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

A review of uncertainty quantification in medical image analysis: Probabilistic and non-probabilistic methods

The comprehensive integration of machine learning healthcare models within clinical practice remains suboptimal, notwithstanding the proliferation of high-performing solutions reported in the literature. A predominant factor hindering widespread adoption pertains to an insufficiency of evidence affirming the reliability of the aforementioned models. Recently, uncertainty quantification methods have been proposed as a potential solution to quantify the reliability of machine learning models and thus increase the interpretability and acceptability of the results. In this review, we offer a comprehensive overview of the prevailing methods proposed to quantify the uncertainty inherent in machine learning models developed for various medical image tasks. Contrary to earlier reviews that exclusively focused on probabilistic methods, this review also explores non-probabilistic approaches, thereby furnishing a more holistic survey of research pertaining to uncertainty quantification for machine learning models. Analysis of medical images with the summary and discussion on medical applications and the corresponding uncertainty evaluation protocols are presented, which focus on the specific challenges of uncertainty in medical image analysis. We also highlight some potential future research work at the end. Generally, this review aims to allow researchers from both clinical and technical backgrounds to gain a quick and yet in-depth understanding of the research in uncertainty quantification for medical image analysis machine learning models.

Application of uncertainty quantification to artificial intelligence in healthcare: A review of last decade (2013-2023)

Uncertainty estimation in healthcare involves quantifying and understanding the inherent uncertainty or variability associated with medical predictions, diagnoses, and treatment outcomes. In this era of Artificial Intelligence (AI) models, uncertainty estimation becomes vital to ensure safe decision-making in the medical field. Therefore, this review focuses on the application of uncertainty techniques to machine and deep learning models in healthcare. A systematic literature review was conducted using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Our analysis revealed that Bayesian methods were the predominant technique for uncertainty quantification in machine learning models, with Fuzzy systems being the second most used approach. Regarding deep learning models, Bayesian methods emerged as the most prevalent approach, finding application in nearly all aspects of medical imaging. Most of the studies reported in this paper focused on medical images, highlighting the prevalent application of uncertainty quantification techniques using deep learning models compared to machine learning models. Interestingly, we observed a scarcity of studies applying uncertainty quantification to physiological signals. Thus, future research on uncertainty quantification should prioritize investigating the application of these techniques to physiological signals. Overall, our review highlights the significance of integrating uncertainty techniques in healthcare applications of machine learning and deep learning models. This can provide valuable insights and practical solutions to manage uncertainty in real-world medical data, ultimately improving the accuracy and reliability of medical diagnoses and treatment recommendations.

AugmentA: Patient-specific augmented atrial model generation tool

Digital twins of patients' hearts are a promising tool to assess arrhythmia vulnerability and to personalize therapy. However, the process of building personalized computational models can be challenging and requires a high level of human interaction. We propose a patient-specific Augmented Atria generation pipeline (AugmentA) as a highly automated framework which, starting from clinical geometrical data, provides ready-to-use atrial personalized computational models. AugmentA identifies and labels atrial orifices using only one reference point per atrium. If the user chooses to fit a statistical shape model to the input geometry, it is first rigidly aligned with the given mean shape before a non-rigid fitting procedure is applied. AugmentA automatically generates the fiber orientation and finds local conduction velocities by minimizing the error between the simulated and clinical local activation time (LAT) map. The pipeline was tested on a cohort of 29 patients on both segmented magnetic resonance images (MRI) and electroanatomical maps of the left atrium. Moreover, the pipeline was applied to a bi-atrial volumetric mesh derived from MRI. The pipeline robustly integrated fiber orientation and anatomical region annotations in 38.4 ± 5.7 s. In conclusion, AugmentA offers an automated and comprehensive pipeline delivering atrial digital twins from clinical data in procedural time.

Quantifying the impact of shape uncertainty on predicted arrhythmias

BACKGROUND

Personalised computer models are increasingly used to diagnose cardiac arrhythmias and tailor treatment. Patient-specific models of the left atrium are often derived from pre-procedural imaging of anatomy and fibrosis. These images contain noise that can affect simulation predictions. There are few computationally tractable methods for propagating uncertainties from images to clinical predictions.

