Real-time QT interval measurement

A real-time QT interval measurement algorithm was developed and preliminary tested. Three ECG leads (I, II and V5) were digitized simultaneously at a rate of 500 samples by second during five minutes; twenty patients were studied. The resulting signal was filtered in order to reject noise and avoid baseline wander. A Function of Spatial Velocity was used as a QRS detector, complexes were classified in PVC and non-PVC. Morphological features and a difference area function were used to classify the complexes. When a non-PVC was found, a squared slope function was computed for samples corresponding to the 60% of the next RR interval. The maximum of this function was always after the T wave peak and two samples, one after and other before this maximum, were used to estimate a tangential line to the signal in this zone. The intercept of this tangential line with the isoelectric line was defined as the end of the T wave and the QT interval was the distance between the QRS onset and this point. Two cardiologists measured some QT intervals aided by a graphic computer program separately and the mean value of these measurements were used as a golden rule to evaluate the proposed algorithm. The QT intervals measured were randomly selected. The absolute difference between the algorithm and cardiologists measurements was never greater than 22 ms and the mean difference was 9.946 ms