A Screening System for Myocardial Ischemia Based on Pathophysiological Vectorcardiogram

Ischemic heart disease is a leading cause of death both in developed and developing countries. In this paper, efforts to detect myocardial ischemia are focused on developing a screening system through vectorcardiogram (VCG). Existing systems manifest a low accuracy, as a result of concentrating more on ST segment and statistical features of VCG. In our work, several novel strategies have been introduced, as application of spatial heartbeat amplitude is introduced in preprocessing and pathophysiological VCG parameters are employed in feature extraction, thereby, establishing the screening system with the classification model of random forest. Experiments have been carried out on 132 individuals from two databases. The accuracy achieves 98.07% when applying 10-fold cross validation, and maintain the same level of sensitivity (95.63%) and specificity (99.04%) when testing in simulated clinical situation.