Evaluating effects of atrial structure on re-entrant drivers for atrial fibrillation using image-based computational approaches

Student thesis: Doctoral ThesisDoctor of Philosophy


Atrial fibrillation (AF), the most common sustained cardiac arrhythmia, affects over 33 million people worldwide. Catheter ablation (CA) has become a first-line treatment for drug-refractory AF and is particularly effective in paroxysmal AF patients in the early stages of the disease. However, CA has high recurrence rates in patients with chronic forms of the disease. This is due to the empirical nature of the procedure and lack of mechanistic knowledge of optimal ablation sites and strategies in these patients, whose atria is altered by AF-induced electrical and structural remodelling. The purpose of the work presented in this thesis is to investigate the influence of patient-specific atrial structure on electrophysiological function, specifically the dynamics of re-entrant drivers (RDs) of AF, and to establish a mechanistic link between the atrial structure and location of RDs. Establishing such a link would allow for non-invasive identification of the RD locations, which could ultimately guide CA in chronic AF patients. 
The first part of the study focused on atrial wall thickness (AWT) and to a lesser extent on atrial fibrosis, both of which can be reconstructed from medical imaging data and have been linked to atrial structural remodelling underlying the progression of AF. The effect of the atrial structure was investigated using two sets of computational models: 1) a simple model of an atrial slab with a step-change in AWT and a synthetic fibrotic patch, and 2) patient-specific atrial models with geometry and fibrosis reconstructed from magnetic resonance imaging (MRI) of 6 AF patients. The simple atrial slab model demonstrated that RDs drift towards and then along the AWT step. Furthermore, in the presence of a fibrotic tissue patch near the step, the RDs were attracted to the patch, and the ultimate RD location was determined by both fibrosis and AWT. In patient-specific 3D atrial models, the behaviour of the RDs was governed by the interaction between AWT and fibrosis in the right atrium (RA), owing to its trabecular structure with large AWT gradients. In the left atrium (LA) with more uniform AWT, the extensive fibrosis distribution played a more prominent role in influencing the RD dynamics. 
In the following part, RD locations in LA were investigated by: 1) developing image-based 3D atrial models from patient MRI data, 2) applying the models to dissect the mechanistic links between atrial structure and the RD dynamics and 3) using the modelling outcomes to quantify the ultimate patient-specific RD locations. Simulations in the patient-specific models revealed that the RD dynamics were strongly influenced by the amount and spatial distribution of fibrotic tissue. RDs were typically found at fibrotic regions in AF patients from Utah 3 and 4 categories with high fibrotic burden (>25%), but more often near the pulmonary veins (PVs) in patients from Utah 2 category. The RDs anchored to specific, relatively small regions, labelled as target areas, with a high percentage of target areas located within the fibrotic tissue region in patients from Utah 3 and 4 categories. The patient-specific target areas showed that areas that harboured the RDs were much smaller than the entire fibrotic areas, indicating potential targets for therapy. Finally, CA strategies based on the knowledge of target areas were evaluated in-silico to terminate RDs efficiently. Ablation strategies that connected the target areas with linear lesions to the PVs or the mitral valve have superior anti-fibrillatory effect compared to ablating the target areas alone, as well as compared to clinically applied strategies such as the PV isolation. 
Thus, the novel image-based modelling workflow presented in this thesis has been applied to dissect multiple effects of atrial structure on the genesis of RDs, providing a deeper understanding of the mechanisms of AF sustenance and paving the way to designing patient specific CA treatments.
Date of Award1 Jul 2020
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorOleg Aslanidi (Supervisor), Marta Varela (Supervisor) & Tobias Schaeffter (Supervisor)

Cite this