| dc.contributor.advisor | Christoph, Jan Dr. | |
| dc.contributor.author | Lebert, Jan | |
| dc.date.accessioned | 2024-05-24T15:34:01Z | |
| dc.date.available | 2024-05-31T00:50:08Z | |
| dc.date.issued | 2024-05-24 | |
| dc.identifier.uri | http://resolver.sub.uni-goettingen.de/purl?ediss-11858/15272 | |
| dc.identifier.uri | http://dx.doi.org/10.53846/goediss-10505 | |
| dc.format.extent | 252 | de |
| dc.language.iso | eng | de |
| dc.rights | Attribution-NonCommercial-ShareAlike 4.0 International | * |
| dc.rights.uri | http://creativecommons.org/licenses/by-nc-sa/4.0/ | * |
| dc.subject.ddc | 571.4 | de |
| dc.title | Reconstruction of Cardiac Wave Dynamics from Mechanical Deformation | de |
| dc.type | doctoralThesis | de |
| dc.contributor.referee | Christoph, Jan Dr. | |
| dc.date.examination | 2023-06-15 | de |
| dc.description.abstracteng | Cardiac arrhythmias, such as atrial and ventricular fibrillation, are caused by abnormal electrical activity within the heart muscle, leading to irregular contractions and potentially life-threatening situations. Current diagnostic methods, such as electrocardiograms and minimally invasive catheter mapping, provide limited information about these abnormal electrical dynamics. This thesis explores novel approaches to recover hidden electrical dynamics from indirect and partial measurements, with the ultimate goal of developing more accurate and efficient diagnostic tools for cardiac arrhythmias. The central focus is on a novel approach for non-invasive mapping of cardiac electrical activity by analyzing the heart's mechanical deformation, which is triggered by the underlying electrical excitation. This work investigates the feasibility of this approach, referred to as the inverse mechano-electrical problem, through a combination of computational and experimental studies. First, the viability of deep learning for predicting cardiac dynamics from indirect and partial observations was investigated. Convolutional neural networks (CNNs) were successfully used to predict phase maps and detect phase singularities from optical mapping videos and simulated low-resolution data mimicking catheter mapping data. Next, the inverse mechano-electrical problem was tackled in computer simulations. A synchronization-based data assimilation approach was used to recover complex three-dimensional scroll wave patterns from noisy and partial observations of deformation in a bulk tissue with fiber anisotropy. Subsequently, a deep learning approach using CNNs was developed, achieving superior accuracy in reconstructing cardiac wave dynamics from mechanical deformation. This approach was further validated in highly diverse, smoothed particle hydrodynamics (SPH)-based simulations of bi-ventricular heart geometries, demonstrating the potential for non-invasive mapping of cardiac electrophysiology in a realistic setting. Finally, an experimental platform was established for imaging electro-mechanical dynamics in the heart. Simultaneous multi-camera panoramic optical mapping and volumetric ultrasound were used to capture the surface electrical activity and mechanical deformation of beating isolated hearts. For this purpose, classical and deep learning-based numerical motion tracking techniques specialized on optical mapping data were developed and evaluated. While the deep learning approach successfully predicted electrical wave dynamics from mechanical deformation in simulations, its application to experimental data remains challenging due to limitations in ultrasound imaging speed and motion tracking accuracy. This thesis contributes to the advancement of non-invasive mapping of cardiac electrophysiology by providing a first numerical proof-of-principle and laying the groundwork for the preclinical development of the inverse mechano-electrical approach. Future work will focus on overcoming the limitations encountered in experimental data and translating the approach into clinical settings. | de |
| dc.contributor.coReferee | Enderlein, Jörg Prof. Dr. | |
| dc.subject.eng | arrhythmia | de |
| dc.subject.eng | deep learning | de |
| dc.subject.eng | artificial neural network | de |
| dc.subject.eng | nonlinear systems | de |
| dc.subject.eng | inverse problems | de |
| dc.subject.eng | optical mapping | de |
| dc.subject.eng | motion tracking | de |
| dc.identifier.urn | urn:nbn:de:gbv:7-ediss-15272-2 | |
| dc.affiliation.institute | Göttinger Graduiertenschule für Neurowissenschaften, Biophysik und molekulare Biowissenschaften (GGNB) | de |
| dc.subject.gokfull | Biologie (PPN619462639) | de |
| dc.description.embargoed | 2024-05-31 | de |
| dc.identifier.ppn | 1889836710 | |
| dc.identifier.orcid | 0000-0001-8754-4964 | de |
| dc.notes.confirmationsent | Confirmation sent 2024-05-24T19:45:01 | de |