric+Population&draw=2&rank=1. Supplementary information: The online version contains supplementary material available at 10.1007/s12553-022-00719-x.

Multilevel hybrid accurate handcrafted model for myocardial infarction classification using ECG signals.

Myocardial infarction (MI) is detected using electrocardiography (ECG) signals. Machine learning (ML) models have been used for automated MI detection on ECG signals. Deep learning models generally yield high classification performance but are computationally intensive. We have developed a novel multilevel hybrid feature extraction-based classification model with low time complexity for MI classification. The study dataset comprising 12-lead ECGs belonging to one healthy and 10 MI classes were downloaded from a public ECG signal databank. The model architecture comprised multilevel hybrid feature extraction, iterative feature selection, classification, and iterative majority voting (IMV). In the hybrid handcrafted feature (HHF) generation phase, both textural and statistical feature extraction functions were used to extract features from ECG beats but only at a low level. A new pooling-based multilevel decomposition model was presented to enable them to create features at a high level. This model used average and maximum pooling to create decomposed signals. Using these pooling functions, an unbalanced tree was obtained. Therefore, this model was named multilevel unbalanced pooling tree transformation (MUPTT). On the feature extraction side, two extractors (functions) were used to generate both statistical and textural features. To generate statistical features, 20 commonly used moments were used. A new, improved symmetric binary pattern function was proposed to generate textural features. Both feature extractors were applied to the original MI signal and the decomposed signals generated by the MUPTT. The most valuable features from among the extracted feature vectors were selected using iterative neighborhood component analysis (INCA). In the classification phase, a one-dimensional nearest neighbor classifier with ten-fold cross-validation was used to obtain lead-wise results. The computed lead-wise results derived from all 12 leads of the same beat were input to the IMV algorithm to generate ten voted results. The most representative was chosen using a greedy technique to calculate the overall classification performance of the model. The HHF-MUPTT-based ECG beat classification model attained excellent performance, with the best lead-wise accuracy of 99.85% observed in Lead III and 99.94% classification accuracy using the IMV algorithm. The results confirmed the high MI classification ability of the presented computationally lightweight HHF-MUPTT-based model.

A comparative study of the performance of methods for f-wave extraction.

Objective.This study proposes a novel technique for atrial fibrillatory waves (f-waves) extraction and investigates the performance of the proposed method comparing with different f-wave extraction methods.Approach.We propose a novel technique combining a periodic component analysis (PiCA) and echo state network (ESN) for f-waves extraction, denoted PiCA-ESN. PiCA-ESN benefits from the advantages of using both source separation and nonlinear adaptive filtering. PiCA-ESN is evaluated by comparing with other state-of-the-art approaches, which include template subtraction technique based on principal component analysis, spatiotemporal cancellation, nonlinear adaptive filtering using an echo state neural network, and a source separation technique based on PiCA. Quality assessment is performed on a recently published reference database including a large number of simulated ECG signals in atrial fibrillation (AF). The performance of the f-wave extraction methods is evaluated in terms of signal quality metrics (SNR, ΔSNR) and robustness of f-wave features.Main results.The proposed method offers the best signal quality performance, with a ΔSNR of approximately 22 dB across all 8 sets of the reference database, as well as the most robust extraction of f-wave features, with 75% of all estimates of dominant atrial frequency well below 1 Hz.

Novel DERMA Fusion Technique for ECG Heartbeat Classification.

An electrocardiogram (ECG) consists of five types of different waveforms or characteristics (P, QRS, and T) that represent electrical activity within the heart. Identification of time intervals and morphological appearance of the waves are the major measuring instruments to detect cardiac abnormality from ECG signals. The focus of this study is to classify five different types of heartbeats, including premature ventricular contraction (PVC), left bundle branch block (LBBB), right bundle branch block (RBBB), PACE, and atrial premature contraction (APC), to identify the exact condition of the heart. Prior to the classification, extensive experiments on feature extraction were performed to identify the specific events from ECG signals, such as P, QRS complex, and T waves. This study proposed the fusion technique, dual event-related moving average (DERMA) with the fractional Fourier-transform algorithm (FrlFT) to identify the abnormal and normal morphological events of the ECG signals. The purpose of the DERMA fusion technique is to analyze certain areas of interest in ECG peaks to identify the desired location, whereas FrlFT analyzes the ECG waveform using a time-frequency plane. Furthermore, detected highest and lowest components of the ECG signal such as peaks, the time interval between the peaks, and other necessary parameters were utilized to develop an automatic model. In the last stage of the experiment, two supervised learning models, namely support vector machine and K-nearest neighbor, were trained to classify the cardiac condition from ECG signals. Moreover, two types of datasets were used in this experiment, specifically MIT-BIH Arrhythmia with 48 subjects and the newly disclosed Shaoxing and Ningbo People's Hospital (SPNH) database, which contains over 10,000 patients. The performance of the experimental setup produced overwhelming results, which show around 99.99% accuracy, 99.96% sensitivity, and 99.9% specificity.

