LASSNet: A Four Steps Deep Neural Network for Left Atrial Segmentation and Scar Quantification.

Accurate quantification of left atrium (LA) scar in patients with atrial fibrillation is essential to guide successful ablation strategies. Prior to LA scar quantification, a proper LA cavity segmentation is required to ensure exact location of scar. Both tasks can be extremely time-consuming and are subject to inter-observer disagreements when done manually. We developed and validated a deep neural network to automatically segment the LA cavity and the LA scar. The global architecture uses a multi-network sequential approach in two stages which segment the LA cavity and the LA Scar. Each stage has two steps: a region of interest Neural Network and a refined segmentation network. We analysed the performances of our network according to different parameters and applied data triaging. 200+ late gadolinium enhancement magnetic resonance images were provided by the LAScarQS 2022 Challenge. Finally, we compared our performances for scar quantification to the literature and demonstrated improved performances.

Presence of Left Atrial Fibrosis May Contribute to Aberrant Hemodynamics and Increased Risk of Stroke in Atrial Fibrillation Patients.

Atrial fibrillation (AF) patients are at high risk of stroke, with the left atrial appendage (LAA) found to be the most common site of clot formation. Presence of left atrial (LA) fibrosis has also been associated with higher stroke risk. However, the mechanisms for increased stroke risk in patients with atrial fibrotic remodeling are poorly understood. We sought to explore these mechanisms using fluid dynamic analysis and to test the hypothesis that the presence of LA fibrosis leads to aberrant hemodynamics in the LA, contributing to increased stroke risk in AF patients. We retrospectively collected late-gadolinium-enhanced MRI (LGE-MRI) images of eight AF patients (four persistent and four paroxysmal) and reconstructed their 3D LA surfaces. Personalized computational fluid dynamic simulations were performed, and hemodynamics at the LA wall were quantified by wall shear stress (WSS, friction of blood), oscillatory shear index (OSI, temporal directional change of WSS), endothelial cell activation potential (ECAP, ratio of OSI and WSS), and relative residence time (RRT, residence time of blood near the LA wall). For each case, these hemodynamic metrics were compared between fibrotic and non-fibrotic portions of the wall. Our results showed that WSS was lower, and OSI, ECAP, and RRT was higher in the fibrotic region as compared to the non-fibrotic region, with ECAP (p = 0.001) and RRT (p = 0.002) having significant differences. Case-wise analysis showed that these differences in hemodynamics were statistically significant for seven cases. Furthermore, patients with higher fibrotic burden were exposed to larger regions of high ECAP, which represents regions of low WSS and high OSI. Consistently, high ECAP in the vicinity of the fibrotic wall suggest that local blood flow was slow and oscillating that represents aberrant hemodynamic conditions, thus enabling prothrombotic conditions for circulating blood. AF patients with high LA fibrotic burden had more prothrombotic regions, providing more sites for potential clot formation, thus increasing their risk of stroke.

Characterizing the arrhythmogenic substrate in personalized models of atrial fibrillation: sensitivity to mesh resolution and pacing protocol in AF models.

AIMS: Computationally guided persistent atrial fibrillation (PsAF) ablation has emerged as an alternative to conventional treatment planning. To make this approach scalable, computational cost and the time required to conduct simulations must be minimized while maintaining predictive accuracy. Here, we assess the sensitivity of the process to finite-element mesh resolution. We also compare methods for pacing site distribution used to evaluate inducibility arrhythmia sustained by re-entrant drivers (RDs). METHODS AND RESULTS: Simulations were conducted in low- and high-resolution models (average edge lengths: 400/350 µm) reconstructed from PsAF patients' late gadolinium enhancement magnetic resonance imaging scans. Pacing was simulated from 80 sites to assess RD inducibility. When pacing from the same site led to different outcomes in low-/high-resolution models, we characterized divergence dynamics by analysing dissimilarity index over time. Pacing site selection schemes prioritizing even spatial distribution and proximity to fibrotic tissue were evaluated. There were no RD sites observed in low-resolution models but not high-resolution models, or vice versa. Dissimilarity index analysis suggested that differences in simulation outcome arising from differences in discretization were the result of isolated conduction block incidents in one model but not the other; this never led to RD sites unique to one mesh resolution. Pacing site selection based on fibrosis proximity led to the best observed trade-off between number of stimulation locations and predictive accuracy. CONCLUSION: Simulations conducted in meshes with 400 µm average edge length and ∼40 pacing sites proximal to fibrosis are sufficient to reveal the most comprehensive possible list of RD sites, given feasibility constraints.

