Empirical mode decomposition template matched filter for detection and estimation of T-wave alternans

T-wave alternans (TWA) is a well-recognized marker for malignant cardiac arrhythmia that may lead to sudden cardiac death. TWA detection and estimation becomes a challenge under the presence of process noises, transient outliers and physiological artifacts. We propose an empirical mode decomposition (EMD) based template matched filter approach for TWA analysis, targeting detection of distributed alternans energy rather than peak amplitude. EMD is performed to construct the template for matched filter which is central to alternans estimation and detection. The template is dynamically generated through decomposition into intrinsic mode functions (IMFs) and subsequent reconstruction. Notable improvement is achieved through correlation based reconstruction that shows a performance gain of > 2dB over other approaches including the spectral method (SM).