Year: 2026
Type: conference-paper
Abstract: Myocardial Infarction (MI) remains a leading cause of global mortality, where timely detection through Electrocardiogram (ECG) analysis is critical for effective intervention. Conventional diagnostic techniques and early machine learning methods often fall short in handling the complex and variable nature of ECG signals, particularly in accurately identifying MI cases. This study presents a deep learning-based ECG classification framework utilizing EfficientNet-B0 enhanced with a Multi-Scale Attention (MSA) mechanism to improve the extraction of both localized and global features from ECG images. The model classifies inputs into four clinically significant categories. A lightweight and user-friendly interface enables image upload and real-time diagnosis, supported by integrated explainable AI methods like Grad-CAM and Integrated Gradients to enhance clinical interpretability. Evaluation results demonstrate a superior classification accuracy of 96%, outperforming baseline convolutional models and attention-augmented DenseNet architectures. The findings highlight the potential of attention-enhanced deep networks combined with interpretability tools in supporting accurate and transparent cardiac diagnostics.
Source: Procedia Computer Science
Institution: Amrita Vishwa Vidyapeetham
Language: English
Topic: ECG Monitoring and Analysis
Subfield: Cardiology and Cardiovascular Medicine
Field: Medicine
Domain: Health Sciences
Open Access status: diamond
OpenAlex ID: W7167742820