Heartbeat Features Research Articles

A Study on Arrhythmia via ECG Signal Classification Using the Convolutional Neural Network.

Cardiovascular diseases (CVDs) are the leading cause of death today. The current identification method of the diseases is analyzing the Electrocardiogram (ECG), which is a medical monitoring technology recording cardiac activity. Unfortunately, looking for experts to analyze a large amount of ECG data consumes too many medical resources. Therefore, the method of identifying ECG characteristics based on machine learning has gradually become prevalent. However, there are some drawbacks to these typical methods, requiring manual feature recognition, complex models, and long training time. This paper proposes a robust and efficient 12-layer deep one-dimensional convolutional neural network on classifying the five micro-classes of heartbeat types in the MIT- BIH Arrhythmia database. The five types of heartbeat features are classified, and wavelet self-adaptive threshold denoising method is used in the experiments. Compared with BP neural network, random forest, and other CNN networks, the results show that the model proposed in this paper has better performance in accuracy, sensitivity, robustness, and anti-noise capability. Its accurate classification effectively saves medical resources, which has a positive effect on clinical practice.

Application of ECG Signal Denoising and T-wave Automatic Detection Based on Computer Deep Learning

To address the problems of time-consuming, laborious and costly ECG expert diagnosis in the long-term massive ECG signal automatic denoising and T-wave automatic detection system, resulting in the difficulty to extract features, as well as the poor adaptability and low accuracy of the abnormal diagnosis model, we proposed an ECG signal denoising and T-wave automatic detection method based on deep learning. The method mainly includes four steps as follows: ECG signal denoising preprocessing, segmentation of ECG signal and unification of sampling points, unsupervised heartbeat feature learning, denoising and T-wave automatic detection. The structure of the deep faith network FCMDBN model and the learning denoising and T-wave automatic detection algorithm are proposed. Based on the MIT-BIH heart rate anomaly database, the simulation experiment shows that the signal diagnosis method proposed in this paper has relatively high adaptability and accuracy compared with the denoising and T-wave automatic detection by artificial design based on traditional ECG features.

Heart Sound Classification Using Multi Modal Data Representation and Deep Learning

Heart anomalies are an important class of medical conditions from personal, public health and social perspectives and hence accurate and timely diagnoses are important. Heartbeat features two well known amplitude peaks termed S1 and S2. Some sound classification models rely on segmented sound intervals referenced to the locations of detected S1 and S2 peaks, which are often missing due to physiological causes and/or artifacts from sound sampling process. The constituent and combined models we propose are free from segmentation, which consequently is more robust and meritful from reliability aspects. Intuitive phonocardiogram representation with relatively simple deep learning architecture was found to be effective for classifying normal and abnormal heart sounds. A frequency spectrum based deep learning network also produced competitive classification results. When the classification models were merged in one via SVM, performance was seen to improve further. The SVM classification model, comprised of two time domain submodels and a frequency domain submodel, produced 0.9175 sensitivity, 0.8886 specificity and 0.9012 accuracy.

Analysis and classification of heart diseases using heartbeat features and machine learning algorithms

This study proposed an ECG (Electrocardiogram) classification approach using machine learning based on several ECG features. An electrocardiogram (ECG) is a signal that measures the electric activity of the heart. The proposed approach is implemented using ML-libs and Scala language on Apache Spark framework; MLlib is Apache Spark’s scalable machine learning library. The key challenge in ECG classification is to handle the irregularities in the ECG signals which is very important to detect the patient status. Therefore, we have proposed an efficient approach to classify ECG signals with high accuracy Each heartbeat is a combination of action impulse waveforms produced by different specialized cardiac heart tissues. Heartbeats classification faces some difficulties because these waveforms differ from person to another, they are described by some features. These features are the inputs of machine learning algorithm. In general, using Spark–Scala tools simplifies the usage of many algorithms such as machine-learning (ML) algorithms. On other hand, Spark–Scala is preferred to be used more than other tools when size of processing data is too large. In our case, we have used a dataset with 205,146 records to evaluate the performance of our approach. Machine learning libraries in Spark–Scala provide easy ways to implement many classification algorithms (Decision Tree, Random Forests, Gradient-Boosted Trees (GDB), etc.). The proposed method is evaluated and validated on baseline MIT-BIH Arrhythmia and MIT-BIH Supraventricular Arrhythmia database. The results show that our approach achieved an overall accuracy of 96.75% using GDB Tree algorithm and 97.98% using random Forest for binary classification. For multi class classification, it achieved to 98.03% accuracy using Random Forest, Gradient Boosting tree supports only binary classification.

