[Spectral analysis of the ECG for detecting late potentials as risk factors of life-threatening cardiac arrhythmias].

Late potentials in the terminal phase of the QRS and early S-T segment are looked upon as a risk marker in patients prone to sustained ventricular tachycardia after myocardial infarction. Since the amplitude of late potentials at the body surface is very low (1-5 microV), most studies use signal averaging of the ECG to increase the signal-to-noise ratio. Two different approaches are generally used to analyze the signal-averaged ECG. In the time domain, the individual channels are combined into a vector magnitude and high-pass filtered in a bidirectional mode. Late potentials are suspected if the filtered QRS duration is greater than 120 ms and/or the amplitude in the terminal 40 ms of the QRS complex is less than or equal to 25 microV. The limitations of this method are that the definition of abnormality differs from one study group to another, highpass filters may introduce artificial signals, patients with bundle branch block in general have to be excluded, and the definitions depend upon the noise level. More recently, spectral analysis of the ECG with Fast Fourier Transform (FFT) has been performed. Late potentials are characterized by a higher frequency content in the otherwise low-frequent S-T wave. We analyzed 25 overlapping segments of the terminal QRS and early S-T wave time shifted in steps of 2 ms with FFT (spectro-temporal mapping). This method was shown to overcome some of the limitations of conventional time domain analysis: no highpass filters have to be applied, noise interference can be detected by a characteristic spectral pattern, and patients with bundle branch block need not be excluded.(ABSTRACT TRUNCATED AT 250 WORDS)