Directed graph mapping exceeds phase mapping in discriminating true and false rotors detected with a basket catheter in a complex in-silico excitation pattern

Figures

• 

  Fig. 1. Different steps of the PM analysis. Panel A shows a single time frame of the voltage map of the meandering rotor in the atria. The true rotor is indicated with a black arrow. The SVC, IVC and TV, respectively the Superior Vena Cava, Inferior Vena Cava and the Tricuspid Valve are indicated with black circles. In panel B the locations of the 3 different positions of the 64-pole basket catheter are displayed inside the RA. In panel C, the flattened catheter basket is visualised in black for the 3 positions with an additional periodical column, leading to the 72 nodes in total. In panel C, the interpolation of the phase to the 57,825 electrodes is shown for a single time frame. To better understand the position of these flattened baskets inside the RA, we also indicated the veins. Panel D shows the classifications of the different cores for the three positions detected by PM. The black arrows show the rotation direction of the detected rotors.

• 

  Fig. 2. Illustration of the DGM protocol. In panel A we show 928 nodes inside the RA at the SVC position which were used for DGM. In panel B, we picture the neighbouring protocol for each of these nodes. In panel C we show the distribution of LATs at a single time frame for each node. In panel D, we demonstrate how the directed network is created for a single node. Starting from the node which is indicated with a star, we determined the difference in LAT and the Euclidean distance between this node and each neighbour. In case these values satisfied equation (1), a directed arrow was drawn from the source node to the neighbouring node. This was repeated for all the nodes, creating a directed network as shown in panel E. The directed cycles in this network are determined in panel F.

• 

  Fig. 3. Panel A shows from left to right: the original flattened 72-pole basket catheter projected on the 2D plane; its corresponding 3D location in the atrium; the interpolated 924 electrodes in 3D; the corresponding 957 pole basket catheter in 2D. Panel B shows the linear interpolation of the ECGMs to the 957 electrodes. Based on the filtered ECGMs, the LATs are calculated for each electrode. Panel C illustrates 2 different neighbouring protocols. As the spherical neighbouring protocol does not result in a representative set of neighbours (too many neighbours at the pole, too little at the equator) a circular neighbouring protocol was chosen.

• 

  Fig. 4. Detection categories of DGM in the basket catheter. In panel A we visualised the 928 pole basket catheter in the 3 positions within the right atrium. In panel B the flattened 957 (29 x (32 + 1)) catheter with the corresponding LATs at the simulation time t = 1500 ms is shown. The colours denote the time which passed since the last activation. Panel C shows the different regions detected by DGM on the 957 flattened catheter.

• 

  Fig. 5. Comparison of the detection performance of PM and DGM. For PM, a single set of results was found. The height of the bar corresponds with the detection performance, the colour with the corresponding reentry. The lines in between PM and DGM columns show the performance of DGM under variation of CVmin from 0.01 cmms to 0.04 cmms for a fixed maximal conduction velocity of 0.2 cmms. The last column is the best setting for only finding the true rotor in DGM. Again, the height of the bars represent the performance rate.