Denoising and Features Extraction of ECG Signals in State Space Using Unbiased FIR Smoothing

The electrocardiogram (ECG) signals bear fundamental information for making decisions about different kinds of heart diseases. Therefore, many efforts were made during decades to extract features of heartbeats via ECG records with high accuracy and efficiency using different strategies and methods. In this paper, we solve the problem in discrete-time state-space using a novel q-lag unbiased finite impulse response (UFIR) smoother, which we adapt to the ECG signal shape via the time-varying optimal averaging horizon. It is shown that the adaptive UFIR smoother performs better in applications to ECG signals than the standard techniques such as the Savitsky-Golay, wavelet-based, low-pass, band-pass, notch, and median filters. Applications are given for the PhysioBank data benchmark, which contains several records taken from different databases such as the MIT-BIH Arrhythmia (MITDB). A complete statistical analysis is provided via normalized histograms and statistical classifiers. It is shown in a comparison with other methods that the adaptive UFIR smoother has a higher accuracy in denoising, features extraction, and features classification for ECG records with normal rhythm and atrial fibrillation (AF).

Unlock with Your Heart

In this paper, we propose to use the vibration of the chest in response to the heartbeat as a biometric feature to authenticate the user on mobile devices. We use the built-in accelerometer to capture the heartbeat signals on commercial mobile phones. The user only needs to press the phone on his/her chest, and the system can identify the user within a few heartbeats. To reliably extract heartbeat features, we design a two-step alignment scheme that can handle the natural variability in human heart rates. We further use an adaptive template selection scheme to authenticate the user under different body postures and body states. Based on heartbeat signals collected on twenty users, the experimental results show that our method can achieve an authentication accuracy of 96.49% and the heartbeat features are stable over a period of three months.

System Design Approach for Heartbeat Detection and Classification of Individuals Irrespective of their Physical Condition

In an electrocardiogram (ECG), the heartbeat feature QRS is an important parameter for analysis in any heartbeat classification automated diagnosis system. In this communication the method which we have proposed is by using the counter which is used in parallel. The first level is detection of heartbeats, which uses hashing of ECG features. In the second level, the profiler profiles a person's regular and irregular ECG characteristic behaviour. The proposed method works on data related with ECG, instead of particular features of ECG. Because of parallel processing, data storage unit requirements and the processing time are less. The dependent values in the proposed method vary according to the changes in the ECG waveform. Such type of analysis is suitable for detection of heart disease. The most significant application of such characteristic plotting is to generate an alert signal for irregular ECG behaviour in a person. Such automated system will be useful in remote areas where a cardiologist may not be easily available.

Cardiac arrhythmia classification using statistical and mixture modeling features of ECG signals

In this paper we propose a novel method for accurate classification of cardiac arrhythmias. Morphological and statistical features of individual heartbeats are used to train a classifier. Two RR interval features as the exemplars of time-domain information are utilized in this study. Gaussian mixture modeling (GMM) with an enhanced expectation maximization (EM) solution is used to fit the probability density function of heartbeats. Parameters of GMM together with shape parameters such as skewness, kurtosis and 5th moment are also included in feature vector. These features are then used to train an ensemble of decision trees. MIT-BIH arrhythmia database containing various types of common arrhythmias is employed to test the algorithm. The overall accuracy of 99.70% in “class-oriented” scheme and 96.15% in “subject-oriented” scheme is achieved. Both cases express a significant improvement of accuracy compared to other methods.

