A Hybrid Model for Diagnosis Cardiovascular Disease Using Clinical Features, ECG and MRI

According to the WHO, cardiovascular diseases (CVD) is the biggest cause of death worldwide; over 17 million people died from CVD in the year 2019, which is a major health concerns for public. Effective prevention and treatment of cardiovascular diseases depend on early detection and in this regard, artificial intelligence techniques have demonstrated great promise. In recent years, MRI images have emerged as a powerful non-invasive modality for evaluating heart problems and its functions so this research study uses clinical features, ECG and MRI images to accurately diagnose diseases. This hybrid model utilize a combination of demographic, clinical, and lifestyle factors and depending on the severity of the disease, ECG and MRI scanning is advised. First CNN models is trained using ECG signals and it predicts the class of disease. The second CNN model predicts the classes of MRI data. This hybrid model uses the results of ensemble, CNN on ECG data, and CNN on MRI images to predict the likelihood of disease in the region of interest. Auto encoders are also used to remove the noise from MRI images. The aggregate accuracy of the three sub models is 93.4% which is superior performance than existing methods using clinical features with ECG and MRI data.