Locating Atrial Fibrillation Rotor and Focal Sources Using Iterative Navigation of Multipole Diagnostic Catheters

Purpose Multi-polar diagnostic catheters are used to construct the 3D electro-anatomic mapping of the atrium during atrial fibrillation (AF) ablation procedures; however, it remains unclear how to use the electrograms recorded by these catheters to locate AF-driving sites known as focal and rotor source types. The purpose of this study is to present the first algorithm to iteratively navigate a circular multi-polar catheter to locate AF focal and rotor sources without the need to map the entire atria. Methods Starting from an initial location, the algorithm, which was blinded to the location and type of the AF source, iteratively advanced a Lasso catheter based on its electrogram characteristics. The algorithm stopped the catheter when it located of an AF source and identified the type. The efficiency of the algorithm is validated using a set of simulated focal and rotor-driven arrhythmias in fibrotic human 2D and 3D atrial tissue. Results Our study shows the feasibility of locating AF sources with a success rate of greater than 95.25% within average 7.56 ± 2.28 placements independently of the initial position of the catheter and the source type. Conclusions The algorithm could play a critical role in clinical electrophysiology laboratories for mapping patient-specific ablation of AF sources located outside the pulmonary veins and improving the procedure success. Electronic supplementary material The online version of this article (10.1007/s13239-019-00414-5) contains supplementary material, which is available to authorized users.

3D simulation: The simulations were performed on a real 3D anatomy based on the Harrild and Henriquez model S4 including: the left atrium, right atrium, Bachmann's bundle, pectinate muscles, left atrial appendage, fossa ovalis, and the superior and inferior vena cava. The mean element size of the model is 550 m and the maximum size is 1,650 m. Collagenous septa were generated by removing the electrical connections in the transmural direction perpendicular to the myocardial fiber orientation. The septa was created in the form of 2.5mmx2.5mm blocks at 400 randomly sampled (uniformly distributed) locations, with the length of septa sampled from a Poisson distribution with average length of 2.5mm. The 3D left atrium was segmented from the complete 3D anatomy and the data was used to test our algorithm.

AF Source Initiation:
We generated several test cases with AF rotor and focal sources in 2D and 3D simulations. In all cases, rotors were initiated by a cross-field stimulation protocol 5 and focal sources were initiated by providing a stimulus with an amplitude of 20mV and area of 1.25x1.25mm. To simulate functional figure-of-eight re-entry, we simulated two closed spaced rotor cores. Figure 2 shows 12 of the simulations.

Basket Catheter Simulation
To compare the performance of the ICAN algorithm with the phase-mapping approach we simulated a 64-electrode FirMap atrial basket catheter comprising of 8 splines with 8 evenly spaced electrodes in each spline. We simulated a basket catheter of 40mm in diameter (the highest resolution among the existing FirMap basket catheters) with an inbetween electrode spacing of 6.8 mm in each splines that has been scaled accordingly to the resolution of our 3D anatomy.
In the case of 2D atrial tissue, the catheter was mapped onto the endocardial surface as a rectangular electrode array with spacing 15.7mm x 6.8mm ( Figure 3B). A total of 120 simulations of 2D catheter placement was generated by shifting and rotating the basket catheter on the 2D atrial tissue. Three shifts (spacing of 1cm) were performed in the highest resolution direction and 8 shifts (spacing of 1cm) were performed in the lowest resolution direction. For every basket simulation, five different orientations (0 0 , 30 0 , 45 0 , 60 0 , and 90 0 ) were simulated.
The 3D simulation of the basket catheter was performed considering the realistic nonuniformity of electrode density in basket catheters. The 8 electrodes on one spline were evenly placed according to Eq. (2).
where indicates the coordinates of = {1, … ,8} unipole electrodes on one spinals of a catheter centered at ( 2 , 2 ) , = 20 is the catheter radius, ∆ℎ = 0.25 mm is the spatial resolution, , = 19.5° is the spacing between unipoles, and = 21.8° is to account for the distance between electrodes 1 and 8 from the beginning and ending of the spinal. The spline was then rotated 7 times with an azimuth of 45° to generate the coordinates of the electrodes on the remaining splines. A total of three simulations of 3D catheter placement was generated by rotating the basket catheter with an azimuth of 11.25°.

Detecting AF sources using basket catheter
We performed the following analysis to detect rotor or focal sources using a 64electrode FirMap atrial basket catheter. First, the extracellular voltage from the unipolar electrograms was spatially interpolated using bilinear interpolation to achieve a grid spacing of at least 4mm in all directions.
Phase mapping to detect rotors: The phase map was calculated between ± using the phase-mapping algorithm by Kuklik et. al 18 . In 2D atrial tissue, the phase singularity points were identified from the generated phase maps by applying the Iyer-Gray algorithm S5 . In 3D atrial tissue, we used an approach based on method in Ref. S7 to detect phase singularities. Finally, clusters of phase singularity points with less than 4mm distance from the center of the cluster were identified and those lasting greater than one wave propagation cycle were considered to be rotors 27 .
Velocity-of-divergence mapping to detect focal sources: We followed an approach similar to the method described in Ref. 27 . For every four neighboring unipolar recordings, an average conduction velocity vector was determined by averaging the isochrone vectors (same as in WDV calculation) obtained from the activation times of the three-recording combinations. Next, the divergence field of all the conduction velocity vectors was computed and the locations with divergence values of greater than the 95 th percentile of the values in the field and those lasting more than one wave propagation cycle were identified as focal sources.