Model based estimation of QT intervals in non-invasive fetal ECG signals

Figures

• 

  Fig 1. An example for T wave end estimation in one beat of nfECG. After plotting R(t), using (3), an interval in which a T wave end is expected to exist was calculated; by taking the median of the interval, one point for an end of a T wave was calculated.

• 

  Fig 2. A: A normal nfECG signal (blue graph) is simultaneously plotted with a Doppler signal (black graph). The estimated T-end points (red dots) are compared with the Ac points in a simultaneously recorded Doppler signal. Ac timings are used to pinpoint T-end timings because it is hard to locate them in the noisy nfECG signal. B: A normal nfECG signal (blue graph) is simultaneously plotted with a M-mode Doppler image. The estimated T-end points (red dots) are compared with the Ac points in a simultaneously recorded Doppler image.

• 

  Fig 3. Model-based estimation of end of T waves validated by a fsECG signal. The above figure shows a tracing for nfECG with model-estimated end of T waves (red asterisk). The signal at the bottom shows a tracing of fsECG recorded simultaneously with nfECG. The dashed lines indicate the end of T waves of fsECG. End of T waves in fsECG were measured after drawing a tangent line at the T wave. The intersection of the tangent line with the baseline was considered the end timing of a T wave.

• 

  Fig 4. nfECG (blue graph) is plotted simultaneously with a Doppler signal (black graph) to locate T-end points. The figure shows signals for a subject that suffers from bradycardia and LQTS.

• 

  Fig 5. The figures show Bland Altman plots for QT (left) and QTc (right) intervals of normal subjects. The total number of points in both figures is 17,227. The total number of points that fall within the LoA is 16,367 (95%) and 16,378 (95%) for the QT and QTc intervals respectively. LoA: Limits of agreement.

• 

  Fig 6. The figures show estimated and reference QT and QTc values plotted against GA. The normal cases are shown in circle and the abnormal cases are shown in asterisk. Both model-based and reference values show linear increasing trends. In (A), LQTS and tachycardia are outside the confidence bound in the model-based values plot. On the other hand, LQTS and one case of tachycardia are outside of the confidence bound in the reference values plot. In (B), LQTS and tachycardia are outside the confidence bound in the model-based values plot. In the reference values plot, only the LQTS case is outside the confidence bound. LQTS = Long QT Syndrome.

• 

  Fig 7. Ventricular AP. During depolarization or phase 0, Na+ and Ca+ channels open causing influx of Na+ and Ca+. The depolarization phase increases cellular potential. When the cellular potential reaches a certain level, specific K+ channels open causing a transient flow of K+ out of the cell, phase 1. The transient current is known as Ito. The efflux of Ito causes a slight reduction in the cellular potential. Phase 1 is followed by phase 2 or plateau phase in which the efflux of Ito is balanced by the influx of Ca2+. In phase 3, more K+ channels open to restore the cell into the resting potential. The current in phase 3 is known as Ik and it has two components, rapid (Ikr) and slow (Iks). The efflux of Ik current continues until the cell is restored to its resting potential in which the dominant current is a leak K+ current known as Ik1.