Thesis

Contents

  1. Our research group
  2. 2026-2027 thesis subjects
  3. Research background

You can find an introduction to cardiac arrhythmia below. More detail about our ongoing research can be found on our research page.

Our research group

We are a young and dynamic research group at the department of Physics and Astronomy led by Prof. Vandersickel (nele.vandersickel@ugent.be). We perform research at the edge of science including physics, computer programming and medicine, a truly interdisciplinary setting. Our research focuses on the analysis of cardiac arrhythmia, which is the main cause of death in the Western society. We provide weekly guidance, step by step, while we will also invite you to come up with own ideas and allow room for creativity. We will give you insight in the scientific process of solving problems, while also focusing on personal growth. We aim for the thesis to allow you to do interesting research while also enjoying the process. Ask a former master student how they experienced their master thesis with us.

What are we looking for? We do not care about your previous scores, but we ask a motivated mindset and a love for programming and solving problems.

What will you do? We do not make subjects just to keep you busy, but you will actually contribute to our scientific research. We have selected subjects which are possible to investigate during one year, while also being able to actually make a contribution. If you succeed, your research should be able (maybe joined with other research) to be published in a scientific journal. How cool would that be?

2026-2027 thesis subjects

We present some subjects which seem relevant to us. If you have a specific interest, please also talk to us, and we might be able think of other subjects which match your interests. We can also bring you to the clinic so you can observe ablation procedures and explain and discuss your results with the medical doctors.

Epicardial bridges - finding hidden electical pathways in clinical cardiac activation maps

Background · Cardiac arrhythmias remain one of the leading causes of death in Western society. Despite their clinical importance, the underlying mechanisms and their detection are still not fully understood. One such arrhythmia is atrial tachycardia (AT), a condition in which the heart beats faster than normal due to abnormal electrical activity in the atria.

A common mechanism underlying AT is re-entry, where electrical activation propagates in a self-sustaining loop, repeatedly reactivating the surrounding tissue. These re-entry circuits can be localized by mapping the heart using a catheter. During such procedures, electrical signals are recorded throughout the atrium and converted into local activation times (LATs), which indicate when specific regions of the tissue are activated.

Our group has recently developed novel methods to automatically detect these re-entry circuits from electroanatomical maps, significantly simplifying their identification. However, in some cases, the electrical activity propagates through pathways outside the atrial wall. These so-called epicardial pathways include structures such as the Vein of Marshall, the Coronary sinus, or interatrial connections like the Bachmann bundle. These structures can act as additional conduction pathways and sustain re-entry circuits. A major challenge is that these epicardial pathways are difficult or sometimes impossible to map directly. As a result, their contribution to the arrhythmia must be inferred indirectly from endocardial measurements, leaving gaps in the observed activation patterns.

In this thesis, we aim to develop methods to detect whether a patient’s AT is sustained by epicardial pathways and to localize their entry and exit sites using endocardial electroanatomical maps.

Goal · In the first stage, a method will be developed to determine whether a clinical activation map likely contains an epicardial bridge. Two complementary approaches will be pursued. First, physically motivated features will be extracted from the map, such as gaps in the LAT sequence or spatially disconnected activation regions. These features can be combined into a probabilistic classifier. Second, unsupervised machine learning will be used to compare maps with and without epicardial bridges, identifying which features discriminate between the two cases. This can be done through clustering, principal component analysis (PCA) and uniform manifold approximation and projection (UMAP).

In the second stage, the entry and exit sites of the epicardial bridge will be located. The exit sit, where activation re-enters the endocardium, manifests as a focal source of electrical activity. It can be detected by identifying regions of positive divergence in the activation velocity field or by using persistent homology and looking for high-persistence H₀ generators for the map. The entry site, where activation leaves the endocardium to travel through the epicardial pathway, is considerably harder to detect, as it often leaves no clear signature in the activation map. To find likely origin site, one can use activation gaps and find which entry sites correspond best with the detected LAT gap and the detected exit site.

Why this matters · This addresses a clinically relevant and growing problem. Epicardial pathways are increasingly implicated in atrial tachycardia, yet reliable tools for their detection and localization remain lacking. The methods developed here have direct potential to improve ablation planning for this challenging patient group. The developed methods have the potential to directly improve ablation planning for this challenging patient group. The project involves working with clinical data in collaboration with electrophysiologists, with the opportunity to go to the clinic to discuss your research and see how ablation procedures are performed.