An accurate and efficient method to train classifiers for atrial fibrillation detection in ECGs: Learning by asking better questions.

BACKGROUND: An increasing number of wearables are capable of measuring electrocardiograms (ECGs), which may help in early detection of atrial fibrillation (AF). Therefore, many studies focus on automated detection of AF in ECGs. A major obstacle is the required amount of manually labelled data. This study aimed to provide an efficient and reliable method to train a classifier for AF detection using large datasets of real-life ECGs. METHOD: Human-controlled semi-supervised learning was applied, consisting of two phases: the pre-training phase and the semi-automated training phase. During pre-training, an initial classifier was trained, which was used to predict the classes of new ECG segments in the semi-automated training phase. Based on the degree of certainty, segments were added to the training dataset automatically or after human validation. Thereafter, the classifier was retrained and this procedure was repeated. To test the model performance, a real-life telemetry dataset containing 3,846,564 30-s ECG segments of hospitalized patients (n = 476) and the CinC Challenge 2017 database were used. RESULTS: After pre-training, the average F1-score on a hidden testing dataset was 89.0%. Furthermore, after the pre-training phase 68.0% of all segments in the hidden test set could be classified with an estimated probability of successful classification of 99%, providing an F1-score of 97.9% for these segments. During the semi-automated training phase, this F1-score showed little variation (97.3%-97.9% in the hidden test set), whilst the number of segments which could be automatically classified increased from 68.0% to 75.8% due to the enhanced training dataset. At the same time, the overall F1-score increased from 89.0% to 91.4%. CONCLUSIONS: Human-validated semi-supervised learning makes training a classifier more time efficient without compromising on accuracy, hence this method might be valuable in the automated detection of AF in real-life ECGs.

Study of the Few-Shot Learning for ECG Classification Based on the PTB-XL Dataset.

The electrocardiogram (ECG) is considered a fundamental of cardiology. The ECG consists of P, QRS, and T waves. Information provided from the signal based on the intervals and amplitudes of these waves is associated with various heart diseases. The first step in isolating the features of an ECG begins with the accurate detection of the R-peaks in the QRS complex. The database was based on the PTB-XL database, and the signals from Lead I-XII were analyzed. This research focuses on determining the Few-Shot Learning (FSL) applicability for ECG signal proximity-based classification. The study was conducted by training Deep Convolutional Neural Networks to recognize 2, 5, and 20 different heart disease classes. The results of the FSL network were compared with the evaluation score of the neural network performing softmax-based classification. The neural network proposed for this task interprets a set of QRS complexes extracted from ECG signals. The FSL network proved to have higher accuracy in classifying healthy/sick patients ranging from 93.2% to 89.2% than the softmax-based classification network, which achieved 90.5-89.2% accuracy. The proposed network also achieved better results in classifying five different disease classes than softmax-based counterparts with an accuracy of 80.2-77.9% as opposed to 77.1% to 75.1%. In addition, the method of R-peaks labeling and QRS complexes extraction has been implemented. This procedure converts a 12-lead signal into a set of R waves by using the detection algorithms and the k-mean algorithm.

VPNET: Variable Projection Networks.

In this paper, we introduce VPNet, a novel model-driven neural network architecture based on variable projection (VP). Applying VP operators to neural networks results in learnable features, interpretable parameters, and compact network structures. This paper discusses the motivation and mathematical background of VPNet and presents experiments. The VPNet approach was evaluated in the context of signal processing, where we classified a synthetic dataset and real electrocardiogram (ECG) signals. Compared to fully connected and one-dimensional convolutional networks, VPNet offers fast learning ability and good accuracy at a low computational cost of both training and inference. Based on these advantages and the promising results obtained, we anticipate a profound impact on the broader field of signal processing, in particular on classification, regression and clustering problems.

Digital biomarkers and algorithms for detection of atrial fibrillation using surface electrocardiograms: A systematic review.

AIMS: Automated detection of atrial fibrillation (AF) in continuous rhythm registrations is essential in order to prevent complications and optimize treatment of AF. Many algorithms have been developed to detect AF in surface electrocardiograms (ECGs) during the past few years. The aim of this systematic review is to gain more insight into these available classification methods by discussing previously used digital biomarkers and algorithms and make recommendations for future research. METHODS: On the 14th of September 2020, the PubMed database was searched for articles focusing on algorithms for AF detection in ECGs using the MeSH terms Atrial Fibrillation, Electrocardiography and Algorithms. Articles which solely focused on differentiation of types of rhythm disorders or prediction of AF termination were excluded. RESULTS: The search resulted in 451 articles, of which 130 remained after full-text screening. Not only did the amount of research on methods for AF detection increase over the past years, but a trend towards more complex classification methods is observed. Furthermore, three different types of features can be distinguished: atrial features, ventricular features, and signal features. Although AF is an atrial disease, only 22% of the described methods use atrial features. CONCLUSION: More and more studies focus on improving accuracy of classification methods for AF in ECGs. As a result, algorithms become increasingly complex and less well interpretable. Only a few studies focus on detecting atrial activity in the ECG. Developing innovative methods focusing on detection of atrial activity might provide accurate classifiers without compromising on transparency.

