Morphology-based QT interval measurement using frame-based representation of the ECG signal

We propose a new method for QT interval measurement which first detects the morphology of the T-wave and then finds the T-end according to the detected morphology. The proposed method has more emphasis on detection of the T-wave morphology which helps more accurate detection of the T-end and better lead selection for QT interval measurement. Six types of T-wave morphologies (normal, invert, biphase +-, biphase -+, only upward and only downward) have been considered and a rule-based algorithm employed to detect the T-wave morphology by examining the primitive components. Followed by the detection of the morphology, the T-end is determined as the point on the signal that has maximum distance to the line representing the last segment of the T-wave. The proposed method has been tested against the QT database available on Physionet. The mean difference between the T-end in the reference annotation and the calculated one by our method is 0.9 ms and the standard deviation is 18.9 ms.