— Cardiovascular diseases (CVDs) continue to be the leading cause of death globally, creating an urgent need for faster and more accessible diagnostic tools. Although electrocardiograms (ECGs) play a crucial role in detecting cardiac abnormalities, their interpretation often requires specialized expertise and can be time-consuming. To address these challenges, this paper proposes a robust ensemble machine learning framework for multi-class cardiovascular disease detection using ECG image-derived signals. The system processes standard 12-lead ECG images by performing grayscale conversion, noise and gridline removal, and contour-based lead extraction to obtain one-dimensional cardiac signals. These signals are normalized and refined using Principal Component Analysis (PCA) to improve classification performance. Multiple machine learning algorithms, including K-Nearest Neighbors, Logistic Regression, Support Vector Machine, Random Forest, Naïve Bayes, and XGBoost, are trained and optimized, and their outputs are combined using a stacking-based ensemble approach.
The model classifies ECG data into four categories: Normal, Abnormal Heartbeat, Myocardial Infarction, and History of Myocardial Infarction. The final system is deployed through a Streamlit-based web application, enabling rapid and user-friendly cardiac screening. The results demonstrate the potential of the proposed approach in supporting efficient and scalable heart disease detection.