Spectral mapping of the electrocardiogram with Fourier transform for identification of patients with sustained ventricular tachycardia and coronary artery disease.

In time domain analysis, detection of late potentials is limited by high pass filtering, noise interference and the necessity to exclude patients with bundle branch block. We therefore used frequency analysis with Fourier transform of multiple segments of the surface electrocardiogram (25 segments, size 80 ms, time shift 3 ms) during sinus rhythm after signal averaging. Thirty-two post-myocardial infarction patients with sustained ventricular tachycardia (VT), 19 post-myocardial infarction patients without VT and 17 healthy subjects were studied. A total of 18 patients had bundle branch block. In 24 out of 32 patients with VT, three-dimensional spectral plots were characterized by spectral peaks greater than 10 dB in the range of 40-200 Hz in segments only at the end of QRS and the early ST wave, but not far outside the QRS. In only 2 out of 19 patients without VT and in 1 out of 17 healthy subjects could such peaks be observed. Noise caused spectral peaks throughout all segments. Sixteen out of 18 patients with bundle branch block were correctly classified with spectral mapping. With the Simson method, patients with bundle branch block had to be excluded, abnormal results were found in 10 out of 19 patients with VT, but also in 5 out of 15 patients without VT and in 3 out of 16 healthy subjects. Thus, spectral mapping of the electrocardiogram offers promise for better identification of patients prone to sustained VT in the presence of coronary artery disease.