Monitoreo ambulatorio y generación de alertas tempranas ante anomalías cardiacas usando dispositivos móviles / Ambulatory monitoring and early warning generation of cardiac anomalies using mobile devices / Monitoramento ambulatorial e geração de aviso prévio de anomalias cardíacas usando dispositivos móveis

Resumen En este artículo se presenta un sistema portátil para el monitoreo ambulatorio del ritmo cardiaco y la detección temprana de las arritmias cardiacas de mayor riesgo. El sistema consta de un sensor con tres electrodos superficiales para la captura de la señal ECG, la cual se transmite vía Bluetooth a un dispositivo móvil con Android, en donde se realiza el análisis de la señal capturada durante lapsos de 5 s. El sistema propuesto distingue entre Ritmo Normal [Ritmo Sinusal - RS), Taquicardia Ventricular [TV), Fibrilación Ventricular [FV) y Asistolia, con una precisión del 100%, 55%, 75% y 90% respectivamente. Sin embargo, el sistema puede recuperarse de los errores rápidamente en el análisis de la trama subsecuente. Este trabajo se centra en el uso de dispositivos móviles de uso cotidiano, multitarea y de fácil acceso, implementando algoritmos en el dominio del tiempo para la extracción de parámetros, los cuales son idóneos para ser usados en aplicaciones móviles principalmente por su baja carga computacional y posibilidad de ejecución en tiempo real, permitiendo la detección de anomalías cardiacas de forma automática y rápida sin la necesidad de una supervisión constante por parte de un especialista para el análisis preliminar.

Abstract This paper presents a portable system for ambulatory heart rate monitoring and early detection of cardiac arrhythmias at high risk. The system consists of a sensor with three surface electrodes to capture the ECG signal, which is transmitted via bluetooth to a mobile device with Android, where the analysis is performed of the acquired signal during a time of 5 s. The proposed system distinguishes between Normal Rhythm [Rhythm Sinus - RS), Ventricular Tachycardia [VT), Ventricular Fibrillation [VF) and Asystole with an accuracy of 100%, 55%, 75% and 90% respectively. However, the system can quickly recover from errors in the subsequent analysis frame. This work focuses on using regular mobile devices which have multitasking and easy access characteristics, implementing algorithms in time domain for extracting parameters that are suitable to use in mobile applications, mainly because of their low computational load and possibility of execution in real time, allowing the detection of cardiac abnormalities automatically and quickly without the need of constant supervision by a specialist for preliminary analysis.

Resumo Neste artigo se apresenta um sistema portátil para o monitoramento da freqüência cardíaca ambulatorial e detecção precoce das arritmias cardíacas de mais risco. O sistema possui um sensor com três eletrodos superficiais para pegar o sinal ECG, o qual é transmitido via Bluetooth para um dispositivo móvel com Android, onde se faz a análise do sinal capturado durante um período de 5 s. O sistema proposto distingue entre Normal Ritmo [Ritmo Sinusal - RS), Taquicardia Ventricular [TV), Fibrilação Ventricular [FV) e Assistolia, com uma precisão do 100%, 55%, 75% e 90%, respectivamente. Porém, o sistema pode - se recuperar rapidamente dos erros na análise do quadro subsequente. Este trabalho centra-se no uso de dispositivos móveis de utilização diária, multitarefa e utilização acessível, implementação de algoritmos no domínio do tempo para a extração de parâmetros que são adequados para utilização em aplicações móveis, principalmente pela baixa carga computacional e possibilidade de execução em tempo real, permitindo a detecção de anormalidades cardíacas numa forma automática e rápida sem a necessidade de um controlo constante por um especialista para análise preliminar.

An Energy-Efficient Algorithm for Wearable Electrocardiogram Signal Processing in Ubiquitous Healthcare Applications.

