Morphological Based Method for Automated Extraction and Classification of ECG ST-T Wave

Abstract

The wave starting from the beginning of the S wave until the end of the T wave is known as ST-T. ST-T wave extraction and classification is a very important technique in the diagnosis of myocardial ischemia. The myocardial ischemia which is also called cardiac ischemia can cause damage to the heart muscle. A sudden, severe blockage of a coronary artery may lead to a heart attack or other severe complications, and may also cause serious abnormal heart rhythms which will be reflected in the electrocardiogram trace. This paper aims to automate the real-time technique detection of ST-T waves that help in the diagnosis of myocardial ischemia and to classify the patient state. The proposed method uses the ECG wave morphological features that have been extracted using the detrended cumulative area, which is used to detect the ST-T wave. The proposed technique was tested and validated and it revealed promising results. The proposed method scored a sensitivity of 90.13% for K-mean Clustering and 96.3% for the SVM classifier for the ST interval detection. The method was tested on the European ST-T Database.