A Model Population-Based Approach to Enhance the Detection of Premature Ventricular Contraction of ECGI

Jorge Sánchez1, Ines Llorente2, Santiago Ros3, Felipe Atienza4, Andreu M. Climent2, Maria de la Salud Guillem Sánchez2
1ITACA Institute, Universidad Politècnica de València, 2Universitat Politècnica de València, 3Gregorio Marañón Health Research Institute, 4Hospital General Universitario Gregorio Marañón (Cardiology Department)


Abstract

Premature ventricular contractions (PVCs) are a type of cardiac arrhythmia where heartbeats originate prematurely from the ventricles. Conventional diagnostics, like the 12-lead electrocardiogram, often fall short in precisely locating PVCs, which is crucial for effective treatment. This study introduces the application of electrocardiographic imaging (ECGI) and a novel computer model-based population database to enhance the accuracy of PVC localization and management. The research involved creating a comprehensive database of 618 simulated PVCs, particularly focusing on those originating from the outflow tract. To compute the local activation time (LAT) map, essential for pinpointing PVC origins, we employed an equivalent single-layer source model alongside Tikhonov of zero-order regularization. This allowed for the recovery of both extracellular endocardial and epicardial potentials. Body surface potentials at the torso surface (BSPM) were calculated using a boundary element method. A key development in this study is an estimation algorithm that analyzes the ECGI LAT map in conjunction with the BSPM to locate the origin of PVCs accurately. This algorithm was validated against a separate set of simulated data not included in the initial database. The results showed that standard ECGI typically had a mean geodesic distance error of 33.84±19.23mm in identifying the earliest activation sites. However, our PVC estimation algorithm reduced this error to 9.28±4.56mm, enhancing precision in localizing PVCs in the base, septum, and free wall regions, especially in the septal and basal areas. Additionally, when applied to patient data, our methodology decreased the localization error to 15.52mm compared to 36.76mm with standard ECGI. This research underscores the potential of using a simulated PVC population database and advanced ECGI techniques to improve the diagnosis and treatment of PVCs, marking a significant step forward in personalized cardiac arrhythmia management.