PTB Diagnostic ECG Database Research Articles

Hybrid LSTM models-based detection of left ventricular hypertrophy in electrocardiogram signals

This research investigates various deep learning techniques to automatically classify Left Ventricular Hypertrophy (LVH) from electrocardiogram (ECG) signals. LVH frequently results from persistently high blood pressure, causing the heart pump harder and thicken the ventricular walls. It is associated with an increased risk of heart attacks, heart failure, stroke, and sudden cardiac death. The significance of this research lies in the early and precise detection of LVH, facilitating timely interventions and ultimately improving patient health. The non-invasive nature of ECG monitoring, integrated with the efficiency of deep learning models, contributes to faster and more accessible to enhance diagnostic accuracy and efficiency in identifying LVH. The objective of this research is to assess and compare the performance of GRU3Net, Double-Bilayer LSTM, and Conv2LSTM, Dual-LSTM models in the classification of Left Ventricular Hypertrophy (LVH) based on electrocardiogram (ECG) signals, utilizing a dataset sourced from the PTB Diagnostic ECG Database. The implemented deep learning models yielded noteworthy results. Specifically, the GRU3Net model achieved a high accuracy of 96.1%, showcasing an optimal configuration for overall accuracy. The Double-Bilayer LSTM model followed with an accuracy of 91.7%. However, a decline in accuracy was observed in both the Dual-LSTM and Conv2LSTM models, with the former registering an accuracy of 90.8% and the latter decreasing further to 87.3%.

A multi-channel ECG signal deep compressive sensing method using Treeshaped Autoecoder based on multiscale feature fusion

Conventionally, multi-lead Electrocardiogram (ECG) signals are recorded and stored using high sensing rates and high precision, resulting in huge data volumes and placing greater pressure on storage and transmission resources. Compressive sensing (CS) allows efficient encoding and decoding of signals through sparse representation, measurement and reconstruction for compressed storage and transmission of ECG signals. Traditional CS methods generally requires manually selecting the features or dictionaries used in sparse representations, which lacks adaptivity and flexibility, and the complexity of the reconstruction process limits the application of CS in some scenarios with high real-time requirements. In this paper, we propose a deep compressive sensing framework for processing multi-lead ECG signals by combining CS and deep learning, which is based on multi-scale feature fusion to construct a binary tree-shaped autoencoder architecture to achieve efficient compression and reconstruction of ECG signals. Experiments on the PTB diagnostic ECG database show that the proposed method achieves “very good” reconstruction quality at low sensing rates, with PRD and SNR of 1.77 % and 35.67 dB, respectively, when SR = 20 %. In addition, we investigate how the number of quantization bits affects the quality of reconstructed signals. After analysis, we found that 8-bit quantization can be used in practical applications to ensure the quality of reconstructed ECG signals, while decrease the bit rate of the data, improving the transmission efficiency, and reduce energy consumption.

Convolutional Neural Network for Individual Identification Using Phase Space Reconstruction of Electrocardiogram

Electrocardiogram (ECG) biometric provides an authentication to identify an individual on the basis of specific cardiac potential measured from a living body. Convolutional neural networks (CNN) outperform traditional ECG biometrics because convolutions can produce discernible features from ECG through machine learning. Phase space reconstruction (PSR), using a time delay technique, is one of the transformations from ECG to a feature map, without the need of exact R-peak alignment. However, the effects of time delay and grid partition on identification performance have not been investigated. In this study, we developed a PSR-based CNN for ECG biometric authentication and examined the aforementioned effects. Based on a population of 115 subjects selected from the PTB Diagnostic ECG Database, a higher identification accuracy was achieved when the time delay was set from 20 to 28 ms, since it produced a well phase-space expansion of P, QRS, and T waves. A higher accuracy was also achieved when a high-density grid partition was used, since it produced a fine-detail phase-space trajectory. The use of a scaled-down network for PSR over a low-density grid with 32 × 32 partitions achieved a comparable accuracy with using a large-scale network for PSR over 256 × 256 partitions, but it had the benefit of reductions in network size and training time by 10 and 5 folds, respectively.

