T-wave altemans detection and estimation through Cosine Modulated Filter Bank and Modified Media Average

T-wave alternane (TWA) is considered as a sudden cardiac death susceptibility index. TWA has been shown to presage lethal ventricular arrhythmias in patients with structural heart disease. TWA appears on the surface of the ECG signal as a fluctuation of ventricular repolarization every two beats. This alternant wave has an amplitude on the microvolts order, being imperceptible the naked eye which makes difficult its detection. In this work, we propose the TWA detection and estimation method through the Cosine Modulated Filter Bank (CMFB) as preprocessing stage to minimize the noise present in ECG signal during routine exercise testing. A TWA detection and estimation by using the Modified Media Average (MMA) method has been established. For evaluation process, the ECG records from Physionet databases were used. The combined CMFB and MMA presents a SNR improvement of 5 dB and 7 dB for probability detection (PD) by using the thresholds of 47 μΎ and 60 μΎ respectively than by using only the MMA method. The PD and the mean absolute error (MAE) metrics have been used.