Analysis of ST/QT dynamics using independent component analysis

Abstract

Heart diseases such as arrhythmia and myocardial infarction are the leading causes of death. If these cardiac-related deceases are detected at early stage then the people can be saved from death. With our proposed method, it is possible to estimate the occurrence of these diseases easily at early stage. Generally the variations in QT-interval are differentiated by the occurrence of premature activation (PA) beats in electrocardiogram (ECG). Similarly the elevation or depression of ST-segments is the indication of ischemia. We used the ratio of ST-segment to QT-interval as an index to specify the probability of ischemia and cardiac injury. The ΔST/QT can be used as the index of the severity of the disease. The aim of this paper is to investigate the instability in the ventricular repolarization process by detecting ST/QT. In our proposed method, four 1-min-long ECG signals of same subject were taken and applied independent component analysis (ICA) to extract the cleaned ECG from various artifacts and non-Gaussian noise background. The performance of proposed method can also be measured using SNR. The ST/QT slope signifies the peak rate of ST-segment change with heart rate, whereas The ΔST/QT index represents the average change of ST-segment depression with QT-interval. In our method, we used a well-known BSS algorithm, JADE algorithm, to obtain ICs. The JADE is able to separate ECM artifact from ECG activity in component domain.