State-space modeling (SSM) provides a general framework for many image reconstruction tasks. Error in a priori physiological knowledge of the imaging physics, can bring incorrectness to solutions. Modern deep-learning approaches show great promise but lack interpretability and rely on large amounts of labeled data. In this paper, we present a novel hybrid SSM framework for electrocardiographic imaging (ECGI) to leverage the advantage of state-space formulations in data-driven learning. We first leverage the physics-based forward operator to supervise the learning. We then introduce neural modeling of the transition function and the associated Bayesian filtering strategy. We applied the hybrid SSM framework to reconstruct electrical activity on the heart surface from body-surface potentials. In unsupervised settings of both in-silico and in-vivo data without cardiac electrical activity as the ground truth to supervise the learning, we demonstrated improved ECGI performances of the hybrid SSM framework trained from a small number of ECG observations in comparison to the fixed SSM. We further demonstrated that, when in-silico simulation data becomes available, mixed supervised and unsupervised training of the hybrid SSM achieved a further 40.6% and 45.6% improvements, respectively, in comparison to traditional ECGI baselines and supervised data-driven ECGI baselines for localizing the origin of ventricular activations in real data.

We have previously developed an intraprocedural system to identify VT exit sites in real-time on the patient-specific CT/MRI geometry using 3-lead ECGs (leads III, V2, V6) during catheter ablation for ventricular tachycardia (RAPID-VT).

Background We have previously developed an intraprocedural automatic arrhythmia‐origin localization (AAOL) system to identify idiopathic ventricular arrhythmia origins in real time using a 3‐lead ECG. The objective was to assess the localization accuracy of ventricular tachycardia (VT) exit and premature ventricular contraction (PVC) origin sites in patients with structural heart disease using the AAOL system. Methods and Results In retrospective and prospective case series studies, a total of 42 patients who underwent VT/PVC ablation in the setting of structural heart disease were recruited at 2 different centers. The AAOL system combines 120‐ms QRS integrals of 3 leads (III, V2, V6) with pace mapping to predict VT exit/PVC origin site and projects that site onto the patient‐specific electroanatomic mapping surface. VT exit/PVC origin sites were clinically identified by activation mapping and/or pace mapping. The localization error of the VT exit/PVC origin site was assessed by the distance between the clinically identified site and the estimated site. In the retrospective study of 19 patients with structural heart disease, the AAOL system achieved a mean localization accuracy of 6.5±2.6 mm for 25 induced VTs. In the prospective study with 23 patients, mean localization accuracy was 5.9±2.6 mm for 26 VT exit and PVC origin sites. There was no difference in mean localization error in epicardial sites compared with endocardial sites using the AAOL system (6.0 versus 5.8 mm, P =0.895). Conclusions The AAOL system achieved accurate localization of VT exit/PVC origin sites in patients with structural heart disease; its performance is superior to current systems, and thus, it promises to have potential clinical utility.

Objective: This work investigates the possibility of disentangled representation learning of inter-subject anatomical variations within electrocardiographic (ECG) data. Methods: Since ground truth anatomical factors are generally not known in clinical ECG for assessing the disentangling ability of the models, the presented work first proposes the SimECG data set, a 12-lead ECG data set procedurally generated with a controlled set of anatomical generative factors. Second, to perform such disentanglement, the presented method evaluates and compares deep generative models with latent density modeled by nonparametric Indian Buffet Process to account for the complex generative process of ECG data. Results: In the simulated data, the experiments demonstrate, for the first time, concrete evidence of the possibility to disentangle key generative anatomical factors within ECG data in separation from task-relevant generative factors. We achieve a disentanglement score of 92.1% while disentangling five anatomical generative factors and the task-relevant generative factor. In both simulated and real-data experiments, this work further provides quantitative evidence for the benefit of disentanglement learning on the downstream clinical task of localizing the origin of ventricular activation. Overall, the presented method achieves an improvement of around 18.5%, and 11.3% for the simulated dataset, and around 7.2%, and 3.6% for the real dataset, over baseline CNN, and standard generative model, respectively. Conclusion: These results demonstrate the importance as well as the feasibility of the disentangled representation learning of inter-subject anatomical variations within ECG data. Significance: This work suggests the important research direction to deal with the well-known challenge posed by the presence of significant inter-subject variations during an automated analysis of ECG data.

We have previously developed an intraprocedural automatic arrhythmia origin localization (AAOL) system to localize idiopathic ventricular arrhythmia origins onto patient-specific electroanatomic (EAM) surface in real time using a 3-lead ECG.

