Characterization and automatic identification of ECG conduction abnormalities using segmental multiharmonic Fourier analysis of gated blood-pool scintigrams.

This study of biventricular segmental timing sequences compared patterns of the first-harmonic phase with those of multiharmonic Fourier-analysis-derived parameters to determine their suitability for the classification of patients using an automatic pattern recognition scheme. The study involved nine patients with normal ECGs, six with left bundle branch block, 4 with right bundle branch block, and 6 with right-ventricular pacemakers; all patients had normal left-ventricular function. The segmental sequence described by the time of end systole was similar to that described by the first-harmonic phase, with a lower correlation using the time of maximum filling rate, and a rather poor correlation of the time of maximum ejection rate with the other parameters. Thus, despite theoretical difficulties, timing patterns described by phase best agreed with those of the time of end systole. Intersegmental timing differences were used as criteria for discriminating between groups by means of a sequential logic tree. Using either phase, time of end systole or time of maximum ejection rate, all patients were correctly classified into the four ECG categories. More intersegmental timing differences were useful as criteria for phase (57) than for the time of end systole (45) or the time of maximum ejection rate (30). The first-harmonic phase was more efficient than the time of end systole for separating patient groups. Sequential biventricular timing disorders can be objectively classified using either first-harmonic or multiharmonic Fourier analysis, and their patterns can be automatically recognized and used for classification.