Signal Acquisition-Independent Lossless Electrocardiogram Compression Using Adaptive Linear Prediction.

In this paper, we propose a lossless electrocardiogram (ECG) compression method using a prediction error-based adaptive linear prediction technique. This method combines the adaptive linear prediction, which minimizes the prediction error in the ECG signal prediction, and the modified Golomb-Rice coding, which encodes the prediction error to the binary code as the compressed data. We used the PTB Diagnostic ECG database, the European ST-T database, and the MIT-BIH Arrhythmia database for the evaluation and achieved the average compression ratios for single-lead ECG signals of 3.16, 3.75, and 3.52, respectively, despite different signal acquisition setup in each database. As the prediction order is very crucial for this particular problem, we also investigate the validity of the popular linear prediction coefficients that are generally used in ECG compression by determining the prediction coefficients from the three databases using the autocorrelation method. The findings are in agreement with the previous works in that the second-order linear prediction is suitable for the ECG compression application.

Multi-lead ECG heartbeat classification of heart disease based on HOG local feature descriptor

IntroductionECG data play an important role in the diagnostics of various cardiovascular diseases. Classification of multi-lead ECG signals could be challenging even for well-trained physicians. In this study we propose a new approach for multi-lead ECG classification. MethodFive-types of 15-lead ECG data namely healthy control, bundle branch block, cardiomyopathy, Dysrhythmia, and myocardial infarction patients from two types of datasets, 5319 and 6647 heartbeats from Baqiyatallah and PTB Diagnostic ECG database, were used, respectively. One-dimensional total variation regularization was used to denoising ECG data. Heartbeats were extracted by one cardiologist and saved as images with jpg format. Histogram of oriented gradients method was used to extract feature of images. for classification task support vector machine and fully connected neural network were used. Five-fold cross validation was used for validating the models. ResultFor 15-lead ECG PTB Diagnostic database, the best classification models are SVM model with cubic (accuracy: 99.9%, Range: 99.77% - 100%) and quadratic (accuracy: 99.88%, Range: 99.77%-100%) kernel function, for this dataset fully connected accuracy is 99.4% with range of 99.02%- 99.70%. Regarding to the Baqyatallah dataset SVM with cubic (accuracy: 99.83%, Range:99.72%-100%) and quadratic (accuracy: 99.77%, Range: 99.62%-99.9%) were the best classification model and the accuracy for fully connected neural network was 99.1% with the range of 98.59%-99.62% based on HOG descriptors. Expected sigmodal kernel all classification method have accuracy more than 99%. Discussionsimultaneous use of HOG feature extraction method and appropriate classification algorithm such as SVM or fully connected neural network can classify 15-lead ECG heart-beat for different heart disease with high accuracy and adding other relevant patients’ information can be easily done in order to increase the method performance.

Robust Classification of Cardiac Arrhythmia Using Machine Learning

Abstract: Machine Learning (ML) is developing progressively in the healthcare industry. The classifying of electrocardiogram (ECG) results depending on Cardiac Arrhythmia is one such instance, where ECG record's the heart's rhythm & electric activity. Two ML techniques, Random Forest (RF) and Convolution Neural Network (CNN) are employed in the presented work to yield rapid & effective categorization of heartbeats. The heartbeats are divided into five classes applying the PTB Diagnostic ECG Database plus the MIT-BIH Arrhythmia Collection. Both datasets are massive, with data being unbalanced for all five classes. The balancing of data is done using both oversampling & under sampling approaches to bring the data equal for each class. Based on performance indicators like f1-score, precision, recall, & accuracy, a comparison will be done among RF and CNN methods. Keywords: ECG, Cardiac Arrhythmia, RF, CNN

Agreement process of ECG annotations using thesaurus (list of typical phrases) of ECG conclusions

A thesaurus (logically ordered list) of statements for the ECG annotation has been developed. 549 ECGs from the PTB Diagnostic ECG Database were independently annotated by 3 clinicians using this thesaurus and web interface. The resulting independent annotations did not match (the proportion of matches was from 3 to 100% for different groups of conclusions in pairwise comparison of annotators). The annotations were agreed, for which a special procedure was developed. The procedure includes participation of moderators (highly qualified physicians) and using videoconferencing. Annotations are published in the public domain and can be used both for the purposes of knowledge control and for testing the accuracy of the automatic ECG analysis algorithms and software.

