Calibrating cardiac electrophysiology models using latent Gaussian processes on atrial manifolds.

Models of electrical excitation and recovery in the heart have become increasingly detailed, but have yet to be used routinely in the clinical setting to guide personalized intervention in patients. One of the main challenges is calibrating models from the limited measurements that can be made in a patient during a standard clinical procedure. In this work, we propose a novel framework for the probabilistic calibration of electrophysiology parameters on the left atrium of the heart using local measurements of cardiac excitability. Parameter fields are represented as Gaussian processes on manifolds and are linked to measurements via surrogate functions that map from local parameter values to measurements. The posterior distribution of parameter fields is then obtained. We show that our method can recover parameter fields used to generate localised synthetic measurements of effective refractory period. Our methodology is applicable to other measurement types collected with clinical protocols, and more generally for calibration where model parameters vary over a manifold.

Machine Learned Cellular Phenotypes in Cardiomyopathy Predict Sudden Death.

RATIONALE: Susceptibility to VT/VF (ventricular tachycardia/fibrillation) is difficult to predict in patients with ischemic cardiomyopathy either by clinical tools or by attempting to translate cellular mechanisms to the bedside. OBJECTIVE: To develop computational phenotypes of patients with ischemic cardiomyopathy, by training then interpreting machine learning of ventricular monophasic action potentials (MAPs) to reveal phenotypes that predict long-term outcomes. METHODS AND RESULTS: We recorded 5706 ventricular MAPs in 42 patients with coronary artery disease and left ventricular ejection fraction ≤40% during steady-state pacing. Patients were randomly allocated to independent training and testing cohorts in a 70:30 ratio, repeated K=10-fold. Support vector machines and convolutional neural networks were trained to 2 end points: (1) sustained VT/VF or (2) mortality at 3 years. Support vector machines provided superior classification. For patient-level predictions, we computed personalized MAP scores as the proportion of MAP beats predicting each end point. Patient-level predictions in independent test cohorts yielded c-statistics of 0.90 for sustained VT/VF (95% CI, 0.76-1.00) and 0.91 for mortality (95% CI, 0.83-1.00) and were the most significant multivariate predictors. Interpreting trained support vector machine revealed MAP morphologies that, using in silico modeling, revealed higher L-type calcium current or sodium-calcium exchanger as predominant phenotypes for VT/VF. CONCLUSIONS: Machine learning of action potential recordings in patients revealed novel phenotypes for long-term outcomes in ischemic cardiomyopathy. Such computational phenotypes provide an approach which may reveal cellular mechanisms for clinical outcomes and could be applied to other conditions.

Economic evaluation of a dedicated cardiac resynchronisation therapy preassessment clinic.

INTRODUCTION: Patient evaluation before cardiac resynchronisation therapy (CRT) remains heterogeneous across centres and it is suspected a proportion of patients with unfavourable characteristics proceed to implantation. We developed a unique CRT preassessment clinic (CRT PAC) to act as a final review for patients already considered for CRT. We hypothesised that this clinic would identify some patients unsuitable for CRT through updated investigations and review. The purpose of this analysis was to determine whether the CRT PAC led to savings for the National Health Service (NHS). METHODS: A decision tree model was made to evaluate two clinical pathways; (1) standard of care where all patients initially seen in an outpatient cardiology clinic proceeded directly to CRT and (2) management of patients in CRT PAC. RESULTS: 244 patients were reviewed in the CRT PAC; 184 patients were eligible to proceed directly for implantation and 48 patients did not meet consensus guidelines for CRT so were not implanted. Following CRT, 82.4% of patients had improvement in their clinical composite score and 57.7% had reduction in left ventricular end-systolic volume ≥15%. Using the decision tree model, by reviewing patients in the CRT PAC, the total savings for the NHS was £966 880. Taking into consideration the additional cost of the clinic and by applying this model structure throughout the NHS, the potential savings could be as much as £39 million. CONCLUSIONS: CRT PAC appropriately selects patients and leads to substantial savings for the NHS. Adopting this clinic across the NHS has the potential to save £39 million.

Quantifying atrial anatomy uncertainty from clinical data and its impact on electro-physiology simulation predictions.

