Abstract 16187: Validation of Novel Algorithm to Automate Detection of Fractionated Electrograms During Ventricular Tachycardia (VT) Ablation

Abstract

Introduction: Ablation of ventricular tachycardia (VT) substrate in patient at risk for VT in the setting of ischemic heart disease is a technically challenging procedure. We thought to evaluate a novel algorithm used to automatically identify target electrograms. Methods: 16 consecutive patients (70±10 years of age, 90% male, 34±18% LV EF) had 20 ablations for ischemic VT using CARTO 3 mapping system over 2 years. Left ventricular (LV) substrate was mapped during right ventricular (RV) apical stimulation. Navistar Thermocool 3.5 mm irrigated tip catheters were used in all patients. A novel algorithm counting the number of electrogram deflections (NOD) crossing the 0.05mV noise threshold and duration of time from first to last such deflection during the window of interest (total fractionation time, TFT) was applied to all acquired maps after ablation was complete. Snapshots of 200 electrograms representing the high and low end of TFT and NOD values were presented to a group of 8 electrophysiologists experienced in VT ablation who were asked to select electrograms they would target for substrate ablation. The diagnostic accuracy of TFT and NOD values was then analysed. Results: Across the range of TFT values (0.0-281.0 ms), a cut-off value of 49.0 ms (81.6% sensitivity, 57% specificity) was established as an optimal indicator of an ablation target. Area under the curve for TFT was 0.675 (95% CI: 0.59-0.75, p=0.001). For NOD values (0.0-70.0 deflections), a cut off of 4.5 deflections (88.0% Sensitivity, 57 % specificity) was established as an optimal indicator of an ablation target. The area under the curve for NOD yielded an area of 0.75 (95% CI: 0.68-0.82, P=0.001). For TFT-NOD product as a variable, a cut-off value of 64 (91.0% Sensitivity, 52.4 % specificity) an optimal indicator of an ablation target. The Area under the curve for NOD and TFT multiple was 0.72 (95% CI: 0.65-0.80, P=0.001). Conclusion: A novel algorithm may be able to automatically classify LV substrate during mapping and ablation of ischemic VT with high sensitivity and acceptable specificity.