Spectro-temporal mapping of the surface electrocardiogram in clinically normal dogs

To establish data for selected variables of the signal-averaged ECG and determine whether sedation affects segment lengths and points within the terminal QRS complex where windows began.

Overtly healthy dogs with no detected abnormalities.

Signal-averaged ECG (600 beats) were obtained from dogs in left-lateral recumbency, using X, Y, and Z leads with 26-gauge subdermal platinum electrodes. Frequency-domain analyses were performed by fast Fourier transform (FFT), using a post-processing computer software package. Analyses were extracted from 30-, 40-, and 50-millisecond windowed (Blackman-Harris) segments of the terminal 10 or 20 milliseconds of the QRS complex and ST segment. Correlation ratios were calculated by cross-correlation of frequency spectra of each FFT in the range of 40 to 200 Hz. The reference spectrum was the last FFT (nearest the T wave). Spectro-temporal maps were plotted for each analysis.

Noise levels were < 1 microV. Sedation resulted in less noise and slower heart rate, but did not affect results. Correlation ratios were not different among 30-, 40-, and 50-millisecond segments, but were lower when 20 (vs 10) milliseconds of the terminal QRS was included. Spectro-temporal maps were similar in all subjects.

Use of sedation facilitated the procedure without affecting results. Segment lengths of 40 and 50 milliseconds, beginning 10 milliseconds prior to the end of the QRS complex and analyzed in the range of 40 to 200 Hz, yielded correlation ratios > 67%, 95% of the time.

Spectro-temporal mapping may identify dogs with ventricular late potentials (high-frequency signals commonly associated with myocardial fibrosis) that may be at increased risk of sudden death.