Cardio-Detect: A Hybrid ML-DL Technique for Identifying Cardiovascular Diseases in ECG Images

Cardiovascular diseases (CVDs) remain one of the leading causes of mortality worldwide. Early and accurate detection is critical to reducing the risk of fatal outcomes. Electrocardiogram (ECG) analysis is a fundamental diagnostic tool used to detect abnormalities associated with myocardial infarction (MI), arrhythmias, and other cardiac conditions. However, manual interpretation of ECGs is time-consuming, susceptible to human error, and requires specialized expertise, highlighting the need for automated diagnostic methods. While deep neural networks (DNNs) offer high accuracy, they often require extensive computational resources, which can hinder real-time deployment. On the other hand, traditional machine learning (ML) approaches struggle to manage the complex patterns inherent in ECG image data.Our model is trained on a diverse dataset comprising ECG images from four categories: patients with myocardial infarction, individuals with a history of MI, those with arrhythmias, and subjects with normal heart rhythms. This comprehensive dataset ensures broad representation of various cardiac conditions. Feature extraction is performed using EfficientNetB0, a lightweight yet powerful convolutional neural network. The model achieves a training accuracy of 91.26%, with improved generalization capabilities attributed to effective preprocessing and data augmentation strategies. By leveraging the efficient feature extraction capabilities of EfficientNetB0 and coupling it with a simple yet effective machine-learned classifier, our model offers a scalable and computationally economical solution for ECG image analysis. This approach facilitates accurate, timely, and accessible cardiovascular disease detection.