Patient-specific computational models of structure and function are increasingly being used to diagnose disease and predict how a patient will respond to therapy. Models of anatomy are often derived after segmentation of clinical images or from mapping systems which are affected by image artefacts, resolution and contrast. Quantifying the impact of uncertain anatomy on model predictions is important, as models are increasingly used in clinical practice where decisions need to be made regardless of image quality. We use a Bayesian probabilistic approach to estimate the anatomy and to quantify the uncertainty about the shape of the left atrium derived from Cardiac Magnetic Resonance images. We show that we can quantify uncertain shape, encode uncertainty about the left atrial shape due to imaging artefacts, and quantify the effect of uncertain shape on simulations of left atrial activation times.

Probabilistic Interpolation of Uncertain Local Activation Times on Human Atrial Manifolds.

OBJECTIVE: Local activation time (LAT) mapping of the atria is important for targeted treatment of atrial arrhythmias, but current methods do not interpolate on the atrial manifold and neglect uncertainties associated with LAT observations. In this paper, we describe novel methods to, first, quantify uncertainties in LAT arising from bipolar electrogram analysis and assignment of electrode recordings to the anatomical mesh, second, interpolate uncertain LAT measurements directly on left atrial manifolds to obtain complete probabilistic activation maps, and finally, interpolate LAT jointly across both the manifold and different S1-S2 pacing protocols. METHODS: A modified center of mass approach was used to process bipolar electrograms, yielding a LAT estimate and error distribution from the electrogram morphology. An error distribution for assigning measurements to the anatomical mesh was estimated. Probabilistic LAT maps were produced by interpolating on a left atrial manifold using Gaussian Markov random fields, taking into account observation errors and characterizing LAT predictions by their mean and standard deviation. This approach was extended to interpolate across S1-S2 pacing protocols. RESULTS: We evaluated our approach using recordings from three patients undergoing atrial ablation. Cross-validation showed consistent and accurate prediction of LAT observations both at different locations on the left atrium and for different S1-S2 intervals. SIGNIFICANCE: Interpolation of scalar and vector fields across anatomical structures from point measurements is a challenging problem in biomedical engineering, compounded by uncertainties in measurements and meshes. New methods and approaches are required, and in this paper, we have demonstrated an effective method for probabilistic interpolation of uncertain LAT.

Evaluation of a real-time magnetic resonance imaging-guided electrophysiology system for structural and electrophysiological ventricular tachycardia substrate assessment.

AIMS: Potential advantages of real-time magnetic resonance imaging (MRI)-guided electrophysiology (MR-EP) include contemporaneous three-dimensional substrate assessment at the time of intervention, improved procedural guidance, and ablation lesion assessment. We evaluated a novel real-time MR-EP system to perform endocardial voltage mapping and assessment of delayed conduction in a porcine ischaemia-reperfusion model. METHODS AND RESULTS: Sites of low voltage and slow conduction identified using the system were registered and compared to regions of late gadolinium enhancement (LGE) on MRI. The Sorensen-Dice similarity coefficient (DSC) between LGE scar maps and voltage maps was computed on a nodal basis. A total of 445 electrograms were recorded in sinus rhythm (range: 30-186) using the MR-EP system including 138 electrograms from LGE regions. Pacing captured at 103 sites; 47 (45.6%) sites had a stimulus-to-QRS (S-QRS) delay of ≥40 ms. Using conventional (0.5-1.5 mV) bipolar voltage thresholds, the sensitivity and specificity of voltage mapping using the MR-EP system to identify MR-derived LGE was 57% and 96%, respectively. Voltage mapping had a better predictive ability in detecting LGE compared to S-QRS measurements using this system (area under curve: 0.907 vs. 0.840). Using an electrical threshold of 1.5 mV to define abnormal myocardium, the total DSC, scar DSC, and normal myocardium DSC between voltage maps and LGE scar maps was 79.0 ± 6.0%, 35.0 ± 10.1%, and 90.4 ± 8.6%, respectively. CONCLUSION: Low-voltage zones and regions of delayed conduction determined using a real-time MR-EP system are moderately associated with LGE areas identified on MRI.

Analytical approaches for myocardial fibrillation signals.