Left Atrial Enhancement Correlates With Myocardial Conduction Velocity in Patients With Persistent Atrial Fibrillation.

BACKGROUND: Conduction velocity (CV) heterogeneity and myocardial fibrosis both promote re-entry, but the relationship between fibrosis as determined by left atrial (LA) late-gadolinium enhanced cardiac magnetic resonance imaging (LGE-CMRI) and CV remains uncertain. OBJECTIVE: Although average CV has been shown to correlate with regional LGE-CMRI in patients with persistent AF, we test the hypothesis that a localized relationship exists to underpin LGE-CMRI as a minimally invasive tool to map myocardial conduction properties for risk stratification and treatment guidance. METHOD: 3D LA electroanatomic maps during LA pacing were acquired from eight patients with persistent AF following electrical cardioversion. Local CVs were computed using triads of concurrently acquired electrograms and were co-registered to allow correlation with LA wall intensities obtained from LGE-CMRI, quantified using normalized intensity (NI) and image intensity ratio (IIR). Association was evaluated using multilevel linear regression. RESULTS: An association between CV and LGE-CMRI intensity was observed at scales comparable to the size of a mapping electrode: -0.11 m/s per unit increase in NI (P < 0.001) and -0.96 m/s per unit increase in IIR (P < 0.001). The magnitude of this change decreased with larger measurement area. Reproducibility of the association was observed with NI, but not with IIR. CONCLUSION: At clinically relevant spatial scales, comparable to area of a mapping catheter electrode, LGE-CMRI correlates with CV. Measurement scale is important in accurately quantifying the association of CV and LGE-CMRI intensity. Importantly, NI, but not IIR, accounts for changes in the dynamic range of CMRI and enables quantitative reproducibility of the association.

Preprocedure Application of Machine Learning and Mechanistic Simulations Predicts Likelihood of Paroxysmal Atrial Fibrillation Recurrence Following Pulmonary Vein Isolation.

BACKGROUND: Pulmonary vein isolation (PVI) is an effective treatment strategy for patients with atrial fibrillation (AF), but many experience AF recurrence and require repeat ablation procedures. The goal of this study was to develop and evaluate a methodology that combines machine learning (ML) and personalized computational modeling to predict, before PVI, which patients are most likely to experience AF recurrence after PVI. METHODS: This single-center retrospective proof-of-concept study included 32 patients with documented paroxysmal AF who underwent PVI and had preprocedural late gadolinium enhanced magnetic resonance imaging. For each patient, a personalized computational model of the left atrium simulated AF induction via rapid pacing. Features were derived from pre-PVI late gadolinium enhanced magnetic resonance images and from results of simulations of AF induction. The most predictive features were used as input to a quadratic discriminant analysis ML classifier, which was trained, optimized, and evaluated with 10-fold nested cross-validation to predict the probability of AF recurrence post-PVI. RESULTS: In our cohort, the ML classifier predicted probability of AF recurrence with an average validation sensitivity and specificity of 82% and 89%, respectively, and a validation area under the curve of 0.82. Dissecting the relative contributions of simulations of AF induction and raw images to the predictive capability of the ML classifier, we found that when only features from simulations of AF induction were used to train the ML classifier, its performance remained similar (validation area under the curve, 0.81). However, when only features extracted from raw images were used for training, the validation area under the curve significantly decreased (0.47). CONCLUSIONS: ML and personalized computational modeling can be used together to accurately predict, using only pre-PVI late gadolinium enhanced magnetic resonance imaging scans as input, whether a patient is likely to experience AF recurrence following PVI, even when the patient cohort is small.

Accurate Conduction Velocity Maps and Their Association With Scar Distribution on Magnetic Resonance Imaging in Patients With Postinfarction Ventricular Tachycardias.

