Ventricular tachycardia site of origin prediction by a QRS axis-based algorithm confirmed by heterogenous tissue channel localization obtained from artificial-intelligence-powered CMR analysis

Abstract

Abstract Funding Acknowledgements Type of funding sources: None. Introduction A previously validated algorithm allows to accurately identify ventricular tachycardia (VT) exit site in a 17-segments American Heart Association left ventricle model in patients with structural heart disease (SHD). ADAS3D is a novel artificial-intelligence cardiac magnetic resonance (CMR) post-processing software which enables heterogeneous tissue channel (HTC) identification. Purpose The aim of this study is to compare inferred VT exit site between QRS-axis based model and HTC identification from CMR post-processing. Methods All subjects allocated to a CMR-guided or aided approach from a multicenter randomized clinical trial were analyzed. Patients with available 12-lead ECG of clinical VT obtained before ablation were analyzed in terms of concordance between VT site of origin inferred from a QRS-axis and location of HTCs in a 17-segments model. Results 10 out of 15 patients had 12-lead ECG of clinical VT (80% male, 60% with ischemic cardiomyopathy, mean age 63,5±16,95 years, mean tachycardia cycle length 373,7±99,55 ms). 7 out of 10 patients (70,00%) had at least one HTC located at VT exit site as predicted by a QRS axis-based algorithm. In all remaining cases (3 out of 10) HTC location was found in a segment adjacent to ECG-derived exit site. All patients underwent CMR-guided or CMR-aided VT ablation with acute clinical success (noninducibility of the clinical VT). Conclusion VT site of origin prediction via a QRS axis-based algorithm allows for reasonable prediction of HTC presence at inferred site of origin. Larger studies are needed in order to confirm such findings and their potential clinical implications.