Human recognition using Fisher׳s discriminant analysis of heartbeat interval features and ECG morphology

The author of this paper presents a novel method to use the electrocardiogram (ECG) as a biometric for human recognition. The ECG is a physiological signal that links to the life signs of an individual and does not need vitality testing. Thus, ECG biometric system has potential to prevent the fraudulent attacks. The hybrid approach consisting analytical and appearance methods is used to derive the ECG features. In order to make the method insensitive to signal variations and muscle flexure, the ECG features are linearly projected using Fisher׳s discriminant method. The method selects heartbeat features of lower dimension in the Fisher space that have sufficient discriminatory information between inter-subject ECG signals. The experiment shows that the proposed ECG biometric method achieves the equal error rates (EER) of 0.76% and 0.71% in recognizing people suffering from cardiac arrhythmia and people of good health, respectively. On mixed health statuses, the method achieves an EER of 1.31% confirming a very good performance and robustness of the proposal.

Non-invasive Drosophila ECG recording by using eutectic gallium-indium alloy electrode: a feasible tool for future research on the molecular mechanisms involved in cardiac arrhythmia.

Background Drosophila heart tube is a feasible model for cardiac physiological research. However, obtaining Drosophila electrocardiograms (ECGs) is difficult, due to the weak signals and limited contact area to apply electrodes. This paper presents a non-invasive Gallium-Indium (GaIn) based recording system for Drosophila ECG measurement, providing the heart rate and heartbeat features to be observed. This novel, high-signal-quality system prolongs the recording time of insect ECGs, and provides a feasible platform for research on the molecular mechanisms involved in cardiovascular diseases.MethodsIn this study, two types of electrode, tungsten needle probes and GaIn electrodes, were used respectively to noiselessly conduct invasive and noninvasive ECG recordings of Drosophila. To further analyze electrode properties, circuit models were established and simulated. By using electromagnetic shielded heart signal acquiring system, consisted of analog amplification and digital filtering, the ECG signals of three phenotypes that have different heart functions were recorded without dissection.Results and DiscussionThe ECG waveforms of different phenotypes of Drosophila recorded invasively and repeatedly with n value (n>5) performed obvious difference in heart rate. In long period ECG recordings, non-invasive method implemented by GaIn electrodes acts relatively stable in both amplitude and period. To analyze GaIn electrode, the correctness of GaIn electrode model established by this paper was validated, presenting accuracy, stability, and reliability.ConclusionsNoninvasive ECG recording by GaIn electrodes was presented for recording Drosophila pupae ECG signals within a limited contact area and signal strength. Thus, the observation of ECG changes in normal and SERCA-depleted Drosophila over an extended period is feasible. This method prolongs insect survival time while conserving major ECG features, and provides a platform for electrophysiological signal research on the molecular mechanism involved in cardiac arrhythmia, as well as research related to drug screening and development.

Matlab-based tool for ECG and HRV analysis

Delivery of electroporation pulses in electroporation-based treatments could potentially induce heart-related effects. The objective of our work was to develop a software tool for electrocardiogram (ECG) analysis to facilitate detection of such effects in pre-selected ECG- or heart rate variability (HRV) parameters.Our software tool consists of five distinct modules for: (i) preprocessing; (ii) learning; (iii) detection and classification; (iv) selection and verification; and (v) ECG and HRV analysis. Its key features are: automated selection of ECG segments from ECG signal according to specific user-defined requirements (e.g., selection of relatively noise-free ECG segments); automated detection of prominent heartbeat features, such as Q, R and T wave peak; automated classification of individual heartbeat as normal or abnormal; displaying of heartbeat annotations; quick manual screening of analyzed ECG signal; and manual correction of annotation and classification errors.The performance of the detection and classification module was evaluated on 19 two-hour-long ECG records from Long-Term ST database. On average, the QRS detection algorithm had high sensitivity (99.78%), high positive predictivity (99.98%) and low detection error rate (0.35%). The classification algorithm correctly classified 99.45% of all normal QRS complexes. For normal heartbeats, the positive predictivity of 99.99% and classification error rate of 0.01% were achieved.The software tool provides for reliable and effective detection and classification of heartbeats and for calculation of ECG and HRV parameters. It will be used to clarify the issues concerning patient safety during the electroporation-based treatments used in clinical practice. Preventing the electroporation pulses from interfering with the heart is becoming increasingly important because new applications of electroporation-based treatments are being developed which are using endoscopic, percutaneous or surgical means to access internal tumors or tissues and in which the target tissue can be located in immediate vicinity to the heart.

