Microvolt T-wave Alternans Analysis using

T-wave alternans (TWA) is a marker of cardiac instability and high risk of sudden cardiac death. In this paper we propose a new approach for the TWA detection. For this purpose Lyapunov spectrum was calculated from T-wave time series. To evaluate the detector performance, simulated T-waves based on the real ECG signals were used. Detected and simulated episodes were compared, in terms of sensitivity and positive predictivity. The results show that this method can reliably detect T-wave alternans episodes. Microvolt T-wave alternans is a marker of cardiac instability and high risk of sudden cardiac death (SCD) that is happened in beat to beat ECG signals. SCD due to ventricular arrhythmias is one of the leading causes of cardiac mortality in the world. The most significant method for primary and secondary prevention of SCD is implantable cardiac defibrillator (ICD). ICDs are used for patients with low ejection fraction (EF) that caused reduction of cardiac mortality. However by using EF as risk stratification indicator, the number of treats in comparison to preventions is large and it decreases cost effectiveness of this method. On the other hand ICD implantation is invasive and has side effects on patients. So to reduce number of treats and these side effects, a better classification of patients for ICD implantation is needed. Microvolt T-wave alternans seems to be a good marker of risk stratification. (1) Since 1981, different methods have been proposed for automatic TWA analysis. We mention the main methods introduced in the literature. Spectral method was proposed by Smith et al based on 0.5 cycles-per-beat fluctuation in the beat-to-beat measured T wave energy (2). Complex demodulation method, correlation method, KL transform, Poincare mapping and moving average method are other methods explored respectively in (3-7). In this paper we proposed a new method for detection of T wave episodes in ECG signals. This method is based on Lyapunov exponents. The spectrum of Lyapunov exponents provides a quantitative measure of the sensitivity to initial conditions and is the most useful dynamical diagnostic for chaotic systems (13). In section II we described the materials for methodological study of T wave alternans analysis using simulated episodes. In section III preprocessing stage, Lyapunov exponents and validation methods are described briefly. In section VI the results are presented. The results show that this method can reliably detect T-wave alternans episodes.