Probabilistic estimation of cardiac electrophysiological model parameters serves an important step toward model personalization and uncertain quantification. The expensive computation associated with these model simulations, however, makes direct Markov Chain Monte Carlo (MCMC) sampling of the posterior probability density function (pdf) of model parameters computationally intensive. Approximated posterior pdfs resulting from replacing the simulation model with a computationally efficient surrogate, on the other hand, have seen limited accuracy. In this study, we present a Bayesian active learning method to directly approximate the posterior pdf function of cardiac model parameters, in which we intelligently select training points to query the simulation model in order to learn the posterior pdf using a small number of samples. We integrate a generative model into Bayesian active learning to allow approximating posterior pdf of high-dimensional model parameters at the resolution of the cardiac mesh. We further introduce new acquisition functions to focus the selection of training points on better approximating the shape rather than the modes of the posterior pdf of interest. We evaluated the presented method in estimating tissue excitability in a 3D cardiac electrophysiological model in a range of synthetic and real-data experiments. We demonstrated its improved accuracy in approximating the posterior pdf compared to Bayesian active learning using regular acquisition functions, and substantially reduced computational cost in comparison to existing standard or accelerated MCMC sampling.

We have previously developed an intraprocedural RAPID-VT localization system to identify VT exit sites in real-time on the patient-specific CT/MRI mesh surface using 3-lead ECGs (leads III, V2, V6). To assess prospectively the localization performance and ablation outcomes of the RAPID-VT system. Eleven consecutive consenting patients with ischemic cardiomyopathy undergoing VT ablation had preprocedural CT/MRI scans. A reconstructed LV endocardial mesh surface was registered to the electroanatomic shell during the procedure and imported into the RAPID-VT program. During the procedure, predicted VT exit sites were used to guide mapping and ablation. The localization accuracy was quantified by comparing estimated sites against sites clinically identified using activation, entrainment and pace mapping. Thirteen accurately identified VT exit sites, pooled from 9 patients (mean procedure time 286.7±31.8 mins, were identified by using a combination of activation, entrainment and pace-mapping, and were used as the “ground truth” for comparisons. The RAPID-VT system had a mean localization error of 8.6±3.2 mm for the 13 VTs. Two patients were excluded due to technical problems with data acquisition. After targeting all inducible VTs, no VT was inducible at the end of the procedure in 7/9 patients. The RAPID-VT system based on the patient-specific CT/MRI mesh surface can be feasibly used in real-time to guide VT mapping and ablation.

Background: We developed a novel RIVAL system that consists of a CT-based computational simulation to identify VT circuits and an ECG-based automated approach to localize VT exit sites. Objective: Prospectively assess the ability of the RIVAL system to localize VT circuits and exits. Methods: Patients presenting for VT ablation were enrolled into the study. Pre-procedural cardiac CT was performed and used to conduct heart simulations for predicting VT circuits. The patient’s CT geometry with the predicted VT circuits was registered to the electroanatomic shell created during the procedure and imported into the RIVAL program. During the procedure, exit sites of induced VTs are localized in real-time using the 12-lead ECG onto the patient-specific CT surface. Predicted ablation regions obtained by combining RIVAL-predicted VT circuits with exit sites were analyzed offline to assess localization accuracy to the invasive VT ablation procedure. Results: Four patients with ischemic cardiomyopathy undergoing VT ablation had preprocedural cardiac CT. In P1, two VTs were induced during the procedure. The RIVAL predicted 2 VT circuits and achieved a mean localization error of 7.0mm for VT exit site prediction (Fig1). There was a spatial concordance between the predicted ablation areas and the clinical ablation regions. For P2, three VTs were induced during the VT ablation. A large area of scar was associated with 6 RIVAL-predicted VT circuits. The exit site localization accuracy could not be precisely quantitated because the VTs terminated with ablation at a mid-diastolic site. For P3, no VTs were inducible, however substrate modification (SM) was performed in the anterior-apical LV. The RIVAL predicted only one VT circuit located in the area where SM was performed. No VT was induced for P4 and the RIVAL did not predict any VT circuits. Conclusions: The RIVAL predicts the VT circuit and exit accurately, which may improve the precision of ablation therapies and procedure outcomes.

