Deep Learning Framework for Cardio Vascular Disease Prediction Using ECG Images

Globally, cardiovascular illnesses are the leading cause of death. More lives can be saved if the disease is identified and predicted earlier. A highly skilled clinician is needed to detect diseases, even though the healthcare system uses many technologies such as cardiac MRIs, Echocardiograms and blood tests are used to detect heart diseases. This manual process could produce results that are not accurate. DL and ML algorithms are used for automatic heart disease prediction in an effort to lower medical errors. The diagnostic process can be accelerated by using intelligent systems to support medical professionals. The ECG image dataset was utilized to predict the 4 primary cardiac abnormalities such as Abnormal Heart Beat, Myocardial Infarction, history of Myocardial Infarction and normal person classes. This was achieved by using DL techniques. In this project, a hybrid Inception-V3+VGG-19 model is used for CVD prediction from the ECG image dataset. Inception-V3 provides efficient feature extraction, while VGG-19 captures rich spatial information from ECG Images. By combining the strengths of both architectures, potentially achieve better performance. According to the results, the proposed model outperforms earlier studies in terms of accuracy (96.0%), precision (90.90%), recall (100%), and F1-score (95.23%).