Cardiac Ischemia Detection Using Parameters Extracted from the Intrinsic Mode Functions

Abstract

Cardiac ischemia is the main cause of death in the world, thus the importance of prevention and early detection of these events. Traditionally, ischemia is detected by analyzing the alteration of the ST level in the electrocardiogram (ECG). In this study, we propose two new parameters extracted from the ECG to improve the cardiac ischemia detection. For this, the signal was decomposed using the Empirical Mode Decomposition, and the Intrinsic Mode Functions related to the frequency band of the ST level were selected. From these modes, two parameters were obtained: the Hjorth Activity and the frequency amplitude, using the Hilbert Transform. With these parameters, two analyses were done. First, the parameters obtained during normal periods were compared with those obtained during ischemic events. Second, a temporal series was obtained with both parameters, where the detection was done using an adaptative threshold. Results were obtained using all the patients with MLIII lead of the European ST-T Database. Parameters differed significantly across ischemic and non ischemic episodes, obtaining a sensitivity and positive predictive value of 88%, after removing noisy records. Also, a multi-lead detection was performed in patients with MLIII and V4 leads. The sensitivity and positive predictive value obtained were 92% and 80%, respectively.