A unified procedure for detecting, quantifying, and validating electrocardiogram T-wave alternans

The precision of T-wave alternans (TWA) quantification depends certainly upon the way we choose to align T-waves and to get feedbacks from the electrocardiogram (ECG) quality. Quantifying the ECG TWA based on assigning automatically the required number of T-waves along with applying a proper T-wave alignment approach is the purpose of this paper. The structure of the proposed method mainly consists of seven sections: preprocessing, ECG events detection-delineation, alignment of cycles, T-wave template extraction, T-wave delineation, T-wave left- and right-lobes synchronization, and T-wave alternans quantification-detection by getting feedback from ECG quality value. The proposed method is examined in two ways. First, some artificially generated ECGs with predefined TWA patterns and qualities are analyzed in order to regulate the parameters of the method to achieve the maximum performance. Finally, the method is applied to PhysioNet/Computing in Cardiology Challenge 2008 database. In this stage, the achieved accuracy is about 91.0 %, which shows marginal improvement in the area of TWA quantification.