Predicting Transmural Lesion Formation and Steam‐Pop Occurrence During Bipolar Ablation—Ex Vivo Porcine Model

Background: Bipolar radiofrequency (RF) ablation can create deeper myocardial lesions than unipolar ablation, yet its optimal settings remain undefined. Objective: To develop and validate predictive models for lesion transmurality and steam-pop occurrence during bipolar ablation. Methods: Ex vivo bipolar ablation was applied to porcine myocardium (5-20 mm thickness) with catheters placed bilaterally at 45-degree and 10-g contact-force. RF power (20-50 W) and duration (20-180 s) were systematically varied. Generalized linear models (GLM) were trained on 194 applications to predict transmurality and steam-pop from RF energy, tissue thickness, initial bipolar impedance, 5-s impedance drop (absolute and percentage [PercentImpDrop5]), and RF duration; 111 independent applications served as the validation. Results: = 0.003). For steam-pop, the model combining RF energy and PercentImpDrop5 showed the best discrimination (AUC 0.90 [0.82-0.97], sensitivity 84%, specificity 90%); notably, PercentImpDrop5 alone achieved comparable accuracy (AUC 0.89). Conclusion: Tissue thickness is the dominant determinant of transmural lesion formation, whereas early impedance drop serves as a reliable real-time indicator of steam-pop risk during bipolar RF ablation. These algorithms may help standardize bipolar ablation protocols by enabling prospective titration of energy delivery based on tissue characteristics and intraprocedural impedance monitoring.