Introduction: We previously developed an LGE-MRI-based virtual-heart arrhythmia ablation targeting (VAAT) methodology to non-invasively determine potential ablation targets for infarct-related VT. However, it is unknown whether VAAT’s predictions correspond with surface ECG predictions. Hypothesis: We hypothesized that the VAAT predicted VT circuits and potential ablation lesions would co-localize with ECG-based VT-exit predictions from a previously validated population-derived automated VT exit localization (PAVEL) system. Methods: We retrospectively enrolled 5 post-infarct patients who underwent LV endocardial VT ablation and had pre-procedural 2D LGE-MRIs. The PAVEL system based on a population-derived statistical method was used to localize VT-exit sites onto one of 238 triangles on the patient-specific virtual-heart LV endocardial surface using 8 independent ECG leads (I, II, V1-V6). The VAAT methodology incorporating patient-specific scar and infarct border zone distributions was used to identify potential VT circuits and find ablation lesions. Results: Eleven induced VTs were analyzed. Ten VT-exit sites were localized onto the patient-specific virtual-heart LV endocardial surface by the PAVEL system, and were used for the comparisons. One VT-exit site was too basal to be localized onto the virtual-heart geometry. The spatial resolution of the 10 predicted VT-exit sites was 13.8 ± 1.8 mm. VAAT-predicted VT circuits and ablation lesions correlated well with all 10 predicted VT-exit sites. Lastly, VAAT ablation lesions fell within the regions ablated clinically. Conclusions: The VAAT-predicted VT circuits and ablation lesions matched VT-exit sites predicted by the surface ECG-based PAVEL system. Combining these two complementary technologies may improve accuracy for non-invasively identifying optimal ablation targets to increase ablation efficacy.

Background: To facilitate ablation of ventricular tachycardia (VT), an automated localization system to identify the site of origin of left ventricular activation in real time using the 12-lead ECG was developed. The objective of this study was to prospectively assess its accuracy. Methods: The automated site of origin localization system consists of 3 steps: (1) localization of ventricular segment based on population templates, (2) population-based localization within a segment, and (3) patient-specific site localization. Localization error was assessed by the distance between the known reference site and the estimated site. Results: In 19 patients undergoing 21 catheter ablation procedures of scar-related VT, site of origin localization accuracy was estimated using 552 left ventricular endocardial pacing sites pooled together and 25 VT-exit sites identified by contact mapping. For the 25 VT-exit sites, localization error of the population-based localization steps was within 10 mm. Patient-specific site localization achieved accuracy of within 3.5 mm after including up to 11 pacing (training) sites. Using 3 remotes (67.8±17.0 mm from the reference VT-exit site), and then 5 close pacing sites, resulted in localization error of 7.2±4.1 mm for the 25 identified VT-exit sites. In 2 emulated clinical procedure with 2 induced VTs, the site of origin localization system achieved accuracy within 4 mm. Conclusions: In this prospective validation study, the automated localization system achieved estimated accuracy within 10 mm and could thus provide clinical utility.

BACKGROUND To facilitate catheter ablation of ventricular tachycardia (VT), we previously developed an automated method to identify sources of left ventricular (LV) activation in real time using 12-lead electrocardiography (ECG), the accuracy of which depends on acquisition of a complete electroanatomic (EA) map. OBJECTIVE The purpose of this study was to assess the feasibility of using a registered cardiac computed tomogram (CT) rather than an EA map to permit real-time localization and avoid errors introduced by incomplete maps. METHODS Before LV VT ablation, 10 patients underwent CT imaging and 3-dimensional reconstruction of the cardiac surface to create a triangle mesh surface, which was registered to the EA map during the procedure and imported into custom localization software. The software uses QRS integrals from leads III, V-2, and V-6; derives personalized regression coefficients from pacing at >= 5 sites with known locations; and estimates the location of unknown activation sites on the 3-dimensional patient-specific LV endocardial surface. Localization accuracy was quantified for VT exit sites in millimeters by comparing the calculated against the known locations. RESULTS The VT exit site was identified for 20 VTs using activation and entrainment mapping, supplemented by pace-mapping at the scar margin. The automated localization software achieved incremental accuracy with additional pacing sites and had a mean localization error of 6.9 +/- 5.7 mm for the 20 VTs. CONCLUSION Patient-specific CT geometry is feasible for use in real-time automated localization of ventricular activation and may avoid reliance on a complete EA map.

