DETECTION OF CARDIOVASCULAR DISEASES IN ECG IMAGES USING MACHINE LEARNING AND DEEP LEARNING METHODS

Abstract

Cardiovascular diseases (heart diseases) are the essential driver of mortality worldwide. The sooner they can be expected and classified, the more noteworthy the quantity of lives that can be saved. An electrocardiogram (ECG) is a pervasive, financially savvy, and harmless instrument for evaluating the heart's electrical movement and is used for the distinguishing proof of cardiovascular diseases. This paper utilized deep learning ways to deal with four critical cardiovascular diseases: deviant heartbeat, myocardial localized necrosis, history of myocardial dead tissue, and ordinary people, using a public ECG pictures dataset of heart patients. The exchange learning approach was inspected using the low-scale pre-trained deep neural networks Squeeze Net and Alex Net. A novel convolutional neural network (CNN) engineering was presented for the expectation of heart irregularities. Third, the recently portrayed pre-trained models and our recommended CNN model filled in as element extraction apparatuses for standard machine learning algorithms, including support vector machine, K-nearest neighbours, decision tree, random forest, and Naïve Bayes. The exploratory outcomes demonstrate that the presentation measurements of the proposed CNN model outperform those of existing works, accomplishing 998.23% accuracy, 98.22% recall, 98.31% precision, and 98.21% F1 score. Moreover, the recommended CNN model achieves an ideal score of 99.79% for highlight extraction while utilising the NB algorithm.