Thesis

Please check below for more info on our current research

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?

2025-26 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.

Cardiac Christmas Trees – Exploring Wave Breakup Around Scars in 3D Heart Models

Summary · In certain heart rhythm disorders (arrhythmias), electrical activity in the heart becomes chaotic due to rotors—spiraling waves of excitation that can self-sustain and disrupt normal contraction. In 3D cardiac tissue, these rotors can extend into structures called scroll waves, which anchor, twist, and evolve through the thickness of the heart wall. These scroll waves can be surprisingly stable—or break up into complex patterns that make arrhythmias harder to treat. One key factor influencing this behavior is scar tissue in the heart, often formed after heart attacks or due to disease. But how exactly does the shape of a scar affect wave dynamics?

Mission · This project aims to explore how asymmetrically shaped scars in cardiac tissue affect the stability and breakup of scroll waves. Using state-of-the-art computational simulations, you will model scroll waves in 3D cardiac tissue and introduce scars of various shapes and sizes that you will systematically vary in size. Using our software you will investigate whether wave breakup occurs and multiple wavefronts emerge, forming "garlands" around the scar—earning the name Cardiac Christmas Trees. You will then relate the observed wave behavior to fundamental tissue properties such as Action Potential Duration (APD) –how long a cell stays excited– Conduction Velocity (CV) –how fast the wave travels through tissue–.

Why this matters · Understanding how scar geometry influences scroll wave stability could help explain why some arrhythmias are more dangerous or more resistant to treatment than others. It may also lead to improved planning of ablation therapy, where targeted destruction of tissue is used to interrupt these arrhythmias.

Build the Future of Cardiac Arrhythmia Detection: Develop an Automated Testing Framework for Cutting-Edge Software to Treat Cardiac Arrhythmia

Background · The Directed Graph Mapping (DGM) software is an innovative tool developed to detect and analyze cardiac arrhythmias using network theory. As our software evolves, it is crucial to ensure that new versions consistently produce accurate and reliable results when applied to clinical data. To achieve this, we aim to develop a comprehensive testing framework that will automatically validate software updates against our entire database of clinical cases.

The DGM library contains various algorithms that can be used for different cardiac arrhythmia. These algorithms include searching for rotational activity in the heart (re-entry), searching for electrical sinks and sources (focal activity), creating a helmholtz decomposition of the graph or transforming the data into a vectorfield. Another crucial component of directed graph mapping is, as the name implies, the creating of a directed graph. As such we would also like to validate its creation. Thus, with this thesis, you will learn about various types of cardiac arrhythmia (AT, VT, AF ...) and get the opportunity to contribute something that will be used far into the future.

The images below show some of the results that a testing framework can verify. These images show respectively:

  1. Rotational activity detected by the cycle search algorithm.
  2. Focal search on a 2D plane showing sources in green and sinks in purple.
  3. A graph created by DGM for a simulated spherical heart with AT.
  4. A vectorfield created by DGM.

Summary · An initial testing framework already exists, allowing comparisons between different versions of the software by running algorithms and analyzing the results. However, this framework currently includes only a single test. The goal of this thesis is to significantly expand and improve this testing system by developing additional automated test pipelines that target the main components of the codebase and core outcomes.

Your initial goal will be to develop automated pipelines to test core functionalities and key results of the software on the entire clinical database. This also includes creating a test report that gives a concise overview of the results.

Depending on the intrests of the thesis student, the framework can be expanded to include one or more of the following features:

  • Intermediate Results Validation: Integrate intermediate tests within the pipelines to capture and compare intermediate outputs at fixed points during execution.
  • Enhanced Testing Coverage: Expand the range of algorithms and components being tested to cover a broader spectrum of the codebase.
  • Visualization and Reporting: Create clear, user-friendly visualizations and test reports to summarize results and highlight discrepancies.
  • Statistical Analysis: Calculate relevant statistics and performance metrics to assess the consistency and accuracy of the results.
  • Reproducibility and Stability Verification: Incorporate our previous research papers and analysis pipelines to ensure that past results remain consistent across newer software versions. This will enhance reproducibility, especially when new cases are added to the database. This task is particularly valuable for gaining experience with diverse concepts related to cardiac arrhythmias.
  • Algorithm Comparison: Implement the ability to test different algorithms against each other on the same database to evaluate their relative performance and consistency.
  • Model Validation and Cross-Validation: Implement statistical validation methods (e.g., cross-validation) to assess the robustness of algorithms when comparing outputs across versions.
  • Version Tracking and Reporting: Maintain a detailed log of each test run, including software version, algorithms used, and outcomes, to ensure full traceability and reproducibility

Enhancing the Transition from Simulated to Clinical Data in Atrial Tachycardia Diagnosis

Summary · Currently, many methods we develop to diagnose atrial tachycardia show promising results for simulated data, but fail to do so on clinical data. We believe the main issue is the absence of lines of functional or scar-related conduction block in our simulations. These areas lead to incorrect estimation of the physiological wavefront propagation, causing our algorithms to produce erroneous detections. Conduction block is defined as a region where excitation waves cannot travel through. It can be functional in nature, meaning the refractory period of some wave is responsible for the block of another wave. The other type is scar-related, where the conduction block is related to scarring of the cardiac tissue.

Mission · To simulate this, you will use the Finitewave Python package, a very user-friendly tool developed by our team for simple excitation wave simulations. You will start as simple as possible, using the principles of topology to deform the atrial surface into its simplest form: a square sheet with 2 holes. These simulations can then serve as a benchmark for our current AT detection algorithms and allow us to improve them by implementing additional filtering steps to model natural wavefront propagation accurately. We strongly believe this project is highly relevant, as it is the final obstacle that stops us from automatically diagnosing re-entry-based atrial tachycardia.

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