Objective: This work presents a novel approach to handle the inter-subject variations existing in the population analysis of ECG, applied for localizing the origin of ventricular tachycardia (VT) from 12-lead electrocardiograms (ECGs). Methods: The presented method involves a factor disentangling sequential autoencoder (f-SAE) - realized in both long short-term memory (LSTM) and gated recurrent unit (GRU) networks - to learn to disentangle the inter-subject variations from the factor relating to the location of origin of VT. To perform such disentanglement, a pair-wise contrastive loss is introduced. Results: The presented methods are evaluated on ECG dataset with 1012 distinct pacing sites collected from scar-related VT patients during routine pace-mapping procedures. Experiments demonstrate that, for classifying the origin of VT into the predefined segments, the presented f-SAE improves the classification accuracy by 8.94% from using prescribed QRS features, by 1.5% from the supervised deep CNN network, and 5.15% from the standard SAE without factor disentanglement. Similarly, when predicting the coordinates of the VT origin, the presented f-SAE improves the performance by 2.25 mm from using prescribed QRS features, by 1.18 mm from the supervised deep CNN network and 1.6 mm from the standard SAE. Conclusion: These results demonstrate the importance as well as the feasibility of the presented f-SAE approach for separating inter-subject variations when using 12-lead ECG to localize the origin of VT. Significance: This work suggests the important research direction to deal with the well-known challenge posed by inter-subject variations during population analysis from ECG signals.

OBJECTIVES The objective of this study was to present a new system, the Automatic Arrhythmia Origin Localization (AAOL) system, which used incomplete electroanatomic mapping (EAM) for localization of idiopathic ventricular arrhythmia (IVA) origin on the patient-specific geometry of left ventricular, right ventricular, and neighboring vessels. The study assessed the accuracy of the system in localizing IVA source sites on cardiac structures where pace mapping is challenging. BACKGROUND An intraprocedural automated site of origin localization system was previously developed to identify the origin of early left ventricular activation by using 12-lead electrocardiograms (ECGs). However, it has limitations, as it could not identify the site of origin in the right ventricle and relied on acquiring a complete EAM. METHODS Twenty patients undergoing IVA catheter ablation had a 12-lead ECG recorded during clinical arrhythmia and during pacing at various locations identified on EAM geometries. The new system combined 3-lead (III, V-2, and V-6) 120-ms QRS integrals and patient-specific EAM geometry with pace mapping to predict the site of earliest ventricular activation. The predicted site was projected onto EAM geometry. RESULTS Twenty-three IVA origin sites were clinically identified by activation mapping and/or pace mapping (8, right ventricle; 15, left ventricle, including 8 from the posteromedial papillary muscle, 2 from the aortic root, and 1 from the distal coronary sinus). The new system achieved a mean localization accuracy of 3.6 mm for the 23 mapped IVAs. CONCLUSIONS The new intraprocedural AAOL system achieved accurate localization of IVA origin in ventricles and neighboring vessels, which could facilitate ablation procedures for patients with IVAs. (C) 2021 by the American College of Cardiology Foundation.

Introduction: Few intraprocedural localization systems have been developed to predict idiopathic ventricular arrhythmia (IVA) source sites. However, an accurate and bi-ventricular patient-specific automated site of origin localization system remains elusive. To address this issue, we have developed a new automatic arrhythmia origin localization (AAOL) system that determines the sites of earliest activation in both ventricles and provides superior accuracy. Hypothesis: We hypothesized that the AAOL system can use electroanatomic mapping (EAM) geometry and accurately localize IVA source sites on patient-specific geometry of LV, RV and neighboring vessels using 3-lead ECGs. Methods: Twenty patients undergoing IVA catheter ablation had a 12-lead ECG recorded during clinical arrhythmia and during pacing at various locations identified on EAM geometries. The AAOL system combined 3-lead (III, V2, V6) 120-ms QRS integrals and patient-specific EAM geometry with intracardiac pacing to predict the site of earliest ventricular activation. The predicted site was projected onto the EAM geometry using the EAM triangular-mesh site nearest to the tip of the predicted site. Results: Twenty-three IVA source sites were clinically identified by activation mapping and/or pace mapping (8 RV, 15 LV, including 8 from the posteromedial papillary muscle; 2 from the aortic root; and 1 from the distal coronary sinus). The new system achieved a mean localization accuracy of 3.6 mm for the 23 mapped IVAs (Figure 1D), better than that achieved by previous systems. Conclusions: The new AAOL system offers highly accurate localization of IVA source sites in both ventricles and neighboring vessels, which could facilitate ablation procedures for patients with IVAs.

