AUTHOR=Monaci Sofia , Gillette Karli , Puyol-Antón Esther , Rajani Ronak , Plank Gernot , King Andrew , Bishop Martin TITLE=Automated Localization of Focal Ventricular Tachycardia From Simulated Implanted Device Electrograms: A Combined Physics–AI Approach JOURNAL=Frontiers in Physiology VOLUME=Volume 12 - 2021 YEAR=2021 URL=https://www.frontiersin.org/journals/physiology/articles/10.3389/fphys.2021.682446 DOI=10.3389/fphys.2021.682446 ISSN=1664-042X ABSTRACT=Background. Focal ventricular tachycardia (VT) is a life-threating arrhythmia, responsible for high morbidity rates and sudden cardiac death (SCD). Radiofrequency ablation is the only curative therapy against incessant VT; however, its success is dependent upon accurate localisation of its source, which is highly invasive and time-consuming. Objective. The goal of our study is, as a proof-of-concept, demonstrating the possibility of utilising electrograms (EGM) recordings from cardiac implantable electronic devices (CIEDs). To achieve this, we utilise fast and accurate whole torso electrophysiological (EP) simulations in conjunction with convolutional neural networks (CNNs) to automate the localisation of focal VTs using simulated EGMs. Methods. A highly detailed 3D torso model was used to simulate ~4000 focal VTs, evenly distributed across the left ventricle (LV), utilising a rapid reaction-eikonal environment. Solutions were subsequently combined with lead field computations on the torso to derive accurate electrocardiograms (ECG) and EGM traces, which were used as inputs to CNNs to localise focal sources. We compared the localisation performance of a previously developed CNN architecture (Cartesian probability-based) with our novel CNN algorithm utilising universal ventricular coordinates (UVCs). Results. Implanted device EGMs successfully localised VT sources with localisation error (8.74mm) comparable to ECG-based localisation (6.69mm). Our novel UVC CNN architecture outperformed the existing Cartesian probability-based algorithm (errors = 4.06mm and 8.07mm for ECGs and EGMs, respectively). Overall, localisation was relatively insensitive to noise and changes in body compositions, however displacements in ECG electrodes and CIED leads caused performance to decrease (errors 16-25mm). Conclusions. EGM recordings from implanted devices may be used to successfully, and robustly, localize focal VT sources, and aid ablation planning.