An algorithm for automated detection of ischemic ECG beats using support vector machines

Cardiac beat classification is a key process in the detection of myocardial ischemia episodes in the electrocardiogram (ECG) signal. In this paper, we have developed a new method based on support vector machines for detection of ischemic ECG beats. The proposed method consists of a preprocessing stage for QRS detection, baseline wandering removal, noise suppression and ST segment extraction. In the next stage, the ST segment pattern is down-sampled and subtracted from the down-sampled normal template. In the third stage, the resulted patterns are used for training a support vector machine and ischemic beats are detected. To evaluate the algorithm, a cardiac beat data set is constructed using a number of recordings of the ESC ST-T database. The obtained sensitivity and positive predictivity were 92.13% and 90.34%, respectively. The proposed methodology presents better results than other approaches.