Localization of myocardial infarction with multi-lead ECG based on DenseNet

Background and Objective: Myocardial infarction (MI) is a critical acute ischemic heart disease, which can be early diagnosed by electrocardiogram (ECG). However, the most research of MI localization pay more attention on the specific changes in every ECG lead independent. In our study, the research envisages the development of a novel multi-lead MI localization approach based on the densely connected convolutional network (DenseNet).Methods: Considering the correlation of the multi-lead ECG, the method using parallel 12-lead ECG, systematically exploited the correlation of the inter-lead signals. In addition, the dense connection of DenseNet enhanced the reuse of the feature information between the inter-lead and intra-lead signals. The proposed method automatically captured the effective pathological features, which improved the identification of MI.Results: The experimental results based on PTB diagnostic ECG database showed that the accuracy, sensitivity and specificity of the proposed method was 99.87%, 99.84% and 99.98% for 11 types of MI localization.Conclusions: The proposed method has achieved superior results compared to other localization methods, which can be introduced into the clinical practice to assist the diagnosis of MI.

Classification of Arrhythmia in Heartbeat Detection Using Deep Learning.

The electrocardiogram (ECG) is one of the most widely used diagnostic instruments in medicine and healthcare. Deep learning methods have shown promise in healthcare prediction challenges involving ECG data. This paper aims to apply deep learning techniques on the publicly available dataset to classify arrhythmia. We have used two kinds of the dataset in our research paper. One dataset is the MIT-BIH arrhythmia database, with a sampling frequency of 125 Hz with 1,09,446 ECG beats. The classes included in this first dataset are N, S, V, F, and Q. The second database is PTB Diagnostic ECG Database. The second database has two classes. The techniques used in these two datasets are the CNN model, CNN + LSTM, and CNN + LSTM + Attention Model. 80% of the data is used for the training, and the remaining 20% is used for testing. The result achieved by using these three techniques shows the accuracy of 99.12% for the CNN model, 99.3% for CNN + LSTM, and 99.29% for CNN + LSTM + Attention Model.

Quality Aware Compression of Multilead Electrocardiogram Signal using 2-mode Tucker Decomposition and Steganography

In this article, a quality controlled compression of multilead electrocardiogram (MECG) is proposed, based on tensor analysis, and implemented upon 3D beat tensor of MECG. To reduce computational complications and execution time, a new approach of principal component analysis (PCA), based on 2-mode Tucker Decomposition, is introduced. In order to maintain relevant features of MECG after reconstruction, multi agent supervised learning system (MASLS) based optimal quantization of each fiber of core tensor is introduced, to limit the percentage root mean squared difference (PRD) within a specified value, maintaining high compression ratio (CR). The MASLS is previously trained offline, using features of tensor fibers, along with optimized quantization levels of those fibers, obtained from a particle swarm optimization (PSO), as reference. In addition, to hide patient’s confidential information, steganography is performed within the core tensor followed by generation of a’ secret key’, which is necessary, while decrypting those information during reconstruction. The whole algorithm is implemented on several MECG records, available in PTB Diagnostic ECG database, and compression result is compared by formation of n-beat tensor separately, using ‘n’ number of successive (‘n’ = 5, 10 and 15) beats. After testing on 547 data, average CR of 22, 41.5, 55.4, PRD of 3.62, 4.96, 5.59 and PRD normalized (PRDN) of 3.61, 4.94, 5.57 are achieved for 5, 10, 15-beat tensor, respectively. This proposed algorithm has provided superior result as compared to recently published works on MECG data compression.