METHOD

We describe the left atrium anatomy using our Bayesian shape model that captures anatomical uncertainty in medical images and has been validated on 63 independent clinical images. This algorithm describes the left atrium anatomy using Nmodes=15 principal components, capturing 95% of the shape variance and calculated from 70 clinical cardiac magnetic resonance (CMR) images. Latent variables encode shape uncertainty: we evaluate their posterior distribution for each new anatomy. We assume a normally distributed prior. We use the unscented transform to sample from the posterior shape distribution. For each sample, we assign the local material properties of the tissue using the projection of late gadolinium enhancement CMR (LGE-CMR) onto the anatomy to estimate local fibrosis. To test which activation patterns an atrium can sustain, we perform an arrhythmia simulation for each sample. We consider 34 possible outcomes (31 macro-re-entries, functional re-entry, atrial fibrillation, and non-sustained arrhythmia). For each sample, we determine the outcome by comparing pre- and post-ablation activation patterns following a cross-field stimulus.

RESULTS

We create patient-specific atrial electrophysiology models of ten patients. We validate the mean and standard deviation maps from the unscented transform with the same statistics obtained with 12,000 Monte Carlo (ground truth) samples. We found discrepancies <3% and <2% for the mean and standard deviation for fibrosis burden and activation time, respectively. For each patient case, we then compare the predicted outcome from a model built on the clinical data (deterministic approach) with the probability distribution obtained from the simulated samples. We found that the deterministic approach did not predict the most likely outcome in 80% of the cases. Finally, we estimate the influence of each source of uncertainty independently. Fixing the anatomy to the posterior mean and maintaining uncertainty in fibrosis reduced the prediction of self-terminating arrhythmias from ≃14% to ≃7%. Keeping the fibrosis fixed to the sample mean while retaining uncertainty in shape decreased the prediction of substrate-driven arrhythmias from ≃33% to ≃18% and increased the prediction of macro-re-entries from ≃54% to ≃68%.

CONCLUSIONS

We presented a novel method for propagating shape uncertainty in atrial models through to uncertainty in numerical simulations. The algorithm takes advantage of the unscented transform to compute the output distribution of the outcomes. We validated the unscented transform as a viable sampling strategy to deal with anatomy uncertainty. We then showed that the prediction computed with a deterministic model does not always coincide with the most likely outcome. Finally, we found that shape uncertainty affects the predictions of macro-re-entries, while fibrosis uncertainty affects the predictions of functional re-entries.

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.

Pacemaking function of two simplified cell models

Simplified nonlinear models of biological cells are widely used in computational electrophysiology. The models reproduce qualitatively many of the characteristics of various organs, such as the heart, brain, and intestine. In contrast to complex cellular ion-channel models, the simplified models usually contain a small number of variables and parameters, which facilitates nonlinear analysis and reduces computational load. In this paper, we consider pacemaking variants of the Aliev-Panfilov and Corrado two-variable excitable cell models. We conducted a numerical simulation study of these models and investigated the main nonlinear dynamic features of both isolated cells and 1D coupled pacemaker-excitable systems. Simulations of the 2D sinoatrial node and 3D intestine tissue as application examples of combined pacemaker-excitable systems demonstrated results similar to obtained previously. The uniform formulation for the conventional excitable cell models and proposed pacemaker models allows a convenient and easy implementation for the construction of personalized physiological models, inverse tissue modeling, and development of real-time simulation systems for various organs that contain both pacemaker and excitable cells.

Space-time shape uncertainties in the forward and inverse problem of electrocardiography

In electrocardiography, the "classic" inverse problem is the reconstruction of electric potentials at a surface enclosing the heart from remote recordings at the body surface and an accurate description of the anatomy. The latter being affected by noise and obtained with limited resolution due to clinical constraints, a possibly large uncertainty may be perpetuated in the inverse reconstruction. The purpose of this work is to study the effect of shape uncertainty on the forward and the inverse problem of electrocardiography. To this aim, the problem is first recast into a boundary integral formulation and then discretised with a collocation method to achieve high convergence rates and a fast time to solution. The shape uncertainty of the domain is represented by a random deformation field defined on a reference configuration. We propose a periodic-in-time covariance kernel for the random field and approximate the Karhunen-Loève expansion using low-rank techniques for fast sampling. The space-time uncertainty in the expected potential and its variance is evaluated with an anisotropic sparse quadrature approach and validated by a quasi-Monte Carlo method. We present several numerical experiments on a simplified but physiologically grounded two-dimensional geometry to illustrate the validity of the approach. The tested parametric dimension ranged from 100 up to 600. For the forward problem, the sparse quadrature is very effective. In the inverse problem, the sparse quadrature and the quasi-Monte Carlo method perform as expected, except for the total variation regularisation, where convergence is limited by lack of regularity. We finally investigate an H1/2 regularisation, which naturally stems from the boundary integral formulation, and compare it to more classical approaches.