Rapid progress and emerging trends in miniaturized medical devices have enabled the un-obtrusive monitoring of physiological signals and daily activities of everyone's life in a prominent and pervasive manner. Due to the power-constrained nature of conventional wearable sensor devices during ubiquitous sensing (US), energy-efficiency has become one of the highly demanding and debatable issues in healthcare. This paper develops a single chip-based wearable wireless electrocardiogram (ECG) monitoring system by adopting analog front end (AFE) chip model ADS1292R from Texas Instruments. The developed chip collects real-time ECG data with two adopted channels for continuous monitoring of human heart activity. Then, these two channels and the AFE are built into a right leg drive right leg drive (RLD) driver circuit with lead-off detection and medical graded test signal. Human ECG data was collected at 60 beats per minute (BPM) to 120 BPM with 60 Hz noise and considered throughout the experimental set-up. Moreover, notch filter (cutoff frequency 60 Hz), high-pass filter (cutoff frequency 0.67 Hz), and low-pass filter (cutoff frequency 100 Hz) with cut-off frequencies of 60 Hz, 0.67 Hz, and 100 Hz, respectively, were designed with bilinear transformation for rectifying the power-line noise and artifacts while extracting real-time ECG signals. Finally, a transmission power control-based energy-efficient (ETPC) algorithm is proposed, implemented on the hardware and then compared with the several conventional TPC methods. Experimental results reveal that our developed chip collects real-time ECG data efficiently, and the proposed ETPC algorithm achieves higher energy savings of 35.5% with a slightly larger packet loss ratio (PLR) as compared to conventional TPC (e.g., constant TPC, Gao's, and Xiao's methods).

The effect of head-up tilt upon markers of heart rate variability in patients with atrial fibrillation.

BACKGROUND: Heart rate variability (HRV) analysis is uncommonly undertaken in patients with atrial fibrillation (AF) due to an assumption that ventricular response is random. We sought to determine the effects of head-up tilt (HUT), a stimulus known to elicit an autonomic response, on HRV in patients with AF; we contrasted the findings with those of patients in sinus rhythm (SR). METHODS: Consecutive, clinically indicated tilt tests were examined for 207 patients: 176 in SR, 31 in AF. Patients in AF were compared to an age-matched SR cohort (n = 69). Five minute windows immediately before and after tilting were analyzed using time-domain, frequency-domain and nonlinear HRV parameters. Continuous, noninvasive assessment of blood pressure, heart rate and stroke volume were available in the majority of patients. RESULTS: There were significant differences at baseline in all HRV parameters between AF and age matched SR. HUT produced significant hemodynamic changes, regardless of cardiac rhythm. Coincident with these hemodynamic changes, patients in AF had a significant increase in median [quartile 1, 2] DFA-α2 (+0.14 [-0.03, 0.32], p < .005) and a decrease in sample entropy (-0.17 [-0.50, -0.01], p < .005). CONCLUSION: In the SR cohort, increasing age was associated with fewer HRV changes on tilting. Patients with AF had blunted HRV responses to tilting, mirroring those seen in an age matched SR group. It is feasible to measure HRV in patients with AF and the changes observed on HUT are comparable to those seen in patients in sinus rhythm.

An Energy efficient application specific integrated circuit for electrocardiogram feature detection and its potential for ambulatory cardiovascular disease detection.

A novel algorithm based on forward search is developed for real-time electrocardiogram (ECG) signal processing and implemented in application specific integrated circuit (ASIC) for QRS complex related cardiovascular disease diagnosis. The authors have evaluated their algorithm using MIT-BIH database and achieve sensitivity of 99.86% and specificity of 99.93% for QRS complex peak detection. In this Letter, Physionet PTB diagnostic ECG database is used for QRS complex related disease detection. An ASIC for cardiovascular disease detection is fabricated using 130-nm CMOS high-speed process technology. The area of the ASIC is 0.5 mm(2). The power dissipation is 1.73 µW at the operating frequency of 1 kHz with a supply voltage of 0.6 V. The output from the ASIC is fed to their Android application that generates diagnostic report and can be sent to a cardiologist through email. Their ASIC result shows average failed detection rate of 0.16% for six leads data of 290 patients in PTB diagnostic ECG database. They also have implemented a low-leakage version of their ASIC. The ASIC dissipates only 45 pJ with a supply voltage of 0.9 V. Their proposed ASIC is most suitable for energy efficient telemetry cardiovascular disease detection system.

Broadband noise suppression and feature identification of ECG waveforms using mathematical morphology and embedding theorem.

The paper presents an adaptive morphological filter developed using multiscale mathematical morphology (MM) to reject broadband noise from ECG signals without affecting the feature waveforms. As a pre-processing procedure, the adaptive morphological filter cleans an ECG signal to prepare it for further analysis. The noiseless ECG signal is embedded within a two-dimensional phase space to form a binary image and the identification of the feature waveforms is carried out based on the information presented by the image. The classification of the feature waveforms is implemented by an adaptive clustering technique according to the geometric information represented by the image in the phase space. Simulation studies on ECG records from the MIT-BIH and BIDMC databases have demonstrated the effectiveness and accuracy of the proposed methods.