Wavelet Analysis of Electrocardiogram for Assessment of Risk for Life-Threatening Cardiac Ailments

Abstract

Research Team Dr. Joseph .P. Noonan, Professor, EECS Department, Tufts University Sumugam Balachandrang (MS), Research Associate, EECS Department Doctor. Paul Wang, New England Medical Center Work Summary This work is concerned with formulating an approach to analyze Electrocardiogram for assessment of risk for life threatening cardiac ailments. A pioneering Non-Invasive method attempting to detect multiple cardiac ailments is presented. A flexible software analysis tool has been developed which given an input rhythm strip of conventionally available high-resolution surface electrocardiogram (ECG) performs the wavelet analysis. The tool provides the option of three alignment techniques, namely Alignment by Maximum correlation with the chosen template, Alignment by the Peak of QRS segment, and Alignment by Maximum slope of RS segment for Cycle Averaging .The aligned ECG waveform is then cycle averaged to improve the signal to noise ratio and then transformed to the wavelet domain using a fourth order Daubechie wavelet (db4). Additionally the module offers three High pass filtering options, which can be applied to the averaged waveform prior to wavelet transformation to eliminate low frequencies from the analysis. The important issue of providing a user interface to choose the options to view the Wavelet Transformed ECG is addressed and a comparison of options is empirically evaluated.