| Summary: | Cardiovascular diseases (CVDs), including ischemic heart disease, remain a significant health concern globally, necessitating accurate and efficient diagnostic methods. Previous studies have demonstrated the potential of machine learning algorithms, such as Random Forest (RF) and Convolutional Neural Networks (CNNs), in classifying ECG signals. The RF classifier achieved 93% accuracy in distinguishing ischemic from non-ischemic ECGs, though it required manual feature extraction, which limits its scalability. On the other hand, CNNs have shown remarkable success, with one study achieving 99% accuracy in classifying myocardial infarction using multi-lead ECG signals, and another combining CNN with Autoencoder models to achieve 88.6% accuracy on the ST-T ECG database. Motivated by these findings, this study proposes a deep learning approach utilizing CNNs to detect ischemic activity in ECG readings by treating ECGs as image classification tasks. The developed CNN model demonstrated promising accuracy rates, exceeding 80%, suggesting its potential as a valuable diagnostic tool. However, limitations such as the relatively small dataset and challenges in detecting subtle changes in ECG patterns were acknowledged. Future research should focus on enhancing model robustness, diversifying the training dataset, and expanding the model's applicability to other cardiovascular conditions. © 2024 IEEE.
|
|---|