Hypertrophic cardiomyopathy (HCM) and hypertensive heart disease (HHD) diagnosis using echocardiography and electrocardiography

Abstract

ABSTRACT Echocardiography and electrocardiogram (ECG) are the primary tools used by cardiologists to diagnose cardiovascular heart diseases. Contrary to the critical role of combining the echocardiography and ECG information in clinical examinations, to our knowledge, no study has considered this to classify heart diseases. Left ventricular hypertrophy (LVH) is caused by a variety of origins such as hypertrophic cardiomyopathy (HCM) and hypertensive heart disease (HHD). Differentiating HCM and HHD is challenging . We proposed an HCM and HHD patient classifier to use ECG and echocardiography information. Longitudinal strain and strain rate from echocardiography frames and amplitude and temporal features from ECG signals, are extracted. To eliminate incompetent features, Fisher’s discrimination ratio (FDR), information gain, and Relief-F weights are used. Finally, support vector machine (SVM) and K-nearest neighbours classifiers are used to classify the normal, HCM, and HHD subjects. The results on 30 subjects show that the best classification refers to SVM classifier using five selected features from ECG and echocardiography information using FDR. The precision, sensitivity, and F-measure are 97.62, 93.33 and 95.43%, respectively. According to the results, the combination of echocardiography and ECG information leads to diagnosis improvement compared to the classification based on separated information of ECG and echocardiography.