BACKGROUND: Characterizing myocardial conduction velocity (CV) in patients with ischemic cardiomyopathy (ICM) and ventricular tachycardia (VT) is important for understanding the patient-specific proarrhythmic substrate of VTs and therapeutic planning. The objective of this study is to accurately assess the relation between CV and myocardial fibrosis density on late gadolinium-enhanced cardiac magnetic resonance imaging (LGE-CMR) in patients with ICM. METHODS: We enrolled 6 patients with ICM undergoing VT ablation and 5 with structurally normal left ventricles (controls) undergoing premature ventricular contraction or VT ablation. All patients underwent LGE-CMR and electroanatomic mapping (EAM) in sinus rhythm (2960 electroanatomic mapping points analyzed). We estimated CV from electroanatomic mapping local activation time using the triangulation method that provides an accurate estimate of CV as it accounts for the direction of wavefront propagation. We evaluated the association between LGE-CMR intensity and CV with multilevel linear mixed models. RESULTS: Median CV in patients with ICM and controls was 0.41 m/s and 0.65 m/s, respectively. In patients with ICM, CV in areas with no visible fibrosis was 0.81 m/s (95% CI, 0.59-1.12 m/s). For each 25% increase in normalized LGE intensity, CV decreased by 1.34-fold (95% CI, 1.25-1.43). Dense scar areas have, on average, 1.97- to 2.66-fold slower CV compared with areas without dense scar. Ablation lesions that terminated VTs were localized in areas of slow conduction on CV maps. CONCLUSIONS: CV is inversely associated with LGE-CMR fibrosis density in patients with ICM. Noninvasive derivation of CV maps from LGE-CMR is feasible. Integration of noninvasive CV maps with electroanatomic mapping during substrate mapping has the potential to improve procedural planning and outcomes. Visual Overview: A visual overview is available for this article.

Understanding AF Mechanisms Through Computational Modelling and Simulations.

AF is a progressive disease of the atria, involving complex mechanisms related to its initiation, maintenance and progression. Computational modelling provides a framework for integration of experimental and clinical findings, and has emerged as an essential part of mechanistic research in AF. The authors summarise recent advancements in development of multi-scale AF models and focus on the mechanistic links between alternations in atrial structure and electrophysiology with AF. Key AF mechanisms that have been explored using atrial modelling are pulmonary vein ectopy; atrial fibrosis and fibrosis distribution; atrial wall thickness heterogeneity; atrial adipose tissue infiltration; development of repolarisation alternans; cardiac ion channel mutations; and atrial stretch with mechano-electrical feedback. They review modelling approaches that capture variability at the cohort level and provide cohort-specific mechanistic insights. The authors conclude with a summary of future perspectives, as envisioned for the contributions of atrial modelling in the mechanistic understanding of AF.

Computationally guided personalized targeted ablation of persistent atrial fibrillation.

Atrial fibrillation (AF)-the most common arrhythmia-significantly increases the risk of stroke and heart failure. Although catheter ablation can restore normal heart rhythms, patients with persistent AF who develop atrial fibrosis often undergo multiple failed ablations, and thus increased procedural risks. Here, we present personalized computational modelling for the reliable predetermination of ablation targets, which are then used to guide the ablation procedure in patients with persistent AF and atrial fibrosis. First, we show that a computational model of the atria of patients identifies fibrotic tissue that, if ablated, will not sustain AF. Then, we report the results of integrating the target ablation sites in a clinical mapping system and testing its feasibility in ten patients with persistent AF. The computational prediction of ablation targets avoids lengthy electrical mapping and could improve the accuracy and efficacy of targeted AF ablation in patients while eliminating the need for repeat procedures.

Voltage during atrial fibrillation is superior to voltage during sinus rhythm in localizing areas of delayed enhancement on magnetic resonance imaging: An assessment of the posterior left atrium in patients with persistent atrial fibrillation.

