Reconstructed scar morphology in patient-specific computational heart models has limited impact on the identification of ablation targets through in-silico pace mapping

BACKGROUND: Patient-specific computational modeling for guiding ventricular tachycardia (VT) ablation often requires precise scar reconstruction to simulate reentrant circuits. However, this can be limited by the quality of scar imaging data. In-silico pace mapping, which simulates pacing rather than VT circuits, may offer a more robust approach to identifying ablation targets. OBJECTIVE: To investigate how the anatomical detail of scar reconstructions within computational image-based heart models influences the ability of in-silico pace mapping to identify VT origins. METHODS: VT was simulated in 15 patient-specific models reconstructed from high-resolution contrast-enhanced cardiac magnetic resonance (CMR). The obtained scar anatomy was then altered to mimic heart models constructed based on low-quality imaging and no-scar data. The ECG of each simulated VT was taken as input for the in-silico pace mapping approach, which involved pacing the heart at 1000 random sites surrounding the infarct. Correlations between the VT and paced ECGs were used to compute pace maps. The distance (d) between visually identified exit sites (ground truth) and pacing locations with the strongest correlation was used to assess accuracy of our in-silico approach. RESULTS: The performance of in-silico pace mapping was highest in high-resolution scar models (d = 7.3 ± 7.0 mm), but low-resolution and no-scar models still adequately located exit sites (d = 8.5 ± 6.5 mm and 13.3 ± 12.2 mm, respectively). CONCLUSION: In-silico pace mapping provides a reliable method for identifying VT ablation targets, showing relative insensitivity to scar reconstruction quality. This advantage may support its clinical translation over methods requiring explicit VT simulation.