Individual Identification Using Linear Projection of Heartbeat Features

This paper presents a novel method to use the electrocardiogram (ECG) signal as biometrics for individual identification. The ECG characterization is performed using an automated approach consisting of analytical and appearance methods. The analytical method extracts the fiducial features from heartbeats while the appearance method extracts the morphological features from the ECG trace. We linearly project the extracted features into a subspace of lower dimension using an orthogonal basis that represent the most significant features for distinguishing heartbeats among the subjects. Result demonstrates that the proposed characterization of the ECG signal and subsequently derived eigenbeat features are insensitive to signal variations and nonsignal artifacts. The proposed system utilizing ECG biometric method achieves the best identification rates of 85.7% for the subjects of MIT-BIH arrhythmia database and 92.49% for the healthy subjects of our IIT (BHU) database. These results are significantly better than the classification accuracies of 79.55% and 84.9%, reported using support vector machine on the tested subjects of MIT-BIH arrhythmia database and our IIT (BHU) database, respectively.

Point-process nonlinear autonomic assessment of depressive states in bipolar patients.

This article is part of the Focus Theme of Methods of Information in Medicine on "Biosignal Interpretation: Advanced Methods for Studying Cardiovascular and Respiratory Systems". The goal of this work is to apply a computational methodology able to characterize mood states in bipolar patients through instantaneous analysis of heartbeat dynamics. A Point-Process-based Nonlinear Autoregressive Integrative (NARI) model is applied to analyze data collected from five bipolar patients (two males and three females, age 42.4 ± 10.5 range 32 -56) undergoing a dedicated affective elicitation protocol using images from the International Affective Picture System (IAPS) and Thematic Apperception Test (TAT). The study was designed within the European project PSYCHE (Personalised monitoring SYstems for Care in mental HEalth). RESULTS demonstrate that the inclusion of instantaneous higher order spectral (HOS) features estimated from the NARI nonlinear assessment significantly improves the accuracy in successfully recognizing specific mood states such as euthymia and depression with respect to results using only linear indices. In particular, a specificity of 74.44% using the instantaneous linear features set, and 99.56% using also the nonlinear feature set were achieved. Moreover, IAPS emotional elicitation resulted in a more discriminant procedure with respect to the TAT elicitation protocol. A significant pattern of instantaneous heartbeat features was found in depressive and euthymic states despite the inter-subject variability. The presented point-process Heart Rate Variability (HRV) nonlinear methodology provides a promising application in the field of mood assessment in bipolar patients.

Identifying Individuals Using Eigenbeat Features of Electrocardiogram

The authors of this paper present a new method to characterize the electrocardiogram (ECG) for individual identification. We propose an ECG biometric system which is insensitive to noise signals and muscle flexure. The method utilizes the principal of linearly projecting the heartbeat features into a subspace of lower dimension using an orthogonal basis that represents the most significant features to distinguish the individuals. The performance of the proposed biometric system is evaluated on the subjects of both health statuses such as the ECG recordings of MIT-BIH Arrhythmia database and the ECG recordings of normal subjects prepared at IIT(BHU). The result demonstrates that the derived eigenbeat features from proposed ECG characterization perform better and achieve the recognition accuracy of 91.42% and 95.55% on the subjects of MIT-BIH Arrhythmia database and IIT(BHU) database, respectively.

Instantaneous Assessment of Depressive State in Bipolar Patients by Point-Process Nonlinear Models of Heartbeat Dynamics

