Note: contributions are sorted in alphabetical order, please refer to the program for sessions and order of appearance

Physics based and data- driven modeling analysis and control of cardiac dynamics

Proposed by: Ulrich Parlitz (Max Planck Institute for Dynamics and Self-Organization, Göttingen, Germany), Stefan Luther (Max Planck Institute for Dynamics and Self-Organization, Göttingen, Germany), Maxime Sermesant (Computational Cardiology, Inria, Université Côte d'Azur, France)

Patrick Boyle

Patient-Specific Simulation of Potentially Pre-Arrhythmogenic Substrate in Embolic Stroke of Undetermined Source

P.M. Boyle

pmjboyle@uw.edu

University of Washington

Cardiac magnetic resonance imaging (MRI) has revealed fibrotic remodeling in embolic stroke of undetermined source (ESUS) patients comparable to levels seen in atrial fibrillation (AFib). Our group has recently used computational modeling to understand the absence of arrhythmia in ESUS despite the presence of putatively pro-arrhythmic fibrosis. We have reconstructed MRI-based atrial models 45 ESUS patients, alongside a control set of 45 models reconstructed from MRI scans of AFib patients. The fibrotic substrate’s arrhythmogenic capacity in each of these 90 patient-specific models was assessed computationally using a virtual overdrive pacing protocol. Reentrant drivers (i.e., self-sustaining “rotor” patterns of bioelectric excitation) were induced in 24/45 (53%) ESUS and 22/45 (49%) AFib models. Models in which rotors were induced had more fibrosis (16.7±5.45%) than non-inducible models (11.07±3.61%; P<0.0001); however, inducible subsets of ESUS and AFib models had similar fibrosis levels (P=0.90), meaning the intrinsic pro-arrhythmic substrate properties of fibrosis in ESUS and AFib are essentially indistinguishable. This suggests some ESUS patients have latent pre-clinical fibrotic substrate that could be a future source of arrhythmogenicity. Thus, our work prompts the novel hypothesis that ESUS patients with fibrotic atria are spared from AFib due predominantly to an absence of arrhythmia triggers.

Jan Christoph

A comparison of a physics-based and data-driven inverse reconstruction technique of cardiac excitation wave patterns from mechanical deformation

Jan Christoph, Jan Lebert

jan.christoph@ucsf.edu

University of California, San Francisco

The contractions of the cardiac muscle are caused by nonlinear waves of electrical excitation. Because it is difficult to measure the electrical excitation within the volume of the cardiac muscle, it is compelling to ask whether the tissue deformation can be analyzed using an inverse numerical approach, such that the electrical excitation can be fully reconstructed. In this talk, I will discuss and compare two inverse numerical approaches, one physics-based and one deep learning-based approach, both of which aim to achieve the same goal of computing electrical excitation wave patterns from mechanical deformation. The two approaches were tested using synthetic data, which was generated in computer simulations of bulk-shaped elastic excitable media with anisotropic muscle fibre architecture. While the physics-based approach employed a replicate numerical model that assimilates mechanical observation data and had to be tuned carefully to match the original dynamics, the deep learning approach employed a convolutional neural network with an autoencoder-like architecture that was trained on many thousands of pairs of mechanical and corresponding electrical data. While both approaches can be used to successfully compute electrical excitation wave patterns from tissue deformation, our results show that deep learning outperforms the physics-based approach. Using deep learning, it becomes possible to predict even scroll wave chaos and their vortex filaments with high accuracy, even in the presence of noisy mechanical data and at low spatial resolutions.

Sebastian Herzog

Data-driven modeling of cardiac dynamics by means of neural network hybrids

S.Herzog, F. Wörgötter, R. S. Zimmermann, J. Abele, S. Luther and U. Parlitz

sherzog3@gwdg.de

University of Göttingen, Third Institute of Physics - Biophysics

The mechanical contraction of the pumping heart is driven by electrical excitation waves running across the heart muscle due to the excitable electrophysiology of heart cells. With cardiac arrhythmias these waves turn into stable or chaotic spiral waves whose observation in the heart is very challenging. Data-driven methods based on neural network hybrids consisting of convolutional neural networks for reconstructing and forecasting these spiral waves are presented. The performance our approach is demonstrated using the four variables of the Bueno-Orovio-Fenton-Cherry model describing electrical excitation waves in cardiac tissue.

Victoriya Kashtanova

EP-Net 2.0: Out-of-Domain Generalisation for Deep Learning Models of Cardiac Electrophysiology

Victoriya Kashtanova, Ibrahim Ayed, Nicolas Cedilnik, Patrick Gallinari and Maxime Sermesant

victoriya.kashtanova@inria.fr

Inria, Universite Cote d’Azur, Nice

Cardiac electrophysiology models achieved good progress in simulating cardiac electrical activity. However, it is still challenging to leverage clinical measurements due to the discrepancy between idealised models and patient-specific conditions. In the last few years, data-driven machine learning methods have been actively used to learn dynamics and physical model parameters from data. In this paper, we propose a principled deep learning approach to learn the cardiac electrophysiology dynamics from data in the presence of scars in the cardiac tissue slab. We demonstrate that this technique is indeed able to reproduce the transmembrane potential dynamics in situations close to the training context. We then focus on evaluating the ability of the trained networks to generalize outside their training domain. We show experimentally that our model is able to generalize to new conditions including more complex scar geometries, multiple signal onsets and various conduction velocities.