BACKGROUND: Bipolar electrogram voltage during sinus rhythm (VSR) has been used as a surrogate for atrial fibrosis in guiding catheter ablation of persistent atrial fibrillation (AF), but the fixed rate and wavefront characteristics present during sinus rhythm may not accurately reflect underlying functional vulnerabilities responsible for AF maintenance. OBJECTIVE: The purpose of this study was determine whether, given adequate temporal sampling, the spatial distribution of mean AF voltage (VmAF) better correlates with delayed-enhancement magnetic resonance imaging (MRI-DE)-detected atrial fibrosis than VSR. METHODS: AF was mapped (8 seconds) during index ablation for persistent AF (20 patients) using a 20-pole catheter (660 ± 28 points/map). After cardioversion, VSR was mapped (557 ± 326 points/map). Electroanatomic and MRI-DE maps were co-registered in 14 patients. RESULTS: The time course of VmAF was assessed from 1-40 AF cycles (∼8 seconds) at 1113 locations. VmAF stabilized with sampling >4 seconds (mean voltage error 0.05 mV). Paired point analysis of VmAF from segments acquired 30 seconds apart (3667 sites; 15 patients) showed strong correlation (r = 0.95; P <.001). Delayed enhancement (DE) was assessed across the posterior left atrial (LA) wall, occupying 33% ± 13%. VmAF distributions were (median [IQR]) 0.21 [0.14-0.35] mV in DE vs 0.52 [0.34-0.77] mV in non-DE regions. VSR distributions were 1.34 [0.65-2.48] mV in DE vs 2.37 [1.27-3.97] mV in non-DE. VmAF threshold of 0.35 mV yielded sensitivity of 75% and specificity of 79% in detecting MRI-DE compared with 63% and 67%, respectively, for VSR (1.8-mV threshold). CONCLUSION: The correlation between low-voltage and posterior LA MRI-DE is significantly improved when acquired during AF vs sinus rhythm. With adequate sampling, mean AF voltage is a reproducible marker reflecting the functional response to the underlying persistent AF substrate.

Arrhythmogenic propensity of the fibrotic substrate after atrial fibrillation ablation: a longitudinal study using magnetic resonance imaging-based atrial models.

AIMS: Inadequate modification of the atrial fibrotic substrate necessary to sustain re-entrant drivers (RDs) may explain atrial fibrillation (AF) recurrence following failed pulmonary vein isolation (PVI). Personalized computational models of the fibrotic atrial substrate derived from late gadolinium enhanced (LGE)-magnetic resonance imaging (MRI) can be used to non-invasively determine the presence of RDs. The objective of this study is to assess the changes of the arrhythmogenic propensity of the fibrotic substrate after PVI. METHODS AND RESULTS: Pre- and post-ablation individualized left atrial models were constructed from 12 AF patients who underwent pre- and post-PVI LGE-MRI, in six of whom PVI failed. Pre-ablation AF sustained by RDs was induced in 10 models. RDs in the post-ablation models were classified as either preserved or emergent. Pre-ablation models derived from patients for whom the procedure failed exhibited a higher number of RDs and larger areas defined as promoting RD formation when compared with atrial models from patients who had successful ablation, 2.6 ± 0.9 vs. 1.8 ± 0.2 and 18.9 ± 1.6% vs. 13.8 ± 1.5%, respectively. In cases of successful ablation, PVI eliminated completely the RDs sustaining AF. Preserved RDs unaffected by ablation were documented only in post-ablation models of patients who experienced recurrent AF (2/5 models); all of these models had also one or more emergent RDs at locations distinct from those of pre-ablation RDs. Emergent RDs occurred in regions that had the same characteristics of the fibrosis spatial distribution (entropy and density) as regions that harboured RDs in pre-ablation models. CONCLUSION: Recurrent AF after PVI in the fibrotic atria may be attributable to both preserved RDs that sustain AF pre- and post-ablation, and the emergence of new RDs following ablation. The same levels of fibrosis entropy and density underlie the pro-RD propensity in both pre- and post-ablation substrates.

A novel approach to mapping the atrial ganglionated plexus network by generating a distribution probability atlas.