Atrial and ventricular fibrillation are complex arrhythmias, and their underlying mechanisms remain widely debated and incompletely understood. This is partly because the electrical signals recorded during myocardial fibrillation are themselves complex and difficult to interpret with simple analytical tools. There are currently a number of analytical approaches to handle fibrillation data. Some of these techniques focus on mapping putative drivers of myocardial fibrillation, such as dominant frequency, organizational index, Shannon entropy and phase mapping. Other techniques focus on mapping the underlying myocardial substrate sustaining fibrillation, such as voltage mapping and complex fractionated electrogram mapping. In this review, we discuss these techniques, their application and their limitations, with reference to our experimental and clinical data. We also describe novel tools including a new algorithm to map microreentrant circuits sustaining fibrillation.

Unraveling the Underlying Arrhythmia Mechanism in Persistent Atrial Fibrillation: Results From the STARLIGHT Study.

BACKGROUND: The mechanisms that initiate and sustain persistent atrial fibrillation are not well characterized. Ablation results remain significantly worse than in paroxysmal atrial fibrillation in which the mechanism is better understood and subsequent targeted therapy has been developed. The aim of this study was to characterize and quantify patterns of activation during atrial fibrillation using contact mapping. METHODS: Patients with persistent atrial fibrillation (n=14; mean age, 61±8 years; ejection fraction, 59±10%) underwent simultaneous biatrial contact mapping with 64 electrode catheters. The atrial electrograms were transformed into phase, and subsequent spatiotemporal mapping was performed to identify phase singularities (PSs). RESULTS: PSs were located in both atria, but we observed more PSs in the left atrium compared with the right atrium (779±302, 552±235; P=0.015). Although some PSs of duration sufficient to complete >1 rotation were detected, the maximum PS duration was only 1150 ms, and the vast majority (97%) of PSs persisted for too short a period to complete a full rotation. Although in selected patients there was evidence of PS local clustering, overall, PSs were distributed globally throughout both chambers with no clear anatomic predisposition. In a subset of patients (n=7), analysis was repeated using an alternative established atrial PS mapping technique, which confirmed our initial findings. CONCLUSIONS: No sustained rotors or localized drivers were detected, and instead, the mechanism of arrhythmia maintenance was consistent with the multiple wavelet hypothesis, with passive activation of short-lived rotational activity. CLINICAL TRIAL REGISTRATION: URL: https://www.clinicaltrials.gov. Unique identifier: NCT01765075.

Voltage and pace-capture mapping of linear ablation lesions overestimates chronic ablation gap size.

Aims: Conducting gaps in lesion sets are a major reason for failure of ablation procedures. Voltage mapping and pace-capture have been proposed for intra-procedural identification of gaps. We aimed to compare gap size measured acutely and chronically post-ablation to macroscopic gap size in a porcine model. Methods and results: Intercaval linear ablation was performed in eight Göttingen minipigs with a deliberate gap of ∼5 mm left in the ablation line. Gap size was measured by interpolating ablation contact force values between ablation tags and thresholding at a low force cut-off of 5 g. Bipolar voltage mapping and pace-capture mapping along the length of the line were performed immediately, and at 2 months, post-ablation. Animals were euthanized and gap sizes were measured macroscopically. Voltage thresholds to define scar were determined by receiver operating characteristic analysis as <0.56 mV (acutely) and <0.62 mV (chronically). Taking the macroscopic gap size as gold standard, error in gap measurements were determined for voltage, pace-capture, and ablation contact force maps. All modalities overestimated chronic gap size, by 1.4 ± 2.0 mm (ablation contact force map), 5.1 ± 3.4 mm (pace-capture), and 9.5 ± 3.8 mm (voltage mapping). Error on ablation contact force map gap measurements were significantly less than for voltage mapping (P = 0.003, Tukey's multiple comparisons test). Chronically, voltage mapping and pace-capture mapping overestimated macroscopic gap size by 11.9 ± 3.7 and 9.8 ± 3.5 mm, respectively. Conclusion: Bipolar voltage and pace-capture mapping overestimate the size of chronic gap formation in linear ablation lesions. The most accurate estimation of chronic gap size was achieved by analysis of catheter-myocardium contact force during ablation.

Local activation time sampling density for atrial tachycardia contact mapping: how much is enough?