A bi-atrial statistical shape model for large-scale in silico studies of human atria: Model development and application to ECG simulations

Large-scale electrophysiological simulations to obtain electrocardiograms (ECG) carry the potential to produce extensive datasets for training of machine learning classifiers to, e.g., discriminate between different cardiac pathologies. The adoption of simulations for these purposes is limited due to a lack of ready-to-use models covering atrial anatomical variability. We built a bi-atrial statistical shape model (SSM) of the endocardial wall based on 47 segmented human CT and MRI datasets using Gaussian process morphable models. Generalization, specificity, and compactness metrics were evaluated. The SSM was applied to simulate atrial ECGs in 100 random volumetric instances. The first eigenmode of our SSM reflects a change of the total volume of both atria, the second the asymmetry between left vs. right atrial volume, the third a change in the prominence of the atrial appendages. The SSM is capable of generalizing well to unseen geometries and 95% of the total shape variance is covered by its first 24 eigenvectors. The P waves in the 12-lead ECG of 100 random instances showed a duration of 109.7±12.2 ms in accordance with large cohort studies. The novel bi-atrial SSM itself as well as 100 exemplary instances with rule-based augmentation of atrial wall thickness, fiber orientation, inter-atrial bridges and tags for anatomical structures have been made publicly available. This novel, openly available bi-atrial SSM can in future be employed to generate large sets of realistic atrial geometries as a basis for in silico big data approaches.

Cycle length statistics during human atrial fibrillation reveal refractory properties of the underlying substrate: a combined in silico and clinical test of concept study

AIMS

The treatment of atrial fibrillation beyond pulmonary vein isolation has remained an unsolved challenge. Targeting regions identified by different substrate mapping approaches for ablation resulted in ambiguous outcomes. With the effective refractory period being a fundamental prerequisite for the maintenance of fibrillatory conduction, this study aims at estimating the effective refractory period with clinically available measurements.

METHODS AND RESULTS

A set of 240 simulations in a spherical model of the left atrium with varying model initialization, combination of cellular refractory properties, and size of a region of lowered effective refractory period was implemented to analyse the capabilities and limitations of cycle length mapping. The minimum observed cycle length and the 25% quantile were compared to the underlying effective refractory period. The density of phase singularities was used as a measure for the complexity of the excitation pattern. Finally, we employed the method in a clinical test of concept including five patients. Areas of lowered effective refractory period could be distinguished from their surroundings in simulated scenarios with successfully induced multi-wavelet re-entry. Larger areas and higher gradients in effective refractory period as well as complex activation patterns favour the method. The 25% quantile of cycle lengths in patients with persistent atrial fibrillation was found to range from 85 to 190 ms.

CONCLUSION

Cycle length mapping is capable of highlighting regions of pathologic refractory properties. In combination with complementary substrate mapping approaches, the method fosters confidence to enhance the treatment of atrial fibrillation beyond pulmonary vein isolation particularly in patients with complex activation patterns.

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.

Computer Modeling of the Heart for ECG Interpretation—A Review

Computer modeling of the electrophysiology of the heart has undergone significant progress. A healthy heart can be modeled starting from the ion channels via the spread of a depolarization wave on a realistic geometry of the human heart up to the potentials on the body surface and the ECG. Research is advancing regarding modeling diseases of the heart. This article reviews progress in calculating and analyzing the corresponding electrocardiogram (ECG) from simulated depolarization and repolarization waves. First, we describe modeling of the P-wave, the QRS complex and the T-wave of a healthy heart. Then, both the modeling and the corresponding ECGs of several important diseases and arrhythmias are delineated: ischemia and infarction, ectopic beats and extrasystoles, ventricular tachycardia, bundle branch blocks, atrial tachycardia, flutter and fibrillation, genetic diseases and channelopathies, imbalance of electrolytes and drug-induced changes. Finally, we outline the potential impact of computer modeling on ECG interpretation. Computer modeling can contribute to a better comprehension of the relation between features in the ECG and the underlying cardiac condition and disease. It can pave the way for a quantitative analysis of the ECG and can support the cardiologist in identifying events or non-invasively localizing diseased areas. Finally, it can deliver very large databases of reliably labeled ECGs as training data for machine learning.