Clinical Epicardial Brigge A clinical case of left atrial tachycardia sustained by re-entry through an epicardial pathway. The entry and exit sites are shown.
LAT Histogram of Epicardial Bridge An LAT histogram showing missing LATs which hints at conduction through a hidden pathway.
Exploring phase transitions in cardiac arrhythmia using the classical XY lattice model

Background · The electrical activity of the heart manifests in several distinct configurations: sinus rhythm, tachycardia, and fibrillation. During sinus rhythm, the heart beats in a regular, coordinated fashion. When normal electrical conduction is disrupted, a cardiac arrhythmia arises, taking the form of either tachycardia or fibrillation. Tachycardia is characterized by a regular but abnormally fast heart rhythm, whereas fibrillation exhibits both irregular and accelerated electrical activity. Understanding the transition between these states is still poorly understood in cardiac electrophysiology.

A recent paper highlighted the potential of modeling this transition using the Kosterlitz-Thouless phase transition in the XY-model [1]. The XY-model is a generalization of the classical 2D Ising model, which allows all spin orientations in [0, 2π[. The KT phase transition is a phase transition based on topology, where vortex pairs are created when the temperature exceeds a critical threshold [2]. This KT transition was compared to the transition between sinus rhythm or tachycardia and fibrillation. Many similarities were found, such as power-law behavior of spatial correlation functions and vortex lifetimes. The major difficulty in applying the XY-model to a practical cardiology context lies in the fact that the XY-model describes a system in thermal equilibrium, whereas cardiac tissue is a non-equilibrium system.

Goal · This thesis aims to further investigate the analogy between the KT transition in the XY-model and the transition between cardiac states, based on realistic arrhythmia simulations. This will be done by analyzing snapshots of simulated sinus rhythm, tachycardia, and fibrillation on a simple 2D square tissue in Finitewave. Finitewave is an open-source Python package designed for simulations of cardiac arrhythmia developed by our group. In the first part of the project, the student will collect snapshots of these simulations and interpret them as equilibrium configurations of the XY-model. A methodology will then be developed to extract an effective temperature that is consistent with the given XY-model configuration (for an example, see [3]). The aim is to show that sinus rhythm or tachycardic states correspond to lower effective temperatures than fibrillation, where the effective temperature serves as a quantitative measure of order versus disorder. In the second part of the project, multiple consecutive time frames can be analyzed to determine how the temperature evolves over time and to investigate whether changes in this parameter can capture or anticipate the transition between atrial tachycardia and atrial fibrillation. In the XY-model, ordered and disordered regimes are separated by a critical temperature associated with the Kosterlitz–Thouless transition, suggesting the possibility of having a similar critical threshold in cardiac arrhythmia.

Why this matters · This interdisciplinary project brings together expertise from the biophysics group and the statistical physics group to bridge cardiac electrophysiology and statistical physics. We believe this project is highly relevant due to its fundamental nature. Research in cardiology is often top-down and inspired by digital twins, which are hyperrealistic models of human hearts. In contrast, the approach of our group is bottom-up; we start from simple simulations to investigate the drivers of cardiac arrhythmia. Our group has already been successful by applying the conservation of topological charge to ablation therapy in clinical atrial tachycardia [4]. We are currently applying these ideas to ventricular tachycardia and atrial fibrillation. This thesis perfectly fits the spirit of our group, as we aim to investigate the fundamental drivers of arrhythmic state transitions in the heart. Our end goal with this project would be to provide electrophysiologists with the ability to assess the fibrillation inducibility during surgery based on a critical threshold.

An example of topological vortices in an ordered (A) and disordered (B) state.

[1] Anand Narayan Ganesan, Pawel Kuklik, and Stanley Nattel. “A topological hypothesis for atrial fibrilllation, atrial flutter and focal atrial tachycardia: comparison and contrast with Kosterlitz-Thouless physics”. In: Frontiers in Network Physiology 5 (Jan. 2026). issn: 2674-0109. doi: 10.3389/fnetp.2025.1710567.

[2] J M Kosterlitz and D J Thouless. “Ordering, metastability and phase transitions in two-dimensional systems”. In: Journal of Physics C: Solid State Physics 6.7 (Apr. 1973), pp. 1181–1203. issn: 0022-3719. doi: 10.1088/0022-3719/6/7/010.