Aims: Local activation time (LAT) mapping forms the cornerstone of atrial tachycardia diagnosis. Although anatomic and positional accuracy of electroanatomic mapping (EAM) systems have been validated, the effect of electrode sampling density on LAT map reconstruction is not known. Here, we study the effect of chamber geometry and activation complexity on optimal LAT sampling density using a combined in silico and in vivo approach. Methods and results: In vivo 21 atrial tachycardia maps were studied in three groups: (1) focal activation, (2) macro-re-entry, and (3) localized re-entry. In silico activation was simulated on a 4×4cm atrial monolayer, sampled randomly at 0.25-10 points/cm2 and used to re-interpolate LAT maps. Activation patterns were studied in the geometrically simple porcine right atrium (RA) and complex human left atrium (LA). Activation complexity was introduced into the porcine RA by incomplete inter-caval linear ablation. In all cases, optimal sampling density was defined as the highest density resulting in minimal further error reduction in the re-interpolated maps. Optimal sampling densities for LA tachycardias were 0.67 ± 0.17 points/cm2 (focal activation), 1.05 ± 0.32 points/cm2 (macro-re-entry) and 1.23 ± 0.26 points/cm2 (localized re-entry), P = 0.0031. Increasing activation complexity was associated with increased optimal sampling density both in silico (focal activation 1.09 ± 0.14 points/cm2; re-entry 1.44 ± 0.49 points/cm2; spiral-wave 1.50 ± 0.34 points/cm2, P < 0.0001) and in vivo (porcine RA pre-ablation 0.45 ± 0.13 vs. post-ablation 0.78 ± 0.17 points/cm2, P = 0.0008). Increasing chamber geometry was also associated with increased optimal sampling density (0.61 ± 0.22 points/cm2 vs. 1.0 ± 0.34 points/cm2, P = 0.0015). Conclusion: Optimal sampling densities can be identified to maximize diagnostic yield of LAT maps. Greater sampling density is required to correctly reveal complex activation and represent activation across complex geometries. Overall, the optimal sampling density for LAT map interpolation defined in this study was ∼1.0-1.5 points/cm2.

Personalized Models of Human Atrial Electrophysiology Derived From Endocardial Electrograms.

OBJECTIVE: Computational models represent a novel framework for understanding the mechanisms behind atrial fibrillation (AF) and offer a pathway for personalizing and optimizing treatment. The characterization of local electrophysiological properties across the atria during procedures remains a challenge. The aim of this work is to characterize the regional properties of the human atrium from multielectrode catheter measurements. METHODS: We propose a novel method that characterizes regional electrophysiology properties by fitting parameters of an ionic model to conduction velocity and effective refractory period restitution curves obtained by a s1-s2 pacing protocol applied through a multielectrode catheter. Using an in-silico dataset we demonstrate that the fitting method can constrain parameters with a mean error of 21.9 ± 16.1% and can replicate conduction velocity and effective refractory curves not used in the original fitting with a relative error of 4.4 ± 6.9%. RESULTS: We demonstrate this parameter estimation approach on five clinical datasets recorded from AF patients. Recordings and parametrization took approx. 5 and 6 min, respectively. Models fitted restitution curves with an error of ~ 5% and identify a unique parameter set. Tissue properties were predicted using a two-dimensional atrial tissue sheet model. Spiral wave stability in each case was predicted using tissue simulations, identifying distinct stable (2/5), meandering and breaking up (2/5), and unstable self-terminating (1/5) spiral tip patterns for different cases. CONCLUSION AND SIGNIFICANCE: We have developed and demonstrated a robust and rapid approach for personalizing local ionic models from a clinically tractable.

Simultaneous display of multiple three-dimensional electrophysiological datasets (dot mapping).