Efficient dynamic modelling of ECG with myocardial infarction using interacting multiple model and particle filter

The automatic identification of cardiac condition through dynamic modelling of electrocardiogram (ECG) signal has immense clinical significance as it eliminates the task of manual annotation of ECG recordings. In this study, an interacting multiple model (IMM)-based scheme has been proposed that helps in dynamically modelling and estimating the ECG signal. This model is having the adaptability of interchanging among several morphological representations. It offers the advantage of not necessitating user-specific parameters. This model does not necessitate a priori information about the ECG signal to initialise the filter parameters and delimitation of fiducial points of ECG signal. The particle filter (PF)-based schemes show superiority owing to their freedom from a single assumption on the signal model and noise model. Besides, it has got the potential of simultaneously tracking multiple pathological and morphological changes occurring in biomedical signals. Thus, the parameters of the model are estimated by adopting the PF, so that the myocardial infarction (MI) affected ECG signals can be efficiently tracked. Investigations on ECG signals from the MIT-BIH database and PTB diagnostic ECG database signify that the IMM-PF scheme can represent several MI morphologies with minimum prior information without distorting the helpful diagnostic information of ECG accurately.

Machine learning-based left ventricular hypertrophy detection using multi-lead ECG signal

This work proposes a novel method for the detection of Left Ventricular Hypertrophy (LVH) from a multi-lead ECG signal. Left Ventricle walls become thick due to prolonged hypertension which may fail to pump heart effectively. The imaging techniques can be used as an alternative diagnose LVH; however, they are more expensive and time-consuming than proposed LVH. To overcome this issue, an algorithm to the diagnosis of LVH using ECG signal based on machine learning techniques were designed. In LVH detection, the pathological attributes such as R wave, S wave, inversion of QRS complex, changes in ST segment noticed in the ECG signal. This clinical information extracted as a feature by applying continuous wavelet transform. The signals were reconstructed with the frequency between 10 and 50 Hz from the wavelet. This followed by the detection of R wave and S wave peaks to obtain the relevant LVH diagnostic features. The Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Ensemble of Bagged Tree, AdaBoost classifiers were employed and the results are compared with four neural network classifiers including Multilayer Perceptron (MLP), Scaled Conjugate Gradient Backpropagation Neural Network (SCG NN), Levenberg–Marquardt Neural Network (LMNN) and Resilient Backpropagation Neural network (RPROP). The data source includes Left Ventricular Hypertrophy and healthy ECG signal from PTB diagnostic ECG database and St Petersburg INCART 12-Lead Arrhythmia Database. The results revealed that the proposed work can diagnose LVH successfully using neural network classifiers. The accuracy in detecting LVH is 86.6%, 84.4%, 93.3%,75.6%, 95.6%, 97.8%, 97.8%, 88.9% using SVM, KNN, Ensemble of Bagged Tree, AdaBoost, MLP, SCG NN, LMNN and RPROP classifiers, respectively.

Energy-efficient Real-time Myocardial Infarction Detection on Wearable Devices.

Myocardial Infarction (MI) is a fatal heart disease that is a leading cause of death. The silent and recurrent nature of MI requires real-time monitoring on a daily basis through wearable devices. Real-time MI detection on wearable devices requires a fast and energy-efficient solution to enable long term monitoring. In this paper, we propose an MI detection methodology using Binary Convolutional Neural Network (BCNN) that is fast, energy-efficient and outperforms the state-of-the- art work on wearable devices. We validate the performance of our methodology on the well known PTB diagnostic ECG database from PhysioNet. Evaluation on real hardware shows that our BCNN is faster and achieves up to 12x energy efficiency compared to the state-of-the-art work.

