Conference Agenda

Cardiac radiofrequency ablation (RFA), a cornerstone treatment for arrhythmias, suffers from limited understanding of lesion formation, hindering optimal outcomes. Current RFA models often oversimplify the complex multiphysics interactions within heterogeneous domains, neglecting crucial factors. This research introduces a high-fidelity, multiphysics, and multi-domain computational framework designed to enhance RFA treatment and minimize complications. The framework integrates heat transfer, electrostatics, fluid dynamics, and a three-state cell-death model across electrode, fluid, and tissue regions. Notably, some processes are confined to specific compartments (e.g., fluid dynamics in the fluid, cell-death in the tissue), while others (e.g., electrostatics, heat transfer) span multiple domains. To achieve accurate and scalable simulations, we employ physics- and domain-decomposition (DD) approaches. Specifically, Dirichlet-Neumann and Optimized Schwarz-like DD methods are explored, supported by rigorous convergence analysis. Implemented using high-order numerical methods within the MFEM library, and enhanced by efficient partially assembled operators and ongoing GPU acceleration, this framework demonstrates significant computational efficiency. Importantly, the framework's design allows for extensibility, making it a versatile platform for RF simulations across diverse tissue types (e.g., kidney, uterine, hepatic) and energy sources (RF, microwave, ultrasound, laser), as well as a range of other biomedical modeling applications (e.g., degradation of bio-resorbable stents). By advancing computational modeling and data assimilation, this research aims to bridge the gap between theoretical simulations and clinical practice, facilitating the development of more effective, personalized, and safer ablation therapies.

Background: Pulsed field ablation (PFA) is a new non-thermal ablation method being rapidly adopted for treatment of cardiac arrhythmia. The goal of PFA is to achieve sufficient electric field strength in the target region to produce the desired lesion depth while minimizing thermal damage.

Objective: Use numerical modelling to determine the lethal electric field threshold (LET) and thermal damage for PFA in healthy ventricles and to predict the extent of PFA lesions in infarcted ventricles.

Methods: We used fifteen 40-50 kg Yorkshire swine (10 healthy, 5 infarcted) in the study. PFA was performed using a two-catheter setup: a focal ablation catheter in the left ventricle (LV) and a return catheter in the inferior vena cava (IVC). In healthy ventricles, the pulse amplitude (1000-1500 V) and the number of pulse trains (1, 4, 8, and 16) were varied. In infarcted ventricles, a fixed protocol (1500 V, 8 trains) was used.

The numerical models were developed in COMSOL Multiphysics (v6.1).

To model PFA in healthy ventricles we used schematic biventricular geometry, constructed based on MRI, accounting for cardiac tissue anisotropy. LET was determined by calculating the electric field distribution for each lesion location and determining the threshold at which the predicted lesion volume matched the volume observed on late gadolinium enhancement (LGE) MRI. To calculate temperature increase we first calculated the blood flow in the LV and IVC for a high and low blood flow condition in a stationary study and use the solution in the coupled electro-thermal time-dependent study. Potential thermal damage was assessed using the Arrhenius integral and a thermal dose threshold of 1 s at ≥ 55 °C.

To model PFA in infarcted ventricles pre-ablation LGE MRI were processed using ADAS 3D software to obtain animal-specific ventricular geometry and scar tissue distribution, which was used to assign different electrical conductivities to the healthy myocardium and scar tissue in the model. Stationary study was used to determine the effect of scar tissue on the electric field distribution in LV.

Results: We determined LET for each PFA lesion in healthy tissue at 24 hours, 7 days, and 6 weeks post-ablation. Median LET varied with the number of pulse trains; at 7 days, thresholds were 725, 520, 484, and 394 V/cm for 1, 4, 8, and 16 pulse trains, respectively. Median LET was lower at 24 hours than at 7 days and 6 weeks, reflecting larger lesion volumes observed on LGE MRI at the earlier time point.

The maximum predicted thermal damage volume was minimal - 13 mm³ (less than 2% of total lesion volume) for the highest dose pulse protocol (16 trains) under low blood-flow condition - consistent with the absence of a thermal signature on MRI.

In infarcted ventricles, numerical simulations demonstrated PFA’s ability to create lesions through dense scar tissue. Predicted lesion extents agreed closely with histology.

Conclusion: We determined the LET for ventricular tissue in vivo, showing its dependence on the number of pulse trains and the time of evaluation. Numerical modelling and MRI confirmed the non-thermal nature of our PFA protocols.

In the landscape of thermal ablation strategies, laser ablation (LA) has emerged as a minimally invasive approach that employs focused laser energy to induce hyperthermia, effectively destroying pathological tissue. This mechanism, based on a rapid temperature increase, has demonstrated high spatial selectivity. While LA has been widely used in oncology, its potential for treating cardiac arrhythmias by targeting ectopic foci is gaining increasing attention. Laser light is absorbed by the tissue and converted into heat, resulting in localized thermal damage to the targeted area. Despite its clinical promise, predictive modeling tools for LA in cardiac applications remain limited, especially when compared to the well-developed models available for radiofrequency ablation.

