P922Non-invasive characterisation of left atrial substrate in atrial fibrillation patients using multi-modal magnetic resonance imaging

Abstract

Introduction: Patient-specific catheter ablation protocols, targeting the regions hosting the re-entrant drivers (RDs) driving atrial fibrillation (AF), are expected to improve treatment success. Recent experimental and computational studies have shown that RDs are likely to stabilise in atrial regions with given characteristics, such as the border zones of patchy fibrosis or areas with a high gradient in atrial wall thickness. Magnetic resonance imaging (MRI) has the potential to non-invasively characterise both patchy fibrosis, showing as regions of high intensity in late gadolinium enhanced (LGE) images, and atrial wall thickness (AWT) using a novel black-blood PSIR scanning protocol. Aim(s): We present non-invasive tools for the characterization of the left atrial (LA) substrate in AF patients from MRI and explore correlations between left atrial wall thickness and fibrosis. Method(s): 5 AF patients were scanned using LGE and PSIR MRI protocols in the same scanning session, with data acquisition triggered to the same phase of the cardiac cycle (atrial diastole at the point of maximal atrial volume). LGE intensity maps and AWT maps were produced after manual segmentation and processing by experienced operators. For the AWT maps, we used the same sequence and protocol that we previously employed to compute an atlas of AWT in healthy subjects. We non-linearly registered all fibrosis and thickness maps to the space of the AWT atlas. This allowed the creation of atlases of AWT and normalised LGE intensity using the data from the 5 patients. We also computed the correlation between LGE enhancement and thickness in each patient across the LA. Result(s): High signal intensity regions on LGE (likely to correspond to patchy fibrosis) were found in most subjects in the posterior wall and in the anterior wall close to the appendage. The spatial localisation of thickness was more variable and, in general, lower in the posterior wall than in the rest of the LA. No statistically significant correlation was found between the spatial localisation of thickness and LGE enhancement in any patient (R2 of linear fit < 0.2, jrSpearmanj<0.13). Conclusion(s): Atrial wall thickness and LGE enhancement are not significantly correlated, suggesting that the two factors need to be considered independently in each AF patient for a comprehensive characterisation of the LA substrate. This structural information from MRI can be used to build patient-specific computational models for studying the dynamics of RDs and also has the potential to help inform patient-specific catheter ablation strategies.