AIMS: Complex ablation procedures are supported by accurate representation of an increasing variety of electrophysiological and imaging data within electroanatomic mapping systems (EMS). This study aims to develop a novel method for representing multiple complementary datasets on a single cardiac chamber model. Validation of the system and its application to both atrial and ventricular arrhythmias is examined. METHODS AND RESULTS: Dot mapping was conceived to display multiple datasets by utilizing quantitative surface shading to represent one dataset and finely spaced dots to represent others. Dot positions are randomized within triangular (surface meshes) or tetrahedral (volumetric meshes) simplices making the approach directly transferrable to contemporary EMS. Test data representing uniform electrical activation (n = 10) and focal scarring (n = 10) were used to test dot mapping data perception accuracy. User experience of dot mapping with atrial and ventricular clinical data is evaluated. Dot mapping ensured constant screen dot density for regions of uniform dataset values, regardless of user manipulation of the cardiac chamber. Perception accuracy of dot mapping was equivalent to colour mapping for both propagation direction (1.5 ± 1.8 vs. 4.8 ± 5.3°, P = 0.24) and focal source localization (1.1 ± 0.7 vs. 1.4 ± 0.5 mm, P = 0.88). User acceptance testing revealed equivalent diagnostic accuracy and display fidelity when compared with colour mapping. CONCLUSION: Dot mapping provides the unique ability to display multiple datasets from multiple sources on a single cardiac chamber model. The visual combination of multiple datasets may facilitate interpretation of complex electrophysiological and imaging data.

Analysis of lead placement optimization metrics in cardiac resynchronization therapy with computational modelling.

AIMS: The efficacy of cardiac resynchronization therapy (CRT) is known to vary considerably with pacing location, however the most effective set of metrics by which to select the optimal pacing site is not yet well understood. Computational modelling offers a powerful methodology to comprehensively test the effect of pacing location in silico and investigate how to best optimize therapy using clinically available metrics for the individual patient. METHODS AND RESULTS: Personalized computational models of cardiac electromechanics were used to perform an in silico left ventricle (LV) pacing site optimization study as part of biventricular CRT in three patient cases. Maps of response to therapy according to changes in total activation time (ΔTAT) and acute haemodynamic response (AHR) were generated and compared with preclinical metrics of electrical function, strain, stress, and mechanical work to assess their suitability for selecting the optimal pacing site. In all three patients, response to therapy was highly sensitive to pacing location, with laterobasal locations being optimal. ΔTAT and AHR were found to be correlated (ρ < -0.80), as were AHR and the preclinical activation time at the pacing site (ρ ≥ 0.73), however pacing in the last activated site did not result in the optimal response to therapy in all cases. CONCLUSION: This computational modelling study supports pacing in laterobasal locations, optimizing pacing site by minimizing paced QRS duration and pacing in regions activated late at sinus rhythm. Results demonstrate information content is redundant using multiple preclinical metrics. Of significance, the correlation of AHR with ΔTAT indicates that minimization of QRSd is a promising metric for optimization of lead placement.

Beneficial Effect on Cardiac Resynchronization From Left Ventricular Endocardial Pacing Is Mediated by Early Access to High Conduction Velocity Tissue: Electrophysiological Simulation Study.

BACKGROUND: Cardiac resynchronization therapy (CRT) delivered via left ventricular (LV) endocardial pacing (ENDO-CRT) is associated with improved acute hemodynamic response compared with LV epicardial pacing (EPI-CRT). The role of cardiac anatomy and physiology in this improved response remains controversial. We used computational electrophysiological models to quantify the role of cardiac geometry, tissue anisotropy, and the presence of fast endocardial conduction on myocardial activation during ENDO-CRT and EPI-CRT. METHODS AND RESULTS: Cardiac activation was simulated using the monodomain tissue excitation model in 2-dimensional (2D) canine and human and 3D canine biventricular models. The latest activation times (LATs) for LV endocardial and biventricular epicardial tissue were calculated (LVLAT and TLAT), as well the percentage decrease in LATs for endocardial (en) versus epicardial (ep) LV pacing (defined as %dLV=100×(LVLATep-LVLATen)/LVLATep and %dT=100×(TLATep-TLATen)/TLATep, respectively). Normal canine cardiac anatomy is responsible for %dLV and %dT values of 7.4% and 5.5%, respectively. Concentric and eccentric remodeled anatomies resulted in %dT values of 15.6% and 1.3%, respectively. The 3D biventricular-paced canine model resulted in %dLV and %dT values of -7.1% and 1.5%, in contrast to the experimental observations of 16% and 11%, respectively. Adding fast endocardial conduction to this model altered %dLV and %dT to 13.1% and 10.1%, respectively. CONCLUSIONS: Our results provide a physiological explanation for improved response to ENDO-CRT. We predict that patients with viable fast-conducting endocardial tissue or distal Purkinje network or both, as well as concentric remodeling, are more likely to benefit from reduced ATs and increased synchrony arising from endocardial pacing.