Empirical Mode Decomposition on T-Wave Alternans Detection

T-Wave Alternans (TWA) is a known marker for fatal cardiac arrhythmias which may lead to Sudden Cardiac Death (SCD). Its detection and estimation becomes a challenge under the presence of various types of noise including electrode movements and muscle artifacts. Various pre-processing techniques have been suggested by authors to counter for the issue of noise. One of the relatively new denoising techniques, Empirical Mode Decomposition (EMD) has also been used for extricating the trend of ST-T wave segments. In this paper, we have carried out a comparison of the effect of using EMD for denoising the ST-T complexes before applying the state of the art TWA analysis algorithms i.e. Spectral Method (SM), Correlation Method (CM), Modified Moving Average Method (MMAM) and Median Matched Filter (MMF) method. Application of EMD has noticeably improved detection and estimation accuracy. Application of EMD has improved detection performance of MMF by 50%. Estimator bias and Mean Square Error (MSE) are reduced by more than 30% for MMAM, SM and MMF with EMD. The improvement is most pronounced in MMF and is least in case of CM, while compared with results of directly applying the algorithms without denoising by EMD.