Ventricular tachycardia risk detection using wavelet decomposition of the signal averaged ECG

This study proposes a method for the early detection of VT risk by wavelet decomposition of the signal-averaged high-resolution ECG (SAECG). A quadratic-spline wavelet was used to decompose the SAECG from the three leads. Decompositions from VT patients showed a higher number of irregularities than those from normal subjects. This content was quantified using a normalized shape parameter (NSP): the integral of the absolute value of the decomposed SAECG normalized by the energy. The method is tested on 73 healthy subjects and 59 VT patients. NSP´s from the first four decompositions (DI-D4) were significantly different between the two groups. NSP values for D1 in leads X and Y gave the best discrimination results. The value of the NSP´s, combined with Simson´s QRSd, was used to classify patients in both groups. Results showed that the NSP contains new independent information that may improve the classifications obtained solely by classical methods