Detection of late potentials in the signal-averaged ECG combining time and frequency domain analysis

The paper reports experience of flexible, computer-based instrumentation for high-resolution ECG, designed for laboratory research and clinical applications. It is an attempt to improve methods of detection of late potentials when identifying patients at increased risk of ventricular tachycardia (VT) and sudden cardiac death. Several factors influencing the efficiency of a high-resolution ECG system are examined. Of primary importance are signal, gain, type of filter, length and position of the time interval taken for spectral analysis, type of window used etc. Some possible error sources are analysed. Time-domain analysis is illustrated by representative XYZ-vector magnitude tracings of a non-VT and a VT patient. In an effort to further refine the diagnostic strategy, an approach to data analysis is presented, combining time-domain with frequency-domain measurements. A spectral ratio is proposed that helps recognise patients prone to VT. Two-dimensional plots of this ratio against time-domain parameters reveal clearly distinguishable normal and pathologic groups. It can be concluded that an appropriate frequency analysis can add significant diagnostic power to existing instrumentation for signal-averaged ECG to predict VT risk.