Thomas Lilienkamp

Utilizing transient dynamics: Controlling spiral wave chaos by small perturbations

Thomas Lilienkamp, Ulrich Parlitz

thomas.lilienkamp@ds.mpg.de

Max-Planck-Institute for Dynamics and Self-Organization

The dynamics during life threatening cardiac arrhythmias like ventricular fibrillation is governed by chaotic spiral/scroll wave dynamics. In ex-vivo experiments and numerical simulations, a phenomenon called self-termination can be observed frequently, where the chaotic dynamics terminates by itself without any interaction. We demonstrate what implications this observation has on the structure of the state space, and how this structure can be exploited for an efficient control of the dynamics via small but finite perturbations (localized in space and time). Furthermore, we discuss in general how an optimal configuration of perturbations can be achieved, in order to control the dynamics and terminate the chaotic spiral wave dynamics.

Rupamanjari Majumder

The chaotic route to spiral wave control: an optogenetics approach.

Rupamanjari Majumder, Vladimir Zykov, Eberhard Bodenschatz

rupamanjari.majumder@ds.mpg.de

Max Planck Institute for Dynamics and Self Organisation, Göttingen

Fatal cardiac arrhythmias such as tachycardia and fibrillation are associated with the occurrence of spiral waves, the control of which is essential for the treatment of the disease. To date, the most effective means of controlling spiral waves in the heart have been high-voltage shock-based control methods. These rely on ensuring abrupt electrical synchronisation of the heart tissue. However, due to the many negative side effects associated with these methods, low energy techniques are in great demand. Low-energy techniques bring about termination of spiral waves by forcing them to drift towards non-excitable tissue boundaries with which they collide and extinguish their phase singularities. In particular, such drift can be induced by spatiotemporal modulation of domain excitability in optogenetically modified cardiac tissue. In this talk, I will demonstrate a low-energy optogenetics-based approach to suppress spiral waves by their controlled chaotisation and rapid drift.

Alexander Panfilov

In silico–in vitro approach to study and control cardiac arrhythmias

A.V.Panfilov

Alexander.Panfilov@UGent.be

Ghent University, Gent, Belgium; Leiden University Medical Centrum, Leiden, Netherlands

Sudden cardiac death as a result of cardiac arrhythmias is the leading cause of death in the industrialized countries. Although cardiac arrhythmias has been studied well over a century, their underlying mechanisms remain largely unknown. One of the main problems is that cardiac arrhythmias occur at the level of the whole organ only, while in most of the cases only single cell experiments can be performed. Due to these limitations alternative approaches, such as multiscale biophysical modelling of the heart, are currently of great interest. Such methodology is extremely valuable if it combined with experimental and clinical methodology. In my talk I will present results of research which combine usage of modelling and modern experimental techniques. In particular, I will report on studies in which properties of cardiac tissue were manipulated using optogenetics and show how this technology can be used to study basic properties of cardiac propagation and to control cardiac arrhythmias using Attract-Anchor-Drag method [1]. I will also report on new data driven methods to identify sources of cardiac arrhythmias from clinical mapping data using a novel methodology of DG-mapping developed in our group [2]. [1]. Majumder, R., Feola, I., Teplenin, A. S., de Vries, A. A., Panfilov, A., Pijnappels, D. A. ”Optogenetics enables real-time spatiotemporal control over spiral wave dynamics in an excitable cardiac system.”, ELIFE, 7:e41076,(2018); [2]. Vandersickel N, Van Nieuwenhuyse E, Van Cleemput N, Goedgebeur J, El Haddad M, De Neve J, Demolder A, Strisciuglio T, Duytschaever M, Panfilov A.V. "Directed networks as a novel way to describe and analyze cardiac excitation: Directed Graph mapping", Front. Physiol., v.10,1138,(2019).

Arkady Pertsov

Evolution of Scroll Ring in Myocardial Wall

Arkady M. Pertsov, V.N. Biktashev, H. Dierckx

pertsova@upstate.edu

SUNY Upstate Medical University, Syracuse NY 13210, USA

Scroll waves are implicated in the most dangerous cardiac arrhythmia known as ventricular fibrillation, the main cause of sudden cardiac death. Scroll waves are organized around phase singularity lines, filaments. The filaments are local breaks of the excitation fronts. The fronts are thin compared to ventricular wall and can be viewed as a 2D surface. Consequently, local breaks of the front can be viewed as holes in this surface, which give rise to filaments with close loop topology (rings). One of the key parameters controlling filament dynamics is filament tension which can be positive or negative depending on tissue excitability. In excitable media with spatially uniform parameters, the positive tension makes rings collapse, healing the front breaks, and terminating the scroll wave activity. It would be reasonable to assume in the normal heart the tension should be positive, which would make the scroll wave formation more difficult and thus protect the heart from arrhythmias. However, the heart wall has a significant inherent non-uniformity, which could affect the filament dynamics and the course of arrhythmia. Such non-uniformity is so-called twisted anisotropy, resulting from large differences in fiber orientation across the thickness of the myocardial wall. Mathematically it can be described by spatial gradients of the diffusivity tensor in the reaction-diffusion equations governing propagation of the action potential in the heart. Here we explore computationally and analytically the dynamics of ring-shaped filaments in the presence of twisted anisotropy. We demonstrate that the gradients of diffusion tensor components cause major deformation of the ring and change its dynamics. In case of positive tension, we demonstrate the for a broad range of initial conditions the filament does not collapse but forms quasi-stable intramural L-shaped filaments. The results are discussed in the context of filament dynamics in transillumination tissue experiments.