Multicnannel selective linear prediction spectrum analysis for detection of cardiac late potentials

In this paper we propose a new technique for the frequency analysis of the signal averaged electrocardiogram (SAECG), based on a parametric model. A part of the spectrua of interest is characterised by a selective linear prediction (SLP)Bodel. A aultichannel extension to the SLP Method is proposed to analyze all the three orthogonal lead data at one stroke. Consideration of three channel SAECG as a vector process results in matrix predictor coefficients and spectral aatrix containing auto and cross spectra of individual channels. Vector sub of the area under the eodel auto spectra is used in the detection of Late Ventricular potentials. In addition, the coherence function represents the aaount of correlation between channels over the frequency of interest.