Within he conceptual framework of the European funded project PSYCHE, we present a novel experimental/methodological approach for the assessment of autonomic patterns of depression in bipolar patients. A novel PointProcess-based Nonlinear Autoregressive Integrative model is applied to analyze heartbeat data, demonstrating that the inclusion of instantaneous nonlinear features significantly improve discrimination accuracy. Introduction: Bipolar disorder is a psychiatric condition in which patients experience episodes of swing, such as oscillations between depressive states and euthymic states (i.e., relatively good affective balance). Despite its prevalence, this illness may go undetected for years. Moreover, the patient’s diagnosis is mainly performed by clinicianadministered rating scales and no biological markers nor physiological signals highlighted in research studies are used for clinical purposes. The goal of our research is to be able to accurately recognize depressive states in bipolar patients from the analysis of the heartbeat dynamics. Data were gathered within the European project PSYCHE (Personalised monitoring SYstems for Care in mental HEalth) using wearable systems such as textile-based sensorized t-shirt. Methods: We propose a novel methodological approach based on a point-process nonlinear derivative model. This powerful, fully-parametric statistical tool accounts for the probabilistic generative mechanism of the heartbeat, further considering a quadratic Wiener-Volterra representation of the first order moment of a physiological plausible inversegaussian statistics and allowing the estimation of the instantaneous autospectrum and bispectrum even in short recordings and under nonstationary conditions. As the framework is defined in continuous time, it is possible to estimate instantaneous heart rate and heart rate variability indices without using any interpolation method. Kolmogorov-Smirnov (KS) test and quantiles’ autocorrelation plots are used to evaluate the model goodness-of-fit and to test independence of the model-transformed intervals, respectively. By integrating the * Corresponding Author. G. Valenza, L. Citi, and R. Barbieri are with the Neuroscience Statistics Research Laboratory, Harvard Medical School, Massachusetts General Hospital, Boston, MA 02114 USA, and also with the Massachusetts Institute of Technology, Cambridge, MA 02139 USA (e-mail: g.valenza@ieee.org; {lciti,barbieri}@neurostat.mit.edu). G. Valenza, A. Lanata and E.P. Scilingo are with the Interdepartmental Research Center E. Piaggio and also with the Department of Information Engineering, University of Pisa, Pisa, Italy (e-mail: {a.lanata,e.scilingo}@centropiaggio.unipi.it). C. Gentili is with the Department of surgical, medical, molecular, and critical area pathology, section of psychology, University of Pisa, Pisa, Italy (e-mail: claudio.gentili@med.unipi.it). obtained instantaneous spectrum, it is possible to compute the HRV index within the LF (0.04-0.15 Hz) and HF (0.150.5 Hz) ranges along with their ratio, which has been taken as index of sympatho-vagal balance. Similarly, the double integration performed on the bispectral plane in the appropriate frequency bands can be interpreted as an index of nonlinear interactions within the sympathetic system (LL index), nonlinear interactions within the parasympathetic system (HH index), and nonlinear interactions between the sympathetic and parasympathetic systems (LH index). Experimental Protocol: The point process was applied to data coming from three bipolar patients having a mood label ( euthymia or depression) undergoing a dedicated affective elicitation protocol. The dedicated affective elicitation protocol started with two five minutes lasting phases of resting state, with eyes closed and open respectively. Subsequently, a passive emotional elicitation by images gathered from the International Affective Picture System and the Thematic Apperception was performed. Results: The optimal model order was chosen by means of the Akaike Information Criterion applied to the first 5-min RR recordings. Three out of the four KS plots were inside the confidence intervals and all the KS distances were < 0.0375. No less than 97% of the autocorrelation points were inside the boundaries. An inter-subject analysis was performed to reveal the common pattern among patients using the well-known Multilayer Perceptron Neural Network. A first feature set, α, was composed by only the standard HRV feature set (mean, standard deviation, LF, HF, LF/HF), whereas the nonlinear indices (LL, LH, and HH) were added to the α set to create the second set, β. A classification accuracy of 93.26% for the α set, and 99.33 % for the β set was achieved. Therefore, a common pattern of heartbeat features was found despite the inter-subject variability. Results clearly demonstrate the importance of the inclusion of the nonlinear indices in improving the discrimination. Moreover, the inclusion of the nonlinear indices gives smaller variance with respect to the classification performed by using only α sets (i.e. the standard features). Given their preliminary nature, these results are very promising. The proposed pointprocess nonlinear analysis represents a pioneering study in the field of assessment in bipolar patients.

