[Automatic classification of the greatly amplified ECG in relation to time and frequency]

The non-invasive recording of cardiac micro-potentials from the surface of the body, in particular of the so-called ventricular late potentials from the highly amplified ECG, makes it possible to identify those patients at high risk from severe arrhythmics or sudden death. By using more extensive automatic parameter extraction methods, we were able to increase the detection rate of pathological ECGs. We were also able to show that the calculation of the power density spectrum, in particular on the basis of maximum entropy spectral estimation, in combination with the variance extraction method of eliminating the influence of residual noise, is capable of providing separate additional valuable information.