Pranav Bhagirath

169 A priori model independent inverse potential mapping: the impact of electrode positioning multiplication with the transfer matrix T . From these BSP, ECG potentials were computed by spatial sampling. In order to approximate real-life conditions, noise was added to the simulated ECG potentials with a signal-to-noise ratio (SNR) of 21 dB. Figure 2. Flowchart visualizing the processes associated with forward simulations and inverse computations. Inverse reconstruction of epicardial potentials from simulated ECG potentials Epicardial potentials ( P epi) were calculated from the ECG potentials ( P bs) using P epi = ( TTT +λ2 I )−1 TTP bswhere T is the forward transfermatrix andλ is the regularisation strength, initially determined by simulation with patient-specific geometries. For all electrode configurations, the correlation between the initial source epicardial potentials (used for the simulation of ECG potentials) and the reconstructed epicardial potentials was computed. In addition, for all electrode configurations, correlation was calculated for 16 identical points on the anterior epicardium. These values were plotted in a graph to visualise correlation trends. To exclude systematic errors due to grid artifacts, all computationswere performedondifferent grids. Four different electrode configurations were used in the simulations ( figure 3 ).