Improved estimation of V-index based on analytic forms of dominant T-wave

Spatial heterogeneity of ventricular repolarization (SHVR) is related to the development of arrhythmias. To assess SHVR, we introduced the V-index, a metric which needs computation of the Dominant T-Wave (DTW) and its derivatives. Theoretically, the larger the number of derivatives, the better the adherence to the modelled T-wave. In practice, only the first derivative is included, as the numerical computation of higher derivatives is corrupted by computation noise. Here, we introduce a parametric method (PM), based on analytic definitions of the DTW, to allow analytical computation of its derivatives. Three analytic forms, based of combination of sigmoidal (S), Gaussian (G) or exponentials (E) functions, were considered. A set of simulated ECGs were generated using a forward ECG model (Matlab version of ECGSIM). SHVR was varied from 5 to 40 ms (5 ms-steps). To simulate real recordings, noise available from the MIT-BIH Noise Stress Test Database was added with different peak-to-peak amplitudes (30, 60, 120 and 180μV). The use of PM allowed the inclusion of a larger number of derivatives in the model and reduced the difference between actual and estimated T-waves, especially for larger SHVR. This reduction was more pronounced for model S and G. However, the model E resulted in a lower estimation bias of V-index with respect to the actual SHVR.