Automated detection of myocardial infarction in ECG using modified Stockwell transform and phase distribution pattern from time-frequency analysis

Myocardial infarction (MI), usually referred as heart attack, takes place when blood circulation stops to specific portion of the heart resulting permanent damage to the heart muscles. It is an important task to identify the occurrence of MI from the ECG recordings efficiently. Most of the detection procedures include advanced signal processing methods, more ECG features and composite classifiers, making the overall procedure complex. This paper aims at automated identification of MI using modified Stockwell transform (MST) based time-frequency analysis and a phase information distribution pattern method. The morphological, pathological and temporal alterations in ECG waveforms resulting from the onset of MI are noticed in the phase distribution pattern of the ECG signal. Two discriminating features, utterly reflecting these alterations, are recognized for 12 leads of the MI affected ECG signal. Prior informations regarding the pathological characteristics of the specific disease are required for the correct detection of MI using few numbers of ECG leads. Thus, in this paper 12 lead ECG signals have been considered for identification of MI. The two-class classification problem with MI class and healthy individual class is performed using the threshold based classification regulation. Both healthy control and MI affected ECG signals are collected from the PTB diagnostic ECG database. The accuracy, sensitivity and specificity are found to be 99.93%, 99.97% and 99.30% for detection of MI. The proposed method has got the superiority in terms of simplicity of features, small feature dimension and simpler classification rule ensuring faster, accurate and easier MI detection.

Classification of myocardial infarction based on hybrid feature extraction and artificial intelligence tools by adopting tunable-Q wavelet transform (TQWT), variational mode decomposition (VMD) and neural networks.

Cardiovascular diseases (CVD) is the leading cause of human mortality and morbidity around the world, in which myocardial infarction (MI) is a silent condition that irreversibly damages the heart muscles. Currently, electrocardiogram (ECG) is widely used by the clinicians to diagnose MI patients due to its inexpensiveness and non-invasive nature. Pathological alterations provoked by MI cause slow conduction by increasing axial resistance on coupling between cells. This issue may cause abnormal patterns in the dynamics of the tip of the cardiac vector in the ECG signals. However, manual interpretation of the pathological alternations induced by MI is a time-consuming, tedious and subjective task. To overcome such disadvantages, computer-aided diagnosis techniques including signal processing and artificial intelligence tools have been developed. In this study we propose a novel technique for automatic detection of MI based on hybrid feature extraction and artificial intelligence tools. Tunable quality factor (Q-factor) wavelet transform (TQWT), variational mode decomposition (VMD) and phase space reconstruction (PSR) are utilized to extract representative features to form cardiac vectors with synthesis of the standard 12-lead and Frank XYZ leads. They are combined with neural networks to model, identify and detect abnormal patterns in the dynamics of cardiac system caused by MI. First, 12-lead ECG signals are reduced to 3-dimensional VCG signals, which are synthesized with Frank XYZ leads to build a hybrid 4-dimensional cardiac vector. Second, this vector is decomposed into a set of frequency subbands with a number of decomposition levels by using the TQWT method. Third, VMD is employed to decompose the subband of the 4-dimensional cardiac vector into different intrinsic modes, in which the first intrinsic mode contains the majority of the cardiac vector's energy and is considered to be the predominant intrinsic mode. It is selected to construct the reference variable for analysis. Fourth, phase space of the reference variable is reconstructed, in which the properties associated with the nonlinear cardiac system dynamics are preserved. Three-dimensional (3D) PSR together with Euclidean distance (ED) has been utilized to derive features, which demonstrate significant difference in cardiac system dynamics between normal (healthy) and MI cardiac vector signals. Fifth, cardiac system dynamics can be modeled and identified using neural networks, which employ the ED of 3D PSR of the reference variable as the input features. The difference of cardiac system dynamics between healthy control and MI cardiac vector is computed and used for the detection of MI based on a bank of estimators. Finally, data sets, which include conventional 12-lead and Frank XYZ leads ECG signal fragments from 148 patients with MI and 52 healthy controls from PTB diagnostic ECG database, are used for evaluation. By using the 10-fold cross-validation style, the achieved average classification accuracy is reported to be 97.98%. Currently, ST segment evaluation is one of the major and traditional ways for the MI detection. However, there exist weak or even undetectable ST segments in many ECG signals. Since the proposed method does not rely on the information of ST waves, it can serve as a complementary MI detection algorithm in the intensive care unit (ICU) of hospitals to assist the clinicians in confirming their diagnosis. Overall, our results verify that the proposed features may satisfactorily reflect cardiac system dynamics, and are complementary to the existing ECG features for automatic cardiac function analysis.

