The effect of configuration parameters of time-frequency maps in the detection of intra-QRS electrical transients of the signal-averaged electrocardiogram: Impact in clinical diagnostic performance

Abstract

Time-frequency maps of signal-averaged electrocardiogram based on a short time Fourier transform (STFT) technique analysis was employed to assess the presence of high frequency electrical transients (turbulence) and arrhythmia risk assessment. The optimal configuration set of STFT variables aiming at risk stratification for sustained monomorphic ventricular tachycardia (SMVT) is still undetermined. Different configuration variables, including analyzing time-window widths, starting positions, relative displacements, and zero-padding for STFT time-frequency maps constructions were combined on each analyzing averaged signal from 18 healthy controls and 18 subjects presenting inducible SMVT. Spectral turbulence analysis (STA) was, thus, carried out according to conventional procedures. The optimal configuration set of variables for STA was obtained by assessing the total diagnostic accuracy of all combinations of parameters. The optimal diagnostic performance was found at 86% total diagnostic accuracy as compared to 56% using previous defined normality thresholds (p=0.01). Present configuration set of variables is distinctive from previously defined set of variables and improves risk stratification.

Analysis of Abnormal Intra-QRS Potentials in Signal-Averaged Electrocardiograms Using a Radial Basis Function Neural Network

“…Although time-domain VLP analysis has confirmed its high negative predictive value, it lacks a sufficient positive value to diagnose a person as being ill [2]. There have been a number of studies published on methods for improving VLP detection, e.g., spectral turbulence [6], the wavelet transform-based method [7] and principal component analysis [8], however, no consensus exists yet on their methods and clinical applications.…”