QT interval variations of the surface electrocardiogram (ECG) reflect beat-to-beat fluctuations of the ventricular repolarization. Several studies have shown that temporal repolarization lability as indicated by an increased QT interval variability (QTV) is associated with cardiac mortality. However, measuring subtle beat-to-beat changes in QT interval remains challenging. Although novel QTV techniques have improved sensitivity and robustness, conventional QTV measures still lack significance. This study proposes a novel ECG waveform morphology based QTV measure for the characterization of the ventricular repolarization lability and employs it in patients with myocardial infarction (MI).
We analyzed recordings of 79 MI patients and 69 healthy control subjects included in the Physikalisch-Technische Bundesanstalt (PTB) diagnostic ECG database. To characterize the QT interval waveform, we employed two-dimensional signal warping (2DSW). Based on the two-dimensional template adaptation to every beat, a novel so-called QTfluc parameter has been developed to take into account complex QT interval’s waveform fluctuations. To demonstrate the power of QTfluc we (1) compared MI patients and healthy subjects and (2) examined the stability of QTV measures in relation to QT interval boundary shifts, i.e. we varied the interval length over which morphological changes were expected to affect QTfluc.
A comparison of QTfluc with standard QTV measures showed a significant improvement (effect size increased up to 30%) in discriminating between MI patients and healthy subjects. QT interval boundary shifts showed significant less impact on the stability of QTfluc in comparison to standard QTV measures. Furthermore, QTfluc showed more independent characteristics to QT interval duration and T wave amplitude changes.
The proposed measure shows significant improved characterization of ventricular repolarization lability in MI patients. Moreover, QTfluc shows more stable characteristics and is less dependent on QT interval boarders. Further investigations will focus on the evaluation of additional datasets and the application to other quasiperiodic biosignals.