High Resolution Electrocardiography—Its Application for the Measurement of the QT Interval in the Presence of Low Amplitude T Waves

Abstract

ObjectiveWe describe a computer‐interactive method for determination of QT duration in the presence of unfavorably shaped T waves. We assumed that such a method should allow to determine QT duration with the same error irrespective of the T wave amplitude.MethodsThe study group comprised 20 patients following myocardial infarction and the control group, 10 healthy volunteers. ECGs were recorded from XYZ leads using a computer‐assisted low noise amplifier. After A/D conversion (16 bit, 2 kHz) fiducial points of R and T waves were determined and then the signal was aligned and averaged. A segment of 106.5‐ms duration was chosen that included either the initial portion of QRS complex or the terminal part of the T wave. This segment was scanned with a 32‐ms window every 0.5 ms. Spectrotemporal analysis was performed by use of the fast Fourier transform. The QRS onset and T wave offset were defined as points at which the percentage ratio of power in frequency range 62.7–121.6/0–121.6 Hz reached minimum. Signal‐averaged ECGs were printed out (120 mm/s, 2.1 mm/mV) for manual measurement of QT duration and T wave amplitude. Low amplitude T wave was defined as lower than 0.2 mV, otherwise normal amplitude T waves were recognized. Measurements were repeated separately by two observers for each lead.ResultsLow T waves were observed in 11 (37%) leads from controls and in 25 (42%) from post‐Ml patients. Manual measurements were possible in 82 leads (all normal and 28 low T waves), while the high resolution method allowed us to determine the QT duration in all leads. In the Ml group, the results of independent QT measurements by means of high resolution were: 412.0 ± 29.8 ms versus 411.5 ± 29.3 ms for flat, and 410.7 ± 30.4 versus 409.8 ± 30.0 ms for normal T waves. In controls, corresponding values were: 388.5 ± 32.0 versus 386.3 ± 23.8 ms and 392.9 ± 32.8 versus 392.2 ± 31.7 ms. The mean absolute error of two independent measurements was greater with use of the manual method, and reached 6.9 ms (median 5.5 ms) in normal and 12.9 ms (median 8.5 ms) in low T waves. With use of the new method, the errors were lower: 3.48 ms (median 0 ms) and 2.29 ms (median 0 ms), respectively. There was a close correlation between repeat measurements in normal T waves with both methods (r = 0.956 and r = 0.961; both P < 0.001). In low T waves the correlation was significantly lower (P < 0.05) with the manual (r = 0.879; P < 0.001), as compared with the high resolution method (r = 0.973; P < 0.001).ConclusionThe signal‐averaged, high resolution ECG analysis allows for the determination of QT duration in the presence of a flat T wave. The described method may provide the opportunity to measure QT duration in all leads and, therefore, to calculate spatial QT dispersion in the majority of post‐Ml patients.