Aug 2026· International journal of computer information systems and industrial management applications· Vol 18, pp. 1005-1023· 0 citations
TL;DR
Present a Wavelet-CNN Hybrid (WCNN-H) framework for automatic multi-class classification of cardiovascular diseases from ECG images, which has a high potential for intelligent computer aided cardiovascular screening applications.
Abstract
CVDs are still a significant health problem worldwide and there is a need for an accurate and non-invasive diagnostic system for early clinical intervention. Automatic interpretation of electrocardiogram (ECG) images is difficult due to background interference (grid lines, waveform distortion and scanning artifacts). To overcome these issues, in this work, Present a Wavelet-CNN Hybrid (WCNN-H) framework for automatic multi-class classification of cardiovascular diseases from ECG images. The proposed method is a combination of two dimensional Discrete Wavelet Transform (DWT) with Daubechies (db2) mother wavelet for noise suppression and waveform enhancement and deep residual convolutional neural network for hierarchic feature extraction and classification. The framework was evaluated on the ECG Images Dataset of Cardiac Patients with 928 ECG image records in four diagnostic classes. The experiment achieved an overall classification accuracy of 98.10% using five-fold cross-validation, and precision, recall and F1-score for all classes were above 97%. The proposed framework has a high potential for intelligent computer aided cardiovascular screening applications.
An accurate and effective heart disease diagnosis through the electrocardiogram (ECG) analysis remains the crucial issue in clinical practice due to complexity and nonlinearity of ECG patterns, morphological variability, and noise interference. This study proposes a hybrid framework that combines a convolutional neural network (CNN) with a support vector machine (SVM) for ECG signal classification. In the proposed hybrid model, feature extraction was done with CNN. Then, a multi-verse optimizer was used for feature selection. The classifier of the hybrid model is an SVM. The proposed hybrid model is trained and evaluated on the standard publicly available QT database (QTDB). Spectrogram and scalogram images generated from individual ECG beats were used as input to the proposed model. The 512 features obtained with spectrogram and scalogram images decreased to 259 features. The same dataset was used to determine the classification accuracy, which was 99.29% for scalogram-based images and 92.12% for spectrogram-based images. This notable enhancement shows that, in comparison to spectrograms, scalogram representations offer better distinguish features for classification of ECG signals. Moreover, a proposed method outperformed existing techniques. A paired t-test confirmed that proposed method significantly improved accuracy compared with CNN+SVM (p < 0.0001).
Shelan K. Ahmed, Serwan A. Mohammed, Ahmed Kh. Mohammed· International Journal of Onl...· 0 citations
Automated electrocardiogram (ECG) analysis can support the screening of cardiac rhythm abnormalities when standardized preprocessing and classification are applied. This work presents an image-based ECG arrhythmia classification framework for six rhythm categories: Left Bundle Branch Block (LBBB), Normal, Premature Atrial Contraction (PAC), Premature Ventricular Contractions (PVC), Right Bundle Branch Block (RBBB), and Ventricular Fibrillation. The dataset contains 22,166 ECG waveform images divided into training and test subsets. Images are resized to 64 × 64 pixels and intensity-scaled, while shear, zoom, and horizontal-flip augmentation are applied to the training subset. A sequential CNN comprising three convolutionpooling stages, a 128-unit dense layer, dropout, and a six-unit softmax output performs multiclass classification. The trained network is stored in HDF5 and native Keras formats and deployed through a Flask-based image-upload interface. Experimental evaluation includes epoch-wise learning behavior and three interface-level classification cases. A Normal ECG is classified as Normal and mapped to a healthy-rhythm status, a Ventricular Fibrillation input is classified as Ventricular Fibrillation and mapped to an abnormal-rhythm status, and an LBBB input is classified as PVC and mapped to an abnormal-rhythm status. The results demonstrate the complete inference pipeline from ECG image input to multiclass prediction and user-facing rhythm interpretation, while also revealing class-level confusion among abnormal rhythms.
Akarapu Anuhya, M. Raju, Sirisha Veluri· International Journal of AI...· 0 citations
This paper reviews machine learning techniques for electrocardiogram (ECG)-based classification of cardiac arrhythmias as effective alternatives to conventional diagnostic approaches. Cardiac arrhythmia remains one of the most prevalent cardiovascular conditions worldwide, and ECG signals play a vital role in the early detection and assessment of heart-related abnormalities. However, traditional diagnostic methods are often limited by the shortage of medical experts, the complexity of ECG interpretation, and the similarity of symptoms across different cardiac conditions. This review focuses on key stages of ECG-based machine learning classification, including signal preprocessing, feature extraction, algorithmic analysis, and performance evaluation. Twenty-five research studies are examined, with particular attention to commonly applied classification methods, including Support Vector Machine, Random Forest, K-Nearest Neighbor, Artificial Neural Network, and Convolutional Neural Network. The review highlights the strengths and limitations of these approaches and compares their reported performance in arrhythmia classification.
