Electromechanical Coupling

Cardiac tissue is an excitable and deformable medium in which tissue mechanics, cellular electrophysiology, and energetics are mutually coupled. Pathophysiological perturbations in coupling may result in electrical and mechanical dysynchrony, structural and functional remodelling of the cardiac substrate, which induce or exacerbate life-threatening arrhythmias and heart failure. Despite decades of research, the fundamental principles underlying electrical, mechanical, and energetic dynamics and self-organization that contribute to the onset and progression of arrhythmias and heart failure remain largely elusive. This lack of progress is mainly because measuring cardiac mechanics, electrophysiology, and energetics simultaneously, as well as resolving wave dynamics inside the heart muscle, remain significant scientific and technological challenges. The Max Planck Research Group Biomedical Physics has pioneered the development of multi-modal imaging technologies to visualize three-dimensional electromechanical waves inside the cardiac muscle and is translating them from bench to bedside. In addition to the use of particularly powerful state-of-the-art measurement technology, new sophisticated methods of data analysis are being developed and applied, including novel approaches from the field of machine learning.

Electromechanical Imaging of Human VF

The three-dimensional spatio-temporal organization of human ventricular fibrillation is of immense scientific and medical interest. Over the past century, several hypotheses have been formulated about the nature of cardiac fibrillation [1]. However, available imaging techniques so far were only able to resolve the dynamics on the surface but not inside the heart muscle.
We have developed advanced 4D ultrasound-based imaging technology that resolves intramural mechanical activation with unprecedented spatial-temporal resolution [2]. In an ongoing clinical study in collaboration with H. Baraki and I. Kutschka (University Medical Center, Göttingen), we demonstrate the first visualization of electromechanic waves during human ventricular fibrillation inside the heart muscle. We are able to resolve the three dimensional dynamics and interaction of intramural rotors and phase singular filaments, and observe intermittency in spatial-temporal complexity. In collaboration with Max Planck Innovation GmbH and our clinical partners at the Yale University and UCSD, we are further advancing 4D electromechanic imaging for arrhythmia and heart failure research, and exploring further applications for advanced diagnostics and therapeutic interventions, including ventricular ablation and cardiac resynchronization therapy.

Multi-modal cardiac imaging

We have developed multi-modal optical mapping, which allows the simultaneous measurement of membrane potential, intracellular calcium, and mechanical motion in intact, isolated hearts using fluorescent dyes [2,3]. This technique has enabled studying excitation-contraction coupling on tissue and organ level and led to the discovery of coexisting electro-mechanical vortices. We plan to further advance multi-modal optical mapping to characterize metabolic function in collaboration with our partners at Yale University. Optical mapping has been used to validate ultrasound-based 4D electromechanical imaging. The figure shows strain wave imaging during ventricular tachycardia in an isolated, Langendorff-perfused porcine heart. The analysis of the strain enables the estimate of an mechanical activation time inside the tissue that represents the propagation of the three-dimensional mechanical wave. The measurement of transient wave phenomena in the ventricular wall transcends the capabilities of state-of-the-art cardiac mapping systems. In order to ensure reproducible and sustainable biomedical research, BMPG has developed the open source workflow data management system LinkAhead, advances data management concepts [5] and supports the MPG’s OpenScience initiative.

Reconstrucring in-depth activity of chaotic 3D spatio-temporal excitable media based on surface data

Motivated by potential applications in cardiac research, we set ourselves the task of reconstructing the dynamics in a spatiotemporally chaotic, excitable three-dimensional medium from partial observations at the surface [6]. Three artificial neural network methods (a spatiotemporal convolutional long-short-term-memory, an autoencoder, and a diffusion model based on the U-Net architecture) were trained to predict the dynamics in deeper layers of a cube-like three-dimensional excitable medium from observational data at the surface using data generated by the Barkley model

$ \frac{\partial u}{\partial t} = D\cdot \nabla^2u+\frac{1}{\varepsilon}u(1-u)\left(u-\frac{v+b}{a}\right) $

$ \frac{\partial v}{\partial t} = u^3-v $

on a 3D domain with no-flux boundary conditions. The figure shows that despite the high-dimensional chaotic dynamics of this system, such cross-prediction is possible, but non-trivial and as expected, its quality decreases with increasing prediction depth.

References

1. W.-J. Rappel, Phys. Rep. 978, 1 (2022)
2. J. Christoph et al., Nature 555, 667 (2018)
3. V. Kappadan, A. Sohi, U. Parlitz et al., J. Physiol. 601, 1353 (2023)
4. S. Luther, J. Am. Coll. Cardiol. EP. 11, 682 (2025)
5. H. tom Wörden, F. Spreckelsen, S. Luther, U. Parlitz, A. Schlemmer, Data 9, 24 (2024)
6. R. Stenger, S. Herzog, I. Kottlarz, B. Rüchardt, S. Luther, F. Wörgötter, U. Parlitz, Chaos 33, 013134 (2023)