Automatic Cardiac Arrhythmia Classification Using Residual Network Combined With Long Short-Term Memory

Abstract

Diagnosis and classification of arrhythmia, which is associated with abnormal electrical activities in the heart, are critical for clinical treatments. Previous studies focused on the diagnosis of atrial fibrillation, which is the most common arrhythmia in adults. The classification performance achieved by studies on other arrhythmia types is not satisfactory for clinical use owing to the small number of classes (minority classes). In this study, we propose a novel framework for automatic classification that combines a residual network with a squeeze and excitation block, and a bidirectional long short-term memory. 8-class, 4-class, and 2-class performances were evaluated on the MIT-BIH arrhythmia database (MITDB), MIT-BIH atrial fibrillation database (AFDB), and PhysioNet/Computing in the cardiology challenge 2017 database (CinC DB), respectively, and they were superior to performance achieved by conventional methods. In addition, the class-wise F1-score in the minority classes was higher than those of the methods adopted in existing studies. To measure the generalization ability of the proposed framework, AFDB and CinC DB were tested using a MITDBtrained model, and superior performance was achieved compared with ShallowConvNet and DeepConvNet. We performed a crosssubject experiment using AFDB and obtained a statistically higher performance using the proposed method compared with typical machine learning methods. The proposed framework can enable the direct diagnosis of arrhythmia types in clinical trials based on the accurate detection of the minority class.

A Convolutional Neural Network-Based Maximum Power Point Voltage Forecasting Method for Pavement PV Array

“…Ideally, a complex model can obtain a better fit but may encounter vanishing gradient problems. ResNet [29] forms a residual block by adding shortcut connections, as shown in Fig. 4, which ensures that the input data are fed to distant layers.…”

“…To check the prediction effect of the maximum power point voltage prediction model, the proposed CNN prediction method is compared with ResNet [29] and MLP [30] in the same dataset, and the results are shown in Fig. 11.…”

Integrating deep learning architectures for improved arrhythmia detection in electrocardiogram signals

“…An explicit interpretation of the convolutional layer as a Finite Layer Impulse Response Filter (FIR) and exploiting this interpretation to develop a STFT-based 1D convolutional layer (Conv1D) to extract the spectrum from the input ECG signal entry is also provided. Yun Kwan Kim.. et.al [5] In this study in , a new automatic classification framework combining residual network with compression and excitation block as well as two-dimensional long-term memory is proposed. Performance at levels eight, four, and two was evaluated on the MIT-BIH Arrhythmia Database (MITDB), the MIT-BIH Atrial Fibrillation Database (AFDB), and the PhysioNet/Internal Calculator.…”

A Study on the Residential Environment Polarization and Characteristics in Seoul

“…The manual inspection and interpretation of ECGs are time-consuming and burdensome processes, as ECGs they are recorded over a long period of time. To address these challenges, several studies have developed CDSSs using automated algorithms to achieve enhanced arrhythmia detection and diagnostic accuracy [8], [9].…”

“…Many studies have utilized machine learning methods using preprocessing, feature extraction, and feature segmentation techniques for automated arrhythmia classification [10]. To develop robust Holter monitoring systems for arrhythmia classification, many studies have used deep learning approaches [8], [10], leading to noteworthy performance improvements. For instance, Acharya et al [11] proposed a convolutional neural network (CNN) including eleven layers to classify four classes and achieved an F1 score of 83.00%.…”