A dictionary matrix generation based compression and bitwise embedding mechanisms for ECG signal classification

The Electrocardiogram (ECG) signal processing is one of the exciting research areas in recent days. Ensuring security to the patient’s confidential information is a demanding critical task in many healthcare systems. So, the traditional works developed the security mechanisms for embedding the original ECG signal with the image, audio, or video. But, it does not focus on reducing the size of the original message before transmitting it to others. Also, it has significant limitations of inefficient security, increased complexity, and reduced classification accuracy. To rectify this issue, our research proposed the new embedding mechanism to improve the security of patient’s health information. In this system, the original ECG signals compressed at the initial stage by using the proposed Dictionary Matrix Generation (DMG) algorithm. Then, the compressed signals embedded within the cover image by using the Bitwise Embedding (BE) mechanism. At the receiver side, the bedded goal is de-embedded and decompressed by using the DMG and BE algorithms. The features such as spectral and peak values of the signal are extracted for increasing the efficiency of classification. Classification and detection of abnormality present in ECG signal of patient is the most essential part. To achieve this, we proposed the Modified Dynamic Classification (MDC) algorithm based on the features. In this work, the novelty is implemented in the compression, embedding, and classification stages. The proposed system reduces the data loss during transmission, memory storage and time complexity. The overall process evaluated by using PTB diagnostic ECG database. In experiments, the proposed classification technique provides the accuracy of 98.39% and it proved that the proposed method had highest performances than existing methods such as PNN, SVM and RF classification.

Identity Recognition Based on the QRS Complex Dynamics of Electrocardiogram

Biometric identification based on electrocardiogram signals has attracted increasing attention. As the most dominant feature of the electrocardiogram signal, the QRS complex(i.e., the combination of ECG Q, R, and S waves) has been used for identification in some studies. This study aims to investigate the intra-individual stability of the QRS complex dynamics to assess its potential for human identification. The QRS complex dynamics are used as the unique feature to classify the QRS complex, which differs from the time/frequency domain features used in the literature. It is the fundamental feature of the QRS complex and contains the underlying information of the QRS complex. The dynamics of training QRS complexes are extracted and expressed as a constant radial basis function network by using deterministic learning. A set of estimators is constructed to represent the training QRS complexes using constant radial basis function networks. By comparing this set of estimators with the test QRS complex, a set of recognition errors is generated, and the average L 1 norms of the errors are taken as the similarity measure between the dynamics of the training QRS complexes and that of the test QRS complex. Therefore, the test QRS complex can be recognized according to the smallest error principle. The electrocardiogram is classified according to the vote of the test QRS complexes recognition results. A private database and PTB diagnostic ECG database are used to test the proposed method. Experimental results on the private database (PTB database) showed that the average identification accuracy was 96.12% (97.42%) for 5-fold cross-validation based on one-lead electrocardiogram and 99.50% (99.23%) for 2-fold cross-validation based on two-lead electrocardiogram, respectively. These show that the dynamics of the QRS complex are well-differentiated for different individuals.

