Identification of myocardial infarction (MI) using spatio-temporal heart dynamics

摘要: Cardiovascular disorders, such as myocardial infarction (MI) are the leading causes of mortality in world. This paper presents an approach that uses novel spatio-temporal patterns vectorcardiogram (VCG) signals for identification various types MI. In contrast to traditional electrocardiogram (ECG) approaches, 3D cardiac VCG signal is partitioned into 8 octants localized analysis heart's electrical activities. The proposed method was tested using PhysioNet PTB database 368 MIs and 80 healthy control (HC) recordings, each which includes 12-lead ECG 3-lead VCG. Significant differences found spatial distribution between MI HC groups. Furthermore, classification regression tree (CART) used demonstrate octant features can distinguish from HCs with a sensitivity (accuracy identification) 97.28% specificity 95.00%, promising compared previously reported results other databases. indicate present provides effective way monitoring, post-processing, interpretation data, hopefully impact current diagnostic practice.