| Abstract [eng] |
Background Images from computed tomography (CT) imported into specialized image processing systems perform image segmentation and integration with electroanatomic maps. Integrating non-invasive artificial intelligence (AI) and decrement evoked potential (DEEP) mapping into ablation procedures may improve substrate targeting in ischemic ventricular tachycardia (VT) ablation. Methods Patients (n=4) with ischemic cardiomyopathy and scar-related VT were included between 2024 and 2025 into the pilot observational study. Pre-procedural CT scans were processed using an AI-guided imaging platform to identify scar distribution and predicted isthmuses. High-density DEEP mapping was performed intra-procedurally. Ablation primarily targeted DEEP sites co-localizing AI-defined isthmuses. Non-inducibility was tested after the ablation of all primary targets using the 400/S2/S3/S4 stimulation protocol and pharmacological provocation testing. Results All patients were male with a mean age of 69.2 ± 5.7 years, mean left ventricular ejection fraction (LVEF) of 32.2 ± 13.2%, and a mean number of 13,67 ± 18,93 discharges of implantable cardioverter-defibrillators (ICDs). Patients presented with an average scar volume of 26.59 ± 6.96 ml, 10.5 ± 5.72 pre-defined VT-dependent isthmuses and 2 ± 1.41 primary inducible VTs per patient. Pre-procedural PAINESD risk score averaged 14.75 ± 5.12. Primarily ablation tactic resulted in VT non-inducibility in all patients. During a median follow-up of 10 ± 2.83 months (range 6-12 months), no sustained VT recurrences were observed. Freedom from non-sustained VT was 50% at 7 months. Conclusion This pilot experience suggests that combining AI-derived anatomical isthmus prediction with DEEP functional mapping is feasible and may enhance substrate targeting in ischemic VT ablation. Larger comparative studies are required to determine its impact on long-term outcomes. |
