A MRI-derived 3D patient specific model for fibrosis quantification in atrial fibrillation

Introduction: Atrial fibrillation (AF) is the most common type of arrhythmia and the mechanisms that sustain it are not yet clearly identified. Catheter ablation is a promising therapy for AF. However, to achieve durable restoration of sinus rhythm, multiple procedures may be required. Early studies have suggested both extensive atrial tissue fibrosis and association between scar gaps and pulmonary vein (PV) reconnection sites as possible causes of the poor outcomes of the AF catheter ablation. In this study, in order to assist the electrophysiologist in patient selection and ablation procedure planning, we developed a 3D patient–specific left atrium (LA) model integrating anatomical and structural information derived from magnetic resonance angiography (MRA) and delayed–enhanced MR imaging (DE–MRI). Materials and Methods: Thirty–five patients with paroxysmal AF were enrolled in the study and MRA and DE–MRI images were acquired. A patient–specific anatomical model was derived by MRA data, applying an edge–based level set approach guided by a phase–based edge detector (figure A, upper panels). A multimodality affine registration based on mutual information was then applied to register MRA into the spatial domain of DE–MRI (figure A, bottom panels). Once affine registration parameters were obtained, the corresponding intensity gray level information derived from the DE–MRI was overlapped on the registered 3D surface LA model, allowing the 3D visualization of LA fibrosis location and extent (figure B). In order to obtain a qualitative validation, the 3D LA models were compared with voltage maps reconstructed during the ablation procedures. Results: The 3D patient specific model obtained through MRA segmentation and registration of DE–MRI data was feasible in all patients. Time required for the analysis was about 30 min for each patient. An example of the qualitative comparison between high enhanced regions in the 3D LA model with fibrosis information and the low voltage areas in the electroanatomical map is shown in figure C. Conclusion: Preliminary qualitative validation of the 3D LA model including structural information seems a promising tool for a correct fibrosis localization and quantification. Next steps include assessment of the proposed tool to quantify scar location and extent for patient selection and catheter ablation planning. JOURNAL/ehjci/beta/01619449-201605001-00094/math_94MM1/v/2017-03-08T095554Z/r/image-png