[3] Andres M. Belaza et al. “Social stability and extended social balance—Quantifying the role of inactive links in social networks”. In: Physica A: Statistical Mechanics and its Applications 518 (Mar. 2019), pp. 270–284. issn: 0378-4371. doi: 10.1016/j.physa.2018.11.055.

[4] Mattias Duytschaever et al. “Atrial Topology for a Unified Understanding of Typical and Atypical Flutter”. In: Circulation: Arrhythmia and Electrophysiology 17.11 (Nov. 2024). issn: 1941-3084. doi: 10.1161/circep.124.013102.

Testing Extended Phase Mapping on exotic activation patterns

Summary · CCardiac arrhythmias are described as abnormal electrical activity in the heart that disturbs the natural rhythm. This disturbance is caused by mechanisms that maintain residual electrical activity in the tissue. The most important mechanism is called reentry, where electrical waves rotate around an obstacle or around their own centre within the cardiac tissue. Current methods have already been developed to identify these locations. However, false-positive detections are a still a common occurrence with these conventional methods. Now, we would like to test whether or not our novel method (Extended Phase Mapping) can do better and detect less false-positives.

Goal · Using our software (openDGM), you will apply Extended Phase Mapping and Phase mapping to edge cases such as those described in Podziemski et al. [1], in order to determine whether false-positive detections are still present when using this new method compared to conventional Phase Mapping [1].

The scope of this thesis includes:

  • Understanding of the methods used for simulations and rotation detection
  • Creating simulations that describe the edge cases
  • Setting up pipelines to analyse the results
  • Identifying potential solutions to false-positive detections

This thesis is very programming-heavy. Therefore, we require the student to enjoy programming (Python).

(A) Experimental setup of the comparison of a rotor to 2 planar waves. In (B) one observes Phasemapping, detecting a rotational center through a phase singularity. In (C) Naive implementations of Phase mapping find a rotational center, while no rotations is present, as the left and right side of the tissue is separated by a line of block.

[1] P. Podziemski, S. Zeemering, P. Kuklik, A. van Hunnik, B. Maesen, J. Maessen, H. J. Crijns, S. Verheule, and U. Schotten, “Rotors detected by phase analysis of filtered, epicardial atrial fibrillation electrograms colocalize with regions of conduction block,” Circulation: Arrhythmia and Electrophysiology, vol. 11, no. 10, p. e005858, 2018. doi: 10.1161/CIRCEP.117.005858

Research background

Normal excitation of the heart and cardiac arrhythmia

Cardiac arrhythmias are the leading cause of death in the Western world. The management ofcardiac arrhythmias and related diseases currently accounts for about 9% of the total health-care expenditure across the European Union. In order to understand the mechanism of cardiacarrhythmia, we will first explain how the normal heart works, see Fig 1.

The heart consists of 4 parts: 2 upper chambers, called the atria, and two lower chambers, the ventricles. The contraction of the heart is initiated by an electrical wave which automatically starts at the sinoatrial nodein the right atrium. From there, the electrical wave spreads through the whole right and left atrium. The electrical signal is then delayed at the AV node, before it spreads rapidly through the bundle of his and excites both ventricles. As the electrical wave initiates contraction, the atria contract first, pumping the blood into the ventricles, after which the ventricles contract, pumping the blood into the lungs and the whole body.

A normal rhythm of the heart results in about 60 beats per minute, which accumulate to about 3 billion beats in a lifetime. One can imagine thatit is possible for this electrical wave to be disturbed, resulting in cardiac arrhythmia.

Fig 1. Normal excitation (on the left). The yellow dot indicates the sinoatrial node and the gray dot represents the AV-node. On the right, disturbed excitation leading to cardiac arrhythmia

There are two main driving mechanisms for cardiac arrhythmias: reentry and triggered activity. Anatomical reentry occurs when the electrical wave starts to rotate around an obstacle, e.g. around a valve, a vein or scar tissue, which can be compared with a Mexican wave. This rotation sustains itself and is usually very fast, taking over the normal rhythm of the heart of the sinoatrial node. Depending on the size of the smallest rotation, one can make a distinction between macro.

reentry (large obstacle) or localized reentry (small obstacle). However, reentry is also possible without an obstacle, whereby the wave rotates around an excitable core, which is called functional reentry. Functional reentry is also often denoted a spiral wave or a rotor. The second driving source is triggered activity or automaticity. The latter mechanism gives rise to focal activation, with most likely a concentric activation pattern. One can view this as extra sinoatrial nodes which occur at wrong places in the heart, and if fast, they can also take over the normal rhythm.