Rawaa R. Rfys· Journal of Intelligent Syste...· 0 citations
The healthcare industry has undergone substantial growth over the past few years. Despite the development of medical technology, the diagnosis of cardiac arrhythmia is still a significant challenge for modern diagnostic methods for correct and efficient classification of electrocardiogram (ECG) signals, especially in cases of class imbalance and the complexity of the arrhythmia. This study proposes a novel deep learning model where the class imbalance problem is addressed using the Synthetic Minority Oversampling Technique (SMOTE) and the effect of signal resolution on the classification is investigated using Cubic Spline (CS) interpolation and Lightweight Deep Neural Network (CSLWDNN). Both the MIT-BIH and the INCART datasets were preprocessed and resampled with CS interpolation with several control point configurations. The Spline-187 configuration performed best on MIT-BIH data, with an F1-score of 97.35% and the accuracy of 97.35%. The model performance for INCART dataset is an F1-score of 98.56% and an accuracy of 98.27% at 20 control points. Comparative experiments showed that CS interpolation has better class-wise recall and smoothness of the signal with various input representations. The outcomes indicate that SMOTE balancing coupled with spline-based preprocessing is a valuable scientific and practical approach to build reliable, scalable, and clinically applicable ECG classification systems.
Hussein Qahtan Khalaf, A. Maolood, Nor Shahida Mohd Shah· Engineering and Technology J...· 0 citations
Cardiac arrhythmia, a common sign of cardiovascular disease, is a leading cause of global mortality and morbidity. Timely and accurate detection of electrocardiogram (ECG) cardiac arrhythmia is essential for effective clinical intervention. Here, we present an explainable hybrid deep learning framework for automated ECG cardiac arrhythmia classification. The proposed model integrates one-dimensional convolutional neural network (1D-CNN) for local morphological features, a gated recurrent unit (GRU) to extract temporal dependencies in sequential ECG signals, and the channel attention technique to emphasize clinically important patterns. In addition, a gradient-weighted class activation mapping (Grad-CAM) module is integrated to enhance interpretability by highlighting complex portions of the ECG signals that influence model decision making. The proposed framework is evaluated on three benchmark datasets: Massachusetts Institute of Technology–Beth Israel Hospital (MIT-BIH) Arrhythmia, St. Petersburg INCART, and the MIT-BIH Supraventricular Arrhythmia Database (SVDB). The model achieves classification accuracy values of 99.69%, 99.73%, and 98.77% along with macro-F1 scores of 95.75%, 97.21%, and 94.58% and specificity values of 99.53%, 99.54%, and 98.63%, respectively. The experimental results demonstrated performance improvement of approximately 2%–5% over recent state-of-the-art techniques in terms of key performance metrics. These findings indicate that the proposed framework effectively integrates local morphological feature extraction with temporal modeling, providing a robust and scalable solution for ECG arrhythmia classification. Moreover, its computational efficiency and its use of single-lead ECG signals make it suitable for real-time deployment in wearable devices, remote monitoring systems, and resource-limited clinical settings.
S. Sundaramoorthy, Govardhan Karunanidhi· Frontiers in Physiology· 0 citations
Aim: Valvular heart disease (VHD) is an important factor in cardiovascular mortality. While early diagnosis is essential, existing methods often lack accuracy, leading to misdiagnosis. This study explores the use of deep learning (DL) for multi-class classification of VHD using phonocardiograph (PCG) signals. Methods: The dataset includes normal PCG heart signals and nine VHD classes such as severe and mild aortic stenosis, mild and moderate mitral stenosis, mild, moderate, and severe mitral regurgitation, and moderate and severe tricuspid regurgitation. A novel convolutional neural network (CNN) is developed to classify these ten classes, demonstrating high accuracy even in noisy data acquisition. Unlike traditional machine learning, which relies on handcrafted features such as spectral, wavelet, and mel-frequency cepstral coefficients, the CNN model automatically learns patterns from raw signals, thereby reducing the need for manual feature engineering. The model is further validated using real-time PCG data from heart clinics diagnosed by heart specialists. Results: It achieves excellent performance with 99.81% accuracy, 99.84% precision, 99.82% recall, 99.85% F1-score, and an area under the curve of the receiver operating characteristic of 1.0. Conclusions: This DL model eliminates segmentation and manual pre-processing, enabling early, accurate VHD detection while outperforming existing methods and supporting clinical decisions and patient care.
Sukant Behera, I. S. Misra, K. Siddiqui· Exploration of Cardiology· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
