A multilead approach to T-wave alternans detection combining principal component analysis and the Laplacian likelihood ratio method

Abstract

Several methods have been proposed to automatically detect and estimate T-wave alternans (TWA) in the ECG, all of them operating on a single-lead basis. In this study a multilead detection scheme is proposed. It combines principal component analysis with the Laplacian likelihood ratio method to improve the detection results. A simulation is carried out to evaluate the effect of different types of simulated and physiological noise, and to assess the performance of this detection scheme. According to simulation results, the proposed multilead approach can detect alternans with a SNR 30 dB lower than the single-lead approach, showing a remarkable improvement of the sensitivity to low-level alternans.