To reconstruct electrical activity in the heart from body-surface electrocardiograms (ECGs) is an ill-posed inverse problem. Electrophysiological models have been found effective in regularizing these inverse problems by incorporating a priori knowledge about how the electrical potential in the heart propagates over time. However, these models suffer from model errors arising from, for example, parameters associated with tissue properties and the earliest sites of excitation. We present a Bayesian approach to simultaneously estimate transmembrane potential (TMP) signals and prior model errors, exploiting sparsity of the error in the gradient domain in the form of a novel sparse prior based on variational lower bound of the generalized Gaussian distribution. In synthetic and real-data experiments, we demonstrate the improvement of accuracy in TMP reconstruction brought by simultaneous model error estimation. We further provide theoretical and empirical justifications for the change of performances in the presented method at the presence of different model errors.

Objective: Ablation treatment of ventricular arrhythmias can be facilitated by pre-procedure planning aided by electrocardiographic inverse solution, which can help to localize the origin of arrhythmia. Our aim was to improve localization accuracy of the inverse solution by using a novel Bayesian approach. Methods: The inverse problem of electrocardiography was solved by reconstructing epicardial potentials from 120 body-surface electrocardiograms and from patient-specific geometry of the heart and torso for four patients suffering from scar-related ventricular tachycardia who underwent epicardial cathetermapping, which included pace-mapping. Simulations using dipole sources in patient-specific geometry were also performed. The proposed method, using dynamic spatio-temporal a priori constraints of the solution, was compared with classical Tikhonov methods based on fixed constraints. Results: The mean localization error of the proposed method for all available pacing sites (n = 78) was significantly smaller than that achieved by Tikhonov methods; specifically, the localization accuracy for pacing in the normal tissue (n = 17) was 8 +/- 6 mm (mean +/- SD) versus 13 +/- 9 mm (P < 0.00001) reported in the previous study using the same clinical data and Tikhonov regularization. Simulation experiments further supported these clinical findings. Conclusion: The promising results of in vivo and in silico experiments presented in this study provide a strong incentive to pursuing further investigation of data-driven Bayesian methods in solving the electrocardiographic inverse problem. Significance: The proposed approach to localizing origin of ventricular activation sequence may have important applications in pre-procedure assessment of arrhythmias and in guiding their ablation treatment.

Personalization of cardiac models involves the optimization of organ tissue properties that vary spatially over the non-Euclidean geometry model of the heart. To represent the high-dimensional (HD) unknown of tissue properties, most existing works rely on a low-dimensional (LD) partitioning of the geometrical model. While this exploits the geometry of the heart, it is of limited expressiveness to allow partitioning that is small enough for effective optimization. Recently, a variational auto-encoder (VAE) was utilized as a more expressive generative model to embed the HD optimization into the LD latent space. Its Euclidean nature, however, neglects the rich geometrical information in the heart. In this paper, we present a novel graph convolutional VAE to allow generative modeling of non-Euclidean data, and utilize it to embed Bayesian optimization of large graphs into a small latent space. This approach bridges the gap of previous works by introducing an expressive generative model that is able to incorporate the knowledge of spatial proximity and hierarchical compositionality of the underlying geometry. It further allows transferring of the learned features across different geometries, which was not possible with a regular VAE. We demonstrate these benefits of the presented method in synthetic and real data experiments of estimating tissue excitability in a cardiac electrophysiological model.

Introduction: Infusion-needle catheter ablation of VT within deep intramural substrate requires accurate identification of appropriate targets. We have developed an automated method for localizing ...

Deep learning networks have shown state-of-the-art performance in many image reconstruction problems. However, it is not well understood what properties of representation and learning may improve the generalization ability of the network. In this paper, we propose that the generalization ability of an encoder-decoder network for inverse reconstruction can be improved in two means. First, drawing from analytical learning theory, we theoretically show that a stochastic latent space will improve the ability of a network to generalize to test data outside the training distribution. Second, following the information bottleneck principle, we show that a latent representation minimally informative of the input data will help a network generalize to unseen input variations that are irrelevant to the output reconstruction. Therefore, we present a sequence image reconstruction network optimized by a variational approximation of the information bottleneck principle with stochastic latent space. In the application setting of reconstructing the sequence of cardiac transmembrane potential from body-surface potential, we assess the two types of generalization abilities of the presented network against its deterministic counterpart. The results demonstrate that the generalization ability of an inverse reconstruction network can be improved by stochasticity as well as the information bottleneck.