We can divide most arrhythmia in two different types: the regular arrhythmia, called tachycardia, and the irregular arrhythmia, called fibrillation as they make the atrial of ventricular walls fibrillate. Therefore, including the location of the arrhythmia, we distinguish 4 main types of arrhythmia: atrial tachycardia (AT), atrial fibrillation (AF), ventricular tachycardia (VT) and ventricular fibrillation (VF). A tachycardia can have two mechanisms, a regular reentry or a regular single focal source, while fibrillation can be any combination of multiple irregular sources, reentries and focal sources which appear irregularly in the heart.

Treatment of cardiac arrhythmia

It is the task of a cardiac electrophysiologist (EP) to determine the mechanism of a cardiac arrhythmia. In case medication is not successful, the EP can perform ablation therapy in order to stop the arrhythmia.

Fig 2. Ablation of a patient

For this, the electrophysiologist (medical doctor) inserts a catheter (measuring electrodes) in the heart and records the electrical activity in the heart. As a result, the medical doctors can obtain a color coded map, as presented in the Figure \ref{map}. In case a clear reentry is detected in the heart, the EP will create a scar which interrupts the reentry path. In case a focal source is detected, this location will be ablated so the focal source can no longer emit electrical activity. It can be very challenging to precisely determine the mechanism of an arrhythmia in a patient. Moreover, wrong ablation lines can make the heart even more prone to new arrhythmia, while also limiting the contractility of the heart. Unfortunately, in the clinical practice there still exist no good automatic algorithms to determine the location of cardiac arrhythmia. This is exactly the problem which we tackle in our research group.

Fig 3. A color coded map created with the mapping system CARTO by Biosense Webster for an atrial tachycardia. The red dots indicate the ablation points

Directed Graph Mapping - DGM

This map needs to be interpreted by the medical doctor, which requires a high specialism. Interpretation of these catheter maps can be often difficult and prone to interpretation. Different EPs can have a complete different ablation strategy based on the same map. Recently, at Ghent University, we have developed an automatic algorithm which can detect any type of rotational activity or focal activity, which we called directed graph mapping (DGM). For this we have merged concepts of network theory with cardiac arrhythmia. Network theory is used for analysing brain activity, social networks and Google searches to name only a few applications. It has numerous applications in many different fields, but strangely enough it was never applied to the heart, although this is a very natural description of cardiac excitation.

DGM takes as input measurements which have been gathered by the catheter. After the doctor has finished to map the inside of the atria, we can apply our tool (see Figure). The XYZ-coordinates of the electrodes and the Local Activation Time (LAT) (the times when the electrical waves passed in the heart) are loaded in the software. With this information, a directed graph is created. Then we apply special algorithms to find cycles in the network. Afterwards, additional filters are applied (which rates the loops based on how orthogonal they pass the wavefront, we call this phase variance). The software then visualises the rotational circuit or focal source of the arrhythmia, so the doctor can easily know where to ablate. DGM will be patented, and we hope one day to bring our software to the market. However, the are many opportunities to do additional research with this idea, and many our thesis subject are all related to the further development of DGM.

Fig 4. First, the annotated points with corresponding LAT values are loaded (a). Second, for each node, we determine the neighbors and an arrow is drawn in case the difference in LATs values of two neighbors and their corresponding distance is within the considered velocities (b-c). Then, the network is created for each node and merged (d). Afterwards, the cycles are detected (e) and the ones with the highest variance are considered as the responsible of the AT (f)

Computer modeling

Besides DGM, we also focus on computer modeling of the heart. For this we have adopted openCARP, which is a community driven open source code for generating computer simulations of electrical propagation of cardiac excitation. The benefits of using openCARP instead of your own code are the flexibility in changing your model for simulations. Also, you have to write python based scripts for generating your simulations, which can then be shared with the community upon publishing new research findings. Have a look at the website of openCARP (https://opencarp.org/) for further information.

Fig 5. Simulation