ECG Feature Extraction for Automatic Classification of Ischemic Events

Abstract

A fundamental part of the analysis of electrocardiographic signals (ECG) lies in the transformation of the data or samples of the signal, with the aim of improving the effectiveness of the processing and the subsequent interpretation of the results obtained from it. Thus, a certain number of features are extracted, which are expected to have high discriminatory properties. In this way, the next step of the analysis must be the construction of a vector of features, which must be a set of descriptors that completely transmit the essence of the signal being studied. These features can focus on different aspects of the signal, for example, durations of intervals or segments, amplitudes of different wave peaks, deviations with respect to the isoelectric line, and areas under curves, among others. In this paper we present a method that includes the detection of temporal events of the ECG signal, such as ST segment, T wave, QRS complex and QT interval, several features are extracted using these events. The features are validated with the K-means algorithm and the feature vector is used as input in the ischemia classifier.