Temperature distribution plays a pivotal role in determining lesion size and therapeutic efficacy. Small deviations in thermal delivery may lead to incomplete ablation or collateral damage to nearby healthy tissue. In cardiac tissue, the anisotropic and heterogeneous structure further complicates the modeling of heat propagation. Moreover, the optical and thermal properties of myocardial tissue significantly affect energy deposition patterns, making accurate, physics-informed computational modeling essential for virtual planning and outcome prediction.

This study proposes a comprehensive numerical framework tailored for simulating laser-tissue interactions in cardiac environments, aiming to enhance in silico planning capabilities for interventional procedures. A three-dimensional idealized cardiac tissue domain is modeled, incorporating rotational anisotropy and simulating the fully coupled optical–thermal response using a custom finite element implementation. The framework evaluates and contrasts several heat conduction models, starting with the classical Pennes bioheat equation and extending to more advanced formulations, including the Generalized Fourier (GF) model and the Dual-Phase Lag (DPL) model. These higher-order approaches account for finite thermal propagation speed and capture microstructural effects often neglected in traditional Fourier-based models.

To assess cellular damage, the thermal models are integrated with a multiscale three-state cell death dynamics model, which distinguishes between healthy, reversibly injured, and irreversibly damaged cells. This allows a more nuanced evaluation of ablation outcomes compared to conventional thresholds such as the 50 °C isotherm. Parametric analyses reveal that higher-order thermal models provide a more realistic prediction of lesion development, especially under fast heating conditions.

The results underscore the critical role of cardiac tissue anisotropy and validate the potential of advanced thermal modeling in improving lesion prediction accuracy. These findings contribute to the advancement of digital twin technologies in cardiology, offering a foundation for future integration of LA protocols into personalized simulation platforms aimed at guiding clinical decision-making in the treatment of cardiac arrhythmias.

Patients with ischemic cardiomyopathy and previous myocardial infarction often develop regions of fibrotic scar tissue in the heart, which can create a structural environment that supports reentrant ventricular tachycardia (VT) [1]. Accurately identifying these arrhythmogenic regions is essential to guide catheter ablation therapy effectively. Late gadolinium enhancement cardiac magnetic resonance imaging (LGE-CMR) allows for non-invasive visualization of myocardial scarring [2], while electroanatomical mapping (EAM) identifies low-voltage areas that indicate damaged or fibrotic tissue [3]. However, differences can exist between what is seen on MRI and what is detected by voltage mapping, due in part to anatomical factors like wall thickness. Recent work has shown that computational models based on MRI data can help identify patient-specific VT circuits and potential targets for ablation [4].

We developed a novel workflow to better characterize the arrhythmogenic substrate in patients with left ventricular myocardial infarction. LGE-CMR images were used to segment and classify scar regions into dense core and surrounding border zones, assigning different electrical properties—either non-conductive (dense scar) or with reduced conductivity (border zone)—based on MRI signal intensity. These data were then combined with 3D electroanatomical mapping of the left ventricle, obtained with the Carto 3 system, to merge structural and electrical information into a unified model.

Using the reconstructed geometry, we simulated electrical activation across the heart tissue through a simplified electrophysiological model [5], applying programmed stimulation at various points on the endocardial surface to test for reentrant circuit dynamics.

This study introduces a robust and integrated modeling approach that combines imaging, electro-anatomical mapping, and simulation to improve understanding of post-infarction VT. By linking structural and electrical information to a single model, this workflow offers a powerful tool for non-invasive identification of critical conduction pathways and potential VT circuits, with significant implications for improving the planning and precision of ablation procedures.

References

[1] R. Lo, K. K. M. Chia, and H. H. Hsia, «Ventricular Tachycardia in Ischemic Heart Disease», Card Electrophysiol Clin, vol. 9, fasc. 1, pp. 25–46, Mar. 2017.

[2] S. Kuruvilla, N. Adenaw, A. B. Katwal, M. J. Lipinski, C. M. Kramer, and M. Salerno, «Late Gadolinium Enhancement on Cardiac Magnetic Resonance Predicts Adverse Cardiovascular Outcomes in Nonischemic Cardiomyopathy: A Systematic Review and Meta-Analysis», Circ: Cardiovascular Imaging, vol. 7, fasc. 2, pp. 250–258, Mar. 2014.

[3] P. Compagnucci et al., «Recent advances in three-dimensional electroanatomical mapping guidance for the ablation of complex atrial and ventricular arrhythmias», J Interv Card Electrophysiol, vol. 61, fasc. 1, pp. 37–43, Jun. 2021.

[4] H. Ashikaga et al., «Feasibility of image-based simulation to estimate ablation target in human ventricular arrhythmia», Heart Rhythm, vol. 10, fasc. 8, pp. 1109–1116, Aug. 2013.

[5] A. J. Pullan, L. K. Cheng, M. L. Buist. Mathematically Modelling The Electrical Activity Of The Heart: From Cell To Body Surface And Back Again. 2005.