Circle Method for Robust Estimation of Local Conduction Velocity High-Density Maps From Optical Mapping Data: Characterization of Radiofrequency Ablation Sites

Figures

• 

  FIGURE 1. Illustration of the experimental setup and the CM method application to obtain local CV and CV maps. (A) For isolated rat atria, RFA was performed in steps of increasing ablation duration of 0.5, 1, 1.5, 2, 2.5, 3, and 4 s, with 2 min pause between the steps. The ablation was performed on the epicardial side with optical mapping measurement on the endocardial side. The ablation was performed on the ventricular tissue for the whole guinea pig heart and human heart preparations. Collimated and bandpass-filtered LED light was used to excite the V m dye. Emanating fluorescent light was passed through the long-pass filter in front of the EMCCD camera sensor. (B) Illustration of the circle method, evaluating LATs differences at the points along a circle of radius r, centered at a grid point (i,j). Left, example of a wave propagating at 45°. Right, a set of test CV values, S (θ n), which is evaluated by taking the difference in LATs along diameters (blue lines) at different angles θ n . The circle method evaluates the resulting conduction velocity by performing a geometrically weighted average of S (θ n) over an angular range around θ PD (blue shaded region). (C) Top, an example of a…

• 

  FIGURE 2. Numerical simulations of the RFA and effects of added noise on CV for the spatial determination of the ablation area. (A) Left, the LAT map of a simulated curved wavefront propagation with isochrone lines spaced 20 ms apart, and superimposed CV vector map showing CV magnitude and direction. The ablated area is modeled as a circular region, 20 pixels radius at the center. Middle, the corresponding CV maps from the LAT map for the whole domain and sub-domain (80 × 80 pixels) region around the ablated area outlined with an inscribed dashed rectangle. Right, 2D Pearson’s correlation coefficient for different SNR levels ranging from 5 to 60 dB, for the CV maps of different sizes on the left, 236 × 236 (Correlation 1) and 80 × 80 pixels (Correlation 2). The Dice similarity coefficient for the selected 80 × 80 pixels CV map around the ablated area. (B) Zoomed-in CV surrogate maps for different SNRs of 60, 30, 25, 20, 15, 10, and 5 dB, with the delineated boundary of the ablated area. The color bar range in all figures is set from the 5th to 95th percentile.

• 

  FIGURE 3. Effect of ablation on CV slowing on isolated rat atria. (A) LAT map obtained before (Baseline) and after RFA for the total duration of 14.5 s and power of 10 W, approximately at the center of the map, along with the superimposed CV vector map. (B) Respective CV maps obtained with the CM method for selected dashed square regions presented in (A) outline the ablated areas as regions of CV reduction. (C) Relative change, ΔCV, between the CV maps after and before the ablation, shows a large relative decrease of CV in the ablated areas, delineating the ablated area from non-ablated, suitable for characterization of the ablated area boundary. (D) Optical AP traces from a pixel inside the ablated area, normalized before ablation, show the reduction in AP amplitude and morphology changes. The color bar range in all figures is set from the 5th to 95th percentile, except for ΔCV maps’ upper boundary that was set to zero. Radius of 10 pixels (1.3 mm), in the CM, was used for CV calculation.

• 

  FIGURE 4. Effect of ablation on CV slowing on guinea pig ventricle for different RFA durations. (A) LAT maps obtained at baseline (top) and RFA duration of 3 s (middle) and 10.5 s (bottom), respectively. (B) Respective CV maps inside the dashed square regions, where ablation is performed, show CV decrease, with a larger CV decrease after longer RFA. (C) Relative difference in CV after and before ablation, ΔCV maps, outlines the ablated region as a distinct decrease of CV. The color bar range in all figures is set from the 5th to 95th percentile, except for ΔCV maps’ upper boundary that was set to zero. Radius of 10 pixels (1.6 mm), in the CM, was used for CV calculation.

• 

  FIGURE 5. Optical mapping on whole explanted human heart for quantification of RFA-created lesion. RFA was performed on ventricle’s epicardial side in two steps, termed RFA1 and RFA2, with power set to 40 W for 10 s in each step. (A) LAT time maps show wavefront propagation slowing in the upper right region where the ablation was performed. (B) CV maps before and after ablation show characteristic CV slowing in the ablated region. (C) Change in CV obtained as the relative difference after and before ablation clearly outlines the red area where ablation was performed of a distinct CV decrease. (D) Black and white images captured using the optical mapping camera illustrate ablated regions. (E) Optical AP traces from a pixel marked with a yellow dot in D normalized before ablation. The traces show the relative change in fluorescence corresponding to transmembrane voltage change. The color bar range in all figures is set from the 5th to 95th percentile, except for ΔCV maps’ upper boundary that was set to zero. Radius of 4 pixels (1.1 mm), in the CM, was used for CV calculation.

• 

  FIGURE 6. Optical mapping on the LV’s endocardium in wedge preparation for quantification of RFA-created lesion. The RFA was performed on a highly trabeculated endocardium for 20 s, and power set to 40 W. Different light excitation bands were used, a green LED of 525 nm center wavelength and a red LED of 660 nm center wavelength, to measure optical APs emanating from the endocardial surface and deeper transmural layers. (A) Due to the highly heterogeneous endocardium (Supplementary Figure S2), identification of the ablated area from LAT maps is non-trivial. (B,C) CV maps and the relative CV difference after and before the ablation do not clearly outline the ablated area, lacking characteristic CV slowing. (D) Relative difference in CV maps between subsequent beats (wavefront propagation), termed even and odd, elicits the CV differences due to AP amplitude alternans in the ablated area. Green LED light illumination of penetration depth limited to the endocardial surface layer, and the absence of alternans indicates a complete ablation of the surface layer. Illumination with a deep-red light band of deeper penetration depth enables measuring APs from deeper ventricular wall layers. The…

• 

  FIGURE 7. Estimation of the ablated area size based on the CM radius. (A) In numerical simulations, CV linearly increases with CM radius inside the ablated area of 20 pixels in radius, reaching a limit value outside the ablated area. The ablated area radius can be estimated as a deviation, the knee point, from a linear rise, independent of wavefront direction. (B) Heatmap of CV values at the ablated area center of rat atria as a function of RFA duration and CM radius. The heatmap shows the expansion of the ablated area with each subsequent ablation step characterized by distinctive CV slowing. (Bottom) CV curves are shown for different RFA steps, with a knee point for the approximate radius of 25 pixels, or 2 mm. (C) Whole isolated guinea pig heart. The CV curve at the center of ablation (RFA of 3 s) for different CM radii flattens for an approximate radius of 2.5 mm, indicating an RFA-created lesion of 5 mm diameter. (D) Human heart epicardium of LV. From the figure, the CV curve, obtained in the same fashion, flattens for an approximate radius of 1.5 mm. (E) Human heart endocardium of LV. Upper, for deep-red light excitation, the CV curve flattens for an approximate CM radius of 2.5 mm,…

• 

  FIGURE 8. Comparison of CV maps using the CM and FiD method. (A) CV maps generated with the CM method for the simulation, rat’s atria, guinea pig’s ventricle, and human ventricle (epicardium). (B) Respective CV maps obtained with the FiD method. (C) Relative difference of CV maps obtained with the two methods.