Catheter Ablation -> Atrial Fibrillation & Atrial Flutter: -> Mapping & Imaging D-PO01-NC - Non-CME Poster Session (ID 14) Poster

D-PO01-NC-16 - Dielectric Based Transmurality Classifier Validation (ID 1483)

  Y. Reisner: Salary from Employment (Commercial Interest) - Philips. Ownership/Partnership/Principal - CorAssist Cardiovascular.
Audio File Upload


Background: Direct real-time lesion transmurality detection during RF delivery may facilitate ablation procedures
Objective: A new novel application for real-time classification of RF lesion transmurality is described.
Methods: Preparations of swine RV heart sections were studied in vitro. RF ablations using variable powers and durations (20W or 30W, 20s-110s) were performed. A transmurality classifier was generated based on electrical signals recorded with KODEX-EPD system (Philips) from an ablation catheter electrodes (Blazer™ OI, Boston Scientific). The transmurality classifier was compared to transmurality determined from photographed 2,3,5-Triphenyltetrazolium chloride (TTC) stained tissue cross sections. A rim value of < 0.6 mm (average of 3 blind independent observers) was selected as transmurality threshold.
Results: The study included training and validation sets. The training included 448 lesions from 59 different animal RV sections and the validation set 277 lesions from 31 RV sections. A Cross - Validation, leave-one-tissue-out analysis of the training group obtained a sensitivity of 82%, specificity of 83% and a receiver operating characteristic (ROC) area under curve (AUC) of 0.89. The validation set analysis included 253 lesions (84 Transmural, 169 Non-Transmural) while 24 lesions were excluded (16 steam pop, 8 partial data). The transmurality classifier identified transmurality with sensitivity 82.1%, specificity 87.6% and ROC AUC of 0.91.
Conclusion: A novel transmurality classifier with high detection accuracy was generated. The classifier may have clinical application for real time assessment of lesion formation during RF ablation.