Extraction, Comparison, and Evaluation of Electrocardiography and Vectorcardiography Signals Features

Cardiovascular disease is the most common disease of the century and is considered as a major cause of mortality and heart failure in industrial and semi-industrial societies. Using electrical signals produced by the heart muscle have an important role in the recognition and diagnosis of heart diseases.
In this study, the signal recorded via vectorcardiography using an electrode arrangement called Frank, and electrocardiography were performed; a limited number of signals available in the database of the National Metrology Institute of Germany (Physikalisch-Technische Bundesanstalt or PTB) were also used. At the end, the data in each field vectorcardiography and electrocardiographic were assessed and compared. In order to make better use of time and optimize signal analysis, feature extraction operation was performed in the wavelet domain; and then, reducing the characteristics and classification were performed using support vector machine technique.
Findings: There were the accuracy of 83.18% and the validity of 99.06% in vectorcardiography leads and the accuracy of 75.44% and the validity of 97.16% in electrocardiographic leads.
Based on support vector machine classification system, the properties of the Frank system leads tended to better results than conventional 12-leads electrocardiogram (ECG).