Enhanced detection of distinguishing features in signal-averaged electrocardiograms from patients with ventricular tachycardia by combined spatial and spectral analyses of entire cardiac cycle.

BACKGROUND

Signals generated by myocardium responsible for ventricular tachycardia (VT) contribute to the entire QRS complex, ST segment, and T wave and are spatially distributed over the entire torso. However, current methods of signal-averaged ECG analysis restrict interrogation to the terminal QRS complex, do not include data on the body surface distributions of the distinguishing features detected, and have a limited clinical value because of a low positive predictive accuracy. Accordingly, we tested the hypothesis that frequency analysis of the entire cardiac cycle of spatially selected ECGs based on isoharmonic maps of the body surface enhance the detection of the unique spectral features in signal-averaged ECGs that differentiate patients with from those without VT.

METHODS AND RESULTS

Isoharmonic maps of the body surface were calculated during sinus rhythm with the use of forward problem solutions for 32 patients with sustained VT, 30 without VT, and 10 healthy subjects and analyzed over a bandwidth of 0.05 to 470 Hz. Spectra of ECGs at the maximum and minimum of each patient's isoharmonic map of 1 to 7 Hz demonstrated a broadened bandwidth of significant separation (P < .05) for patients with from those without VT compared with the separation achieved with the use of Frank ECGs alone. Furthermore, the statistical significance within the bands of separation was greater for spatially selected ECGs compared with the Frank leads. Frank leads separated patients over the band from 11 to 84 Hz with a mean value of P = .0094. ECGs at the maximum of 1-to-7-Hz isoharmonic maps separated patients over the 8-to-111-Hz band with a mean value of P = .0062 (range, P < .05 to P < .0000001). ECGs at the minimum of 1-to-7-Hz isoharmonic maps extended the low-frequency end of the band of separation, which covered 0 to 69 Hz with a mean value of P = .0039 (range, P < .05 to P < .0000001). Subgroup analysis verified that results were independent of QRS duration.

CONCLUSIONS

Spectral analysis of ECGs that are spatially selected for each patient is superior to orthogonal ECGs and augments detection of distinguishing features in ECGs that identify risk of VT. The new data acquired from analysis of spatially selected ECGs from individual patients provide the information on the specific frequency bands and an improved ECG-lead system required to refine methods of analysis of the signal-averaged ECG.