The Heartbeat Classification Based on PCA

Classify the electrocardiogram (ECG) into different pathophysiological categories is a complex pattern recognition task which has been tried in lots of methods. This paper will discuss a method of principal component analysis (PCA) in exacting the heartbeat features, and a new method of classification that is to calculate the error between the testing heartbeat and reconstructed heartbeat. Training and testing heartbeat is taken from the MIT-BIH Arrhythmia Database, in which 8 types of arrhythmia signals are selected in this paper. The true positive rate (TPR) is 83%.

Correlation-based classification of heartbeats for individual identification

This paper proposes new techniques to delineate P and T waves efficiently from heartbeats. The delineation results have been found to be optimum and stable in comparison to other published results. These delineators are used along with QRS complex to extract various features of classes time interval, amplitude and angle from clinically dominant fiducials on each heartbeat of the electrocardiogram (ECG). A new identification system has been proposed in this study, which uses these features and makes the decision on the identity of an individual with respect to a given database. The system has been tested against a set of 250 ECG recordings prepared from 50 individuals of Physionet. The matching decisions are made on the basis of correlation between heartbeat features among individuals. The proposed system has achieved an equal error rate of less than 1.01 with an accuracy of 99%.

Unsupervised classification of atrial heartbeats using a prematurity index and wave morphology features

ECG heartbeat type detection and classification are regarded as important procedures since they can significantly help to provide an accurate automated diagnosis. This paper addresses the specific problem of detecting atrial premature beats, that had been demonstrated to be a marker for stroke risk or cardiac arrhythmias. The proposed methodology consists of a stage to estimate characteristics such as morphology of P wave and QRS complex as well as indices of prematurity and a non-supervised stage used by the algorithm J-means to separate heartbeat feature vectors into classes. Partition initialization is carried out by a Max-Min approach. Experimental data set is taken from MIT-BIH arrhythmia database. Results evidence the reliability of the method since achieved sensitivity and specificity are high, 92.9 and 99.6%, respectively, for an average output number of 12 discovered clusters that can be considered as appropriate value to separate heartbeat classes from recordings.

Hierarchical Classification of ECG Beat Using Higher Order Statistics and Hermite Model

Objective: The heartbeat classification of the electrocardiogram is important in cardiac disease diagnosis. For detecting QRS complex, conventional detection algorithm have been designed to detect P, QRS, Twave, first. However, the detection of the P and T wave is difficult because their amplitudes are relatively low, and occasionally they are included in noise. Furthermore the conventional multiclass classification method may have skewed results to the majority class, because of unbalanced data distribution. Methods: The Hermite model of the higher order statistics is good characterization methods for recognizing morphological QRS complex. We applied three morphological feature extraction methods for detecting QRS complex: higher-order statistics, Hermite basis functions and Hermite model of the higher order statistics. Hierarchical scheme tackle the unbalanced data distribution problem. We also employed a hierarchical classification method using support vector machines. Results: We compared classification methods with feature extraction methods. As a result, our mean values of sensitivity for hierarchical classification method (75.47%, 76.16% and 81.21%) give better performance than the conventional multiclass classification method (46.16%). In addition, the Hermite model of the higher order statistics gave the best results compared to the higher order statistics and the Hermite basis functions in the hierarchical classification method. Conclusion: This research suggests that the Hermite model of the higher order statistics is feasible for heartbeat feature extraction. The hierarchical classification is also feasible for heartbeat classification tasks that have the unbalanced data distribution.

Unsupervised classification of ventricular extrasystoles using bounded clustering algorithms and morphology matching

Ventricular extrasystoles (VE) are ectopic heartbeats involving irregularities in the heart rhythm. VEs arise in response to impulses generated in some part of the heart different from the sinoatrial node. These are caused by the premature discharge of a ventricular ectopic focus. VEs after myocardial infarction are associated with increased mortality. Screening of VEs is typically a manual and time consuming task that involves analysis of the heartbeat morphology, QRS duration, and variations of the RR intervals using long-term electrocardiograms. We describe a novel algorithm to perform automatic classification of VEs and report the results of our validation study. The proposed algorithm makes use of bounded clustering algorithms, morphology matching, and RR interval length to perform automatic VE classification without prior knowledge of the number of classes and heartbeat features. Additionally, the proposed algorithm does not need a training set.