https://doi.org/10.1007/s00392-024-02526-y
1Universitätsklinikum Schleswig-Holstein Innere Medizin III mit den Schwerpunkten Kardiologie, Angiologie und internistische Intensivmedizin Kiel, Deutschland; 2Christian-Albrechts-Universität zu Kiel Kiel, Deutschland; 3Biosense Webster Norderstedt, Deutschland
Background: Efficacy of the catheter ablation (CA) of ventricular tachycardia (VT) depends on accurate visualization of the critical isthmus during electroanatomic mapping. A major challenge is the imprecise automatic annotation of diastolic potentials (DP), often due to far-field signal annotation. Biosense Webster's Late Annotation mapping (LAM™) algorithm aims to improve late potential annotation during substrate mapping. However, there's limited comparative evaluation with standard Wavefront™ annotation, even when enhanced by manual reannotation.
Objective: This study evaluates LAM™'s effectiveness against Wavefront™ annotation in identifying DP during VT activation mapping.
Methods: High-density activation maps from nine patients, undergone CA for VT were recorded using Carto®3 (Version 7.5; Biosense Webster Inc.). Maps were reproduced offline with three algorithms: Wavefront™, LAM™, and manual reannotation. LAM™'s late boundary threshold was set at 0 ms (mid-QRS complex), and Late Annotation sensitivity was adjusted. Three time zones (late diastolic, systolic, early diastolic) were defined around the QRS complex and window of interest (WOI). Maps were analyzed using Python software, focusing on point count in each zone. DP percentage comparisons were made against manual reannotation, which was considered Gold Standard.
Results: Wavefront™ detected 7,708 late diastolic points (avg. 856.4/patient, 79.13% of Gold Standard) and 3,653 early diastolic points (avg. 405.9/patient, 50.12% of Gold Standard). LAM™ detected 7,071 late diastolic points (avg. 785.7/patient, 78.62% of Gold Standard) and 4,138 early diastolic points (avg. 459.8/patient, 83.44% of Gold Standard). Mann-Whitney U test showed no significant difference in late diastolic point detection (U: 38.0, p: 0.86) between algorithms. However, LAM™ was significantly better in early diastole (U: 14.0, p: 0.02) (Fig. 1 and 2).
Conclusion: LAM™ outperforms Wavefront™ in early DP annotation. Though both algorithms fall short of manual annotation in DP detection, LAM™'s early diastole advantage suggests its potential. Further adjustment of the LAM™ parameter "late boundary" and the WOI may be considered to improve the result and close the gap with manual reannotation, optimizing VT mapping.