INTRODUCTION: The ganglionated plexuses (GPs) of the intrinsic cardiac autonomic system are implicated in arrhythmogenesis. GP localization by stimulation of the epicardial fat pads to produce atrioventricular dissociating (AVD) effects is well described. We determined the anatomical distribution of the left atrial GPs that influence atrioventricular (AV) dissociation. METHODS AND RESULTS: High frequency stimulation was delivered through a Smart-Touch catheter in the left atrium of patients undergoing atrial fibrillation (AF) ablation. Three dimensional locations of points tested throughout the entire chamber were recorded on the CARTO™ system. Impact on the AV conduction was categorized as ventricular asystole, bradycardia, or no effect. CARTO maps were exported, registered, and transformed onto a reference left atrial geometry using a custom software, enabling data from multiple patients to be overlaid. In 28 patients, 2108 locations were tested and 283 sites (13%) demonstrated (AVD-GP) effects. There were 10 AVD-GPs (interquartile range, 11.5) per patient. Eighty percent (226) produced asystole and 20% (57) showed bradycardia. The distribution of the two groups was very similar. Highest probability of AVD-GPs (>20%) was identified in: inferoseptal portion (41%) and right inferior pulmonary vein base (30%) of the posterior wall, right superior pulmonary vein antrum (31%). CONCLUSION: It is feasible to map the entire left atrium for AVD-GPs before AF ablation. Aggregated data from multiple patients, producing a distribution probability atlas of AVD-GPs, identified three regions with a higher likelihood for finding AVD-GPs and these matched the histological descriptions. This approach could be used to better characterize the autonomic network.

Automated fiducial point selection for reducing registration error in the co-localisation of left atrium electroanatomic and imaging data.

Registration of electroanatomic surfaces and segmented images for the co-localisation of structural and functional data typically requires the manual selection of fiducial points, which are used to initialise automated surface registration. The identification of equivalent points on geometric features by the human eye is heavily subjective, and error in their selection may lead to distortion of the transformed surface and subsequently limit the accuracy of data co-localisation. We propose that the manual trimming of the pulmonary veins through the region of greatest geometrical curvature, coupled with an automated angle-based fiducial-point selection algorithm, significantly reduces target registration error compared with direct manual selection of fiducial points.

A technique for visualising three-dimensional left atrial cardiac activation data in two dimensions with minimal distance distortion.

Electro-anatomic mapping and medical imaging systems, used during clinical procedures for treatment of atrial arrhythmias, frequently record and display measurements on an anatomical surface of the left atrium. As such, obtaining a complete picture of activation necessitates simultaneous views from multiple angles. In addition, post-processing of three-dimensional surface data is challenging, since algorithms are typically applicable to planar or volumetric data. We applied a surface flattening methodology to medical imaging data and electro-anatomic mapping data to generate a two-dimensional representation that best preserves distances, since the calculation of many clinically relevant metrics, including conduction velocity and rotor trajectory identification require an accurate representation of distance. Distance distortions were small and improved upon exclusion of the pulmonary veins. The technique is demonstrated using maps of local activation time, based on clinical data, and plotting rotor-core trajectories, using simulated data.

An automated algorithm for determining conduction velocity, wavefront direction and origin of focal cardiac arrhythmias using a multipolar catheter.

Determining locations of focal arrhythmia sources and quantifying myocardial conduction velocity (CV) are two major challenges in clinical catheter ablation cases. CV, wave-front direction and focal source location can be estimated from multipolar catheter data, but currently available methods are time-consuming, limited to specific electrode configurations, and can be inaccurate. We developed automated algorithms to rapidly identify CV from multipolar catheter data with any arrangement of electrodes, whilst providing estimates of wavefront direction and focal source position, which can guide the catheter towards a focal arrhythmic source. We validated our methods using simulations on realistic human left atrial geometry. We subsequently applied them to clinically-acquired intracardiac electrogram data, where CV and wavefront direction were accurately determined in all cases, whilst focal source locations were correctly identified in 2/3 cases. Our novel automated algorithms can potentially be used to guide ablation of focal arrhythmias in real-time in cardiac catheter laboratories.

A software platform for the comparative analysis of electroanatomic and imaging data including conduction velocity mapping.

Electroanatomic mapping systems collect increasingly large quantities of spatially-distributed electrical data which may be potentially further scrutinized post-operatively to expose mechanistic properties which sustain and perpetuate atrial fibrillation. We describe a modular software platform, developed to post-process and rapidly analyse data exported from electroanatomic mapping systems using a range of existing and novel algorithms. Imaging data highlighting regions of scar can also be overlaid for comparison. In particular, we describe the conduction velocity (CV) mapping algorithm used to highlight wavefront behaviour. CV was found to be particularly sensitive to the spatial distribution of the triangulation points and corresponding activation times. A set of geometric conditions were devised for selecting suitable triangulations of the electrogram set for generating CV maps.