Electrocardiogram profiling of myocardial infarction history using MLP and HMLP networks

<span>Narrowing of coronary arteries caused by cholesterol deposits deprives heart tissues of oxygen. In prolonged conditions, these will result in myocardium infarction. The presence of damage tissues modifies the normal sinus rhythm and this can be detected using electrocardiogram (ECG). Hence, this paper characterized history of myocardial infarction from survivors using QRS power ratio features from the ECG. Subsequent profiling is performed using multilayered perceptron (MLP) and hybrid multilayered perceptron (HMLP) networks. ECG with history of anterior and inferior infarctions, along with healthy controls is obtained from PTB Diagnostic ECG Database. The signal is initially pre-processed and the power ratio features are extracted for low- and mid-frequency components. The features are then used as input vector to the MLP and HMLP networks. The optimized MLP has attained accuracies of 99.2% for training and 98.0% for testing. Meanwhile, the optimized HMLP managed to achieve accuracies of 99.4% for training and 97.8% for testing. Despite the similarities in network performance, MLP provides a better alternative due to the reduced computational requirements by as much as 30%.</span>

Automated detection and localization system of myocardial infarction in single-beat ECG using Dual-Q TQWT and wavelet packet tensor decomposition

Background and objectiveIt is challenging to conduct real-time identification of myocardial infarction (MI) due to artifact corruption and high dimensionality of multi-lead electrocardiogram (ECG). In the present study, we proposed an automated single-beat MI detection and localization system using dual-Q tunable Q-factor wavelet transformation (Dual-Q TQWT) denoising algorithm. MethodsAfter denoising and segmentation of ECG, a fourth-order wavelet tensor (leads × subbands × samples × beats) was constructed based on the discrete wavelet packet transform (DWPT), to represent the features considering the information of inter-beat, intra-beat, inter-frequency, and inter-lead. To reduce the tensor dimension and preserve the intrinsic information, the multilinear principal component analysis (MPCA) was employed. Afterward, 84 discriminate features were fed into a classifier of bootstrap-aggregated decision trees (Treebagger). A total of 78 healthy and 328 MI (6 types) records including 57557 beats were chosen from PTB diagnostic ECG database for evaluation. ResultsThe validation results demonstrated that our proposed MI detection and localization system embedded with Dual-Q TQWT and wavelet packet tensor decomposition outperformed commonly used discrete wavelet transform (DWT), empirical mode decomposition (EMD) denoising methods and vector-based PCA method. With the Treebagger classifier, we obtained an accuracy of 99.98% in beat level and an accuracy of 97.46% in record level training/testing for MI detection. We also achieved an accuracy of 99.87% in beat level and an accuracy of 90.39% in record level for MI localization. ConclusionAltogether, the automated system brings potential improvement in automated detection and localization of MI in clinical practice.

Deep Learning with a Recurrent Network Structure in the Sequence Modeling of Imbalanced Data for ECG-Rhythm Classifier

The interpretation of Myocardial Infarction (MI) via electrocardiogram (ECG) signal is a challenging task. ECG signals’ morphological view show significant variation in different patients under different physical conditions. Several learning algorithms have been studied to interpret MI. However, the drawback of machine learning is the use of heuristic features with shallow feature learning architectures. To overcome this problem, a deep learning approach is used for learning features automatically, without conventional handcrafted features. This paper presents sequence modeling based on deep learning with recurrent network for ECG-rhythm signal classification. The recurrent network architecture such as a Recurrent Neural Network (RNN) is proposed to automatically interpret MI via ECG signal. The performance of the proposed method is compared to the other recurrent network classifiers such as Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU). The objective is to obtain the best sequence model for ECG signal processing. This paper also aims to study a proper data partitioning ratio for the training and testing sets of imbalanced data. The large imbalanced data are obtained from MI and healthy control of PhysioNet: The PTB Diagnostic ECG Database 15-lead ECG signals. According to the comparison result, the LSTM architecture shows better performance than standard RNN and GRU architecture with identical hyper-parameters. The LSTM architecture also shows better classification compared to standard recurrent networks and GRU with sensitivity, specificity, precision, F1-score, BACC, and MCC is 98.49%, 97.97%, 95.67%, 96.32%, 97.56%, and 95.32%, respectively. Apparently, deep learning with the LSTM technique is a potential method for classifying sequential data that implements time steps in the ECG signal.