Enabling forward uncertainty quantification and sensitivity analysis in cardiac electrophysiology by reduced order modeling and machine learning

We present a new, computationally efficient framework to perform forward uncertainty quantification (UQ) in cardiac electrophysiology. We consider the monodomain model to describe the electrical activity in the cardiac tissue, coupled with the Aliev-Panfilov model to characterize the ionic activity through the cell membrane. We address a complete forward UQ pipeline, including both: (i) a variance-based global sensitivity analysis for the selection of the most relevant input parameters, and (ii) a way to perform uncertainty propagation to investigate the impact of intra-subject variability on outputs of interest depending on the cardiac potential. Both tasks exploit stochastic sampling techniques, thus implying overwhelming computational costs because of the huge amount of queries to the high-fidelity, full-order computational model obtained by approximating the coupled monodomain/Aliev-Panfilov system through the finite element method. To mitigate this computational burden, we replace the full-order model with computationally inexpensive projection-based reduced-order models (ROMs) aimed at reducing the state-space dimensionality. Resulting approximation errors on the outputs of interest are finally taken into account through artificial neural network (ANN)-based models, enhancing the accuracy of the whole UQ pipeline. Numerical results show that the proposed physics-based ROMs outperform regression-based emulators relying on ANNs built with the same amount of training data, in terms of both numerical accuracy and overall computational efficiency.

Linking statistical shape models and simulated function in the healthy adult human heart

Cardiac anatomy plays a crucial role in determining cardiac function. However, there is a poor understanding of how specific and localised anatomical changes affect different cardiac functional outputs. In this work, we test the hypothesis that in a statistical shape model (SSM), the modes that are most relevant for describing anatomy are also most important for determining the output of cardiac electromechanics simulations. We made patient-specific four-chamber heart meshes (n = 20) from cardiac CT images in asymptomatic subjects and created a SSM from 19 cases. Nine modes captured 90% of the anatomical variation in the SSM. Functional simulation outputs correlated best with modes 2, 3 and 9 on average (R = 0.49 ± 0.17, 0.37 ± 0.23 and 0.34 ± 0.17 respectively). We performed a global sensitivity analysis to identify the different modes responsible for different simulated electrical and mechanical measures of cardiac function. Modes 2 and 9 were the most important for determining simulated left ventricular mechanics and pressure-derived phenotypes. Mode 2 explained 28.56 ± 16.48% and 25.5 ± 20.85, and mode 9 explained 12.1 ± 8.74% and 13.54 ± 16.91% of the variances of mechanics and pressure-derived phenotypes, respectively. Electrophysiological biomarkers were explained by the interaction of 3 ± 1 modes. In the healthy adult human heart, shape modes that explain large portions of anatomical variance do not explain equivalent levels of electromechanical functional variation. As a result, in cardiac models, representing patient anatomy using a limited number of modes of anatomical variation can cause a loss in accuracy of simulated electromechanical function.

Using machine learning to identify local cellular properties that support re-entrant activation in patient-specific models of atrial fibrillation

AIMS

Atrial fibrillation (AF) is sustained by re-entrant activation patterns. Ablation strategies have been proposed that target regions of tissue that may support re-entrant activation patterns. We aimed to characterize the tissue properties associated with regions that tether re-entrant activation patterns in a validated virtual patient cohort.

METHODS AND RESULTS

Atrial fibrillation patient-specific models (seven paroxysmal and three persistent) were generated and validated against local activation time (LAT) measurements during an S1-S2 pacing protocol from the coronary sinus and high right atrium, respectively. Atrial models were stimulated with burst pacing from three locations in the proximity of each pulmonary vein to initiate re-entrant activation patterns. Five atria exhibited sustained activation patterns for at least 80 s. Models with short maximum action potential durations (APDs) were associated with sustained activation. Phase singularities were mapped across the atria sustained activation patterns. Regions with a low maximum conduction velocity (CV) were associated with tethering of phase singularities. A support vector machine (SVM) was trained on maximum local conduction velocity and action potential duration to identify regions that tether phase singularities. The SVM identified regions of tissue that could support tethering with 91% accuracy. This accuracy increased to 95% when the SVM was also trained on surface area.

CONCLUSION

In a virtual patient cohort, local tissue properties, that can be measured (CV) or estimated (APD; using effective refractory period as a surrogate) clinically, identified regions of tissue that tether phase singularities. Combing CV and APD with atrial surface area further improved the accuracy in identifying regions that tether phase singularities.

Translational applications of computational modelling for patients with cardiac arrhythmias

Cardiac arrhythmia is associated with high morbidity, and its underlying mechanisms are poorly understood. Computational modelling and simulation approaches have the potential to improve standard-of-care therapy for these disorders, offering deeper understanding of complex disease processes and sophisticated translational tools for planning clinical procedures. This review provides a clinician-friendly summary of recent advancements in computational cardiology. Organ-scale models automatically generated from clinical-grade imaging data are used to custom tailor our understanding of arrhythmia drivers, estimate future arrhythmogenic risk and personalise treatment plans. Recent mechanistic insights derived from atrial and ventricular arrhythmia simulations are highlighted, and the potential avenues to patient care (eg, by revealing new antiarrhythmic drug targets) are covered. Computational approaches geared towards improving outcomes in resynchronisation therapy have used simulations to elucidate optimal patient selection and lead location. Technology to personalise catheter ablation procedures are also covered, specifically preliminary outcomes form early-stage or pilot clinical studies. To conclude, future developments in computational cardiology are discussed, including improving the representation of patient-specific fibre orientations and fibrotic remodelling characterisation and how these might improve understanding of arrhythmia mechanisms and provide transformative tools for patient-specific therapy.

An audit of uncertainty in multi-scale cardiac electrophysiology models

Models of electrical activation and recovery in cardiac cells and tissue have become valuable research tools, and are beginning to be used in safety-critical applications including guidance for clinical procedures and for drug safety assessment. As a consequence, there is an urgent need for a more detailed and quantitative understanding of the ways that uncertainty and variability influence model predictions. In this paper, we review the sources of uncertainty in these models at different spatial scales, discuss how uncertainties are communicated across scales, and begin to assess their relative importance. We conclude by highlighting important challenges that continue to face the cardiac modelling community, identifying open questions, and making recommendations for future studies. This article is part of the theme issue 'Uncertainty quantification in cardiac and cardiovascular modelling and simulation'.

Creation and application of virtual patient cohorts of heart models

Patient-specific cardiac models are now being used to guide therapies. The increased use of patient-specific cardiac simulations in clinical care will give rise to the development of virtual cohorts of cardiac models. These cohorts will allow cardiac simulations to capture and quantify inter-patient variability. However, the development of virtual cohorts of cardiac models will require the transformation of cardiac modelling from small numbers of bespoke models to robust and rapid workflows that can create large numbers of models. In this review, we describe the state of the art in virtual cohorts of cardiac models, the process of creating virtual cohorts of cardiac models, and how to generate the individual cohort member models, followed by a discussion of the potential and future applications of virtual cohorts of cardiac models. This article is part of the theme issue 'Uncertainty quantification in cardiac and cardiovascular modelling and simulation'.

Constructing a Human Atrial Fibre Atlas

Atrial anisotropy affects electrical propagation patterns, anchor locations of atrial reentrant drivers, and atrial mechanics. However, patient-specific atrial fibre fields and anisotropy measurements are not currently available, and consequently assigning fibre fields to atrial models is challenging. We aimed to construct an atrial fibre atlas from a high-resolution DTMRI dataset that optimally reproduces electrophysiology simulation predictions corresponding to patient-specific fibre fields, and to develop a methodology for automatically assigning fibres to patient-specific anatomies. We extended an atrial coordinate system to map the pulmonary veins, vena cava and appendages to standardised positions in the coordinate system corresponding to the average location across the anatomies. We then expressed each fibre field in this atrial coordinate system and calculated an average fibre field. To assess the effects of fibre field on patient-specific modelling predictions, we calculated paced activation time maps and electrical driver locations during AF. In total, 756 activation time maps were calculated (7 anatomies with 9 fibre maps and 2 pacing locations, for the endocardial, epicardial and bilayer surface models of the LA and RA). Patient-specific fibre fields had a relatively small effect on average paced activation maps (range of mean local activation time difference for LA fields: 2.67-3.60 ms, and for RA fields: 2.29-3.44 ms), but had a larger effect on maximum LAT differences (range for LA 12.7-16.6%; range for RA 11.9-15.0%). A total of 126 phase singularity density maps were calculated (7 anatomies with 9 fibre maps for the LA and RA bilayer models). The fibre field corresponding to anatomy 1 had the highest median PS density map correlation coefficient for LA bilayer simulations (0.44 compared to the other correlations, ranging from 0.14 to 0.39), while the average fibre field had the highest correlation for the RA bilayer simulations (0.61 compared to the other correlations, ranging from 0.37 to 0.56). For sinus rhythm simulations, average activation time is robust to fibre field direction; however, maximum differences can still be significant. Patient specific fibres are more important for arrhythmia simulations, particularly in the left atrium. We propose using the fibre field corresponding to DTMRI dataset 1 for LA simulations, and the average fibre